Tag: Automation

  • Loopr AI: Bellevue AI Profile — Product, Footprint & Verified Record

    Loopr AI: Bellevue AI Profile — Product, Footprint & Verified Record

    Washington AI Companies Census · Verified Record (November 19, 2026)
    Sourced Baseline: As of November 19, 2026, Loopr AI (Loopr AI, Inc.) operates as an independent manufacturing AI company based in Bellevue and the Greater Seattle area (loopr.ai, distinct from parked domain looprai.com). Winner of Startup of the Year at the 2026 GeekWire Awards and backed by a $5.4M August 2025 round led by Growth Warrior Capital ($5.4M to $5.68M cumulative), Loopr AI builds hardware-agnostic edge computer vision software for factory quality inspection.
    PROFILE BRIEF · HOUSE PDF & MARKDOWN TWIN
    Loopr AI — Bellevue AI Profile
    Download the formatted 2-page profile brief (PDF) or open the single-source Markdown twin for Loopr AI. Every claim below links to its verified primary source.

    What Loopr AI Builds

    Loopr AI develops a hardware-agnostic AI Quality Intelligence System built to automate visual defect inspection and quality assurance on factory floors. Rather than forcing manufacturers to rip out production lines for proprietary camera rigs, loopr.ai runs edge computer vision models on plants' existing fixed cameras and operator tablets to inspect parts and surface systemic root-cause trends.

    The platform spans three production checkpoints—incoming parts, inline processes such as welding, painting, and assembly, and final dispatch—while also digitizing operator checklists. Beyond defect detection, Loopr AI markets real-time plant safety monitoring, after-sales and MRO warranty documentation, visual inventory tracking, and upcoming agentic quality workflows.

    • Hardware-agnostic factory vision: Deploys on existing fixed cameras and industrial tablets without requiring specialized proprietary vision hardware.
    • End-to-end quality checkpoints: Covers incoming material checks, inline welding/painting/assembly inspection, final dispatch verification, and root-cause trend analysis.
    • Expanded plant operations: Supports digitized inspection checklists, plant safety monitoring, MRO/warranty documentation, and visual inventory tracking across regulated industrial settings.
    Loopr AI — Verified Washington AI Snapshot
    Verified profile snapshot for Loopr AI (Bellevue, WA) as of November 19, 2026.
    FieldVerified Record (As of Nov 19, 2026)
    CompanyLoopr AI (Loopr AI, Inc.)
    Official Websitehttps://www.loopr.ai/
    Washington HQBellevue, WA (Greater Seattle area, WA)
    Other WA FootprintOfficial Info for AI page lists Greater Seattle area, Washington, USA (exact street address not publicly disclosed); backed by Tacoma Venture Fund; co-hosted Factory AI Tour 2026 with Microsoft
    Founded2021 (Loopr AI Info for AI & GeekWire)
    Founders / LeadershipPriyansha Bagaria (Founder & CEO), Kevin Hannegan (Co-founder / CPO), Anirban Konar (Co-founder / CDO)
    StageSeries A
    Announced Funding$5.4M round (Aug 2025, led by Growth Warrior Capital with Stellar Ventures and Tacoma Venture Fund; $5.4M–$5.68M cumulative)
    Team SizeNot publicly disclosed
    Primary Ecosystem Ties2026 GeekWire Awards Startup of the Year winner; Tacoma Venture Fund; Microsoft Factory AI Tour 2026

    Washington Footprint and Leadership

    Loopr AI's corporate footprint sits in Bellevue and the Greater Seattle area: its official Info for AI page locates the company in the "Greater Seattle area, Washington, USA," regional records place its headquarters in Bellevue, and GeekWire refers to it as a Seattle-area startup (an exact street address is not publicly disclosed). Readers should note that its official domain is loopr.ai, whereas looprai.com is an unrelated parked for-sale page.

    Loopr AI, Inc. was founded in summer 2021 by CEO Priyansha Bagaria, who originally built defect-detection software for her family's manufacturing business in India, alongside founding executives Kevin Hannegan (Chief Product Officer) and Anirban Konar (Chief Data Officer) (Loopr AI About Us; Info for AI). In July 2026, GeekWire Tech Moves reported that Loopr AI hired Monica Lazo as sales director from Boston visual-inspection firm Neurala. Loopr AI does not publicly disclose an official employee headcount.

    Aerial view of the Greater Seattle and Bellevue Lake Washington technology corridor with Mount Rainier on the horizon
    Washington AI ecosystem context: Aerial view of the Greater Seattle and Bellevue Lake Washington technology corridor with Mount Rainier on the horizon.

    Products, Users, and Public Traction

    Loopr AI serves enterprise industrial manufacturers across aerospace (AS9100), heavy engineering, automotive and commercial vehicles, building products, medical devices (ISO 13485), battery and semiconductor production, chemicals, and data center control systems. In GeekWire reporting, Loopr AI was reported to serve 10 Fortune 1000 companies across aerospace, automotive, and chemical manufacturing, though specific Fortune 1000 client names and revenue totals are not publicly disclosed.

    In May 2026, Loopr AI won Startup of the Year at the 2026 GeekWire Awards, prevailing over a finalist field that included ElastixAI, Dropzone AI, Dopl Technologies, and mpathic (GeekWire, Apr 16, 2026). On loopr.ai, the company highlights co-hosting a six-city Factory AI Tour 2026 with Microsoft in September and October 2026, alongside industrial program partnerships and challenges with Marvin (Selected Innovation Partner 2026), TRATON (2026), Safran (2024), and Sandvik (2023). Loopr AI claims on its homepage up to 100% inspection coverage across three checkpoints, 30% fewer quality escapes, and 20% to 30% less scrap and rework.

    Funding, Stage, and Place in Washington's AI Landscape

    In August 2025, GeekWire reported that Loopr AI raised a $5.4M funding round led by Growth Warrior Capital, with participation from Stellar Ventures and Washington-based Tacoma Venture Fund (Plug and Play Tech Center is also listed among backers). Industry records place cumulative funding at $5.4M to $5.68M including earlier seed and accelerator capital and classify the 2025 financing as Series A, while some secondary trackers describe the round as a Seed financing. No newer round or acquisition has been publicly reported through November 2026.

    Within Washington's AI ecosystem, Loopr AI stands out for winning GeekWire's 2026 Startup of the Year on a relatively capital-efficient funding base while deploying edge computer vision into heavy manufacturing. Alongside Puget Sound industrial and physical-AI peers such as Reflection Robotics and Jinn Labs, Loopr AI reflects regional momentum in applied industrial AI paired with Microsoft enterprise go-to-market ties.

    Related Washington AI Profiles & Tygart Media Context

    Verified Primary Sources

  • Reflection Robotics: Seattle AI Profile — Product, Footprint & Verified Record

    Reflection Robotics: Seattle AI Profile — Product, Footprint & Verified Record

    Washington AI Companies Census · Verified Record (November 19, 2026)
    Sourced Baseline: As of November 19, 2026, Reflection Robotics is an active, independent Seattle physical-AI startup building single-demonstration robot foundation models for ultra-precision manufacturing. Led by former NVIDIA robotics research manager Arsalan Mousavian, the company participated in Plug and Play Seattle Batch 3 and pitched at AI House Demo Day in September 2026, with no funding amount publicly disclosed.
    PROFILE BRIEF · HOUSE PDF & MARKDOWN TWIN
    Reflection Robotics — Seattle AI Profile
    Download the formatted 2-page profile brief (PDF) or open the single-source Markdown twin for Reflection Robotics. Every claim below links to its verified primary source.

    What Reflection Robotics Builds

    Reflection Robotics builds robot foundation models for ultra-precision manufacturing automation, aiming to eliminate the multi-week custom programming and teleoperation data collection that keep most factory tasks manual. On reflectionrobotics.co, the company argues that roughly 72% of manufacturing operations are still performed by hand because legacy industrial robots lack the flexibility and sub-millimeter precision needed for tight-tolerance assembly and frequent line changeovers.

    Instead of building bespoke hardware, digital twins, or teleoperation farms requiring thousands of recorded runs, Reflection Robotics structures deployment into a three-step zero-code pipeline called From Demonstration to Production. An operator first demonstrates a physical task once; the robot autonomously collects additional task data in under 30 minutes; Reflection Robotics then post-trains its pre-trained foundation model in the cloud for that specific task; and the job transitions onto the production floor the same day.

    In Plug and Play Seattle's Batch 3 announcement, the program summarized Reflection Robotics as making it faster and cheaper to automate physical manufacturing tasks without custom hardware or bespoke machines. Researchers should note two practical disambiguations: the company's live website is reflectionrobotics.co (whereas reflectionrobotics.com resolves to a parked GoDaddy domain-for-sale page), and Seattle-based Reflection Robotics is completely unrelated to frontier generative AI lab Reflection AI.

    • Single-demonstration onboarding: operator shows a repetitive task once, followed by under 30 minutes of autonomous data collection.
    • Cloud post-training: adapts pre-trained robot foundation models without writing task-specific code, building digital twins, or running teleoperation farms.
    • Sub-millimeter tolerance target: designed for same-day production launch and rapid line changeovers using 15 to 30 minutes of operator time per new task.
    • Hardware-agnostic framing: targets standard industrial robotics setups rather than requiring bespoke single-purpose factory machines.
    Reflection Robotics — Verified Washington AI Snapshot
    Verified profile snapshot for Reflection Robotics (Seattle, WA) as of November 19, 2026.
    FieldVerified Record (As of Nov 19, 2026)
    CompanyReflection Robotics
    Official Websitehttps://www.reflectionrobotics.co/
    Washington HQSeattle, WA
    Other WA FootprintUW CoMotion Labs via Plug and Play Seattle; pitched at AI House Pier 70 (Seattle)
    FoundedNot publicly disclosed
    Founders / LeadershipArsalan Mousavian (Co-Founder & CEO); other co-founders not publicly disclosed
    StageAccelerator (Plug and Play Seattle Batch 3)
    Announced FundingNot publicly disclosed
    Team SizeNot publicly disclosed
    Primary Ecosystem TiesPlug and Play Seattle Batch 3 (March 2026); AI House Demo Day presenter (Sept 2026); Seattle NVIDIA robotics research lineage

    Washington Footprint and Leadership

    Reflection Robotics is based in Seattle and surfaced through two major regional startup showcases during 2026. On March 23, 2026, GeekWire reported that Plug and Play selected Reflection Robotics as one of nine Seattle-based companies in its 10-startup third Seattle accelerator cohort (running 12 weeks through mid-June 2026, with access to Plug and Play's office inside UW CoMotion Labs and space at SNBL Global Gateway in Everett). Six months later, on September 25, 2026, Reflection Robotics was among 25 early-stage startups invited to pitch local and national VCs at Seattle's Pier 70 AI House Demo Day—an event GeekWire noted deliberately included outside startups beyond AI House's own incubator portfolio.

    In GeekWire's September 25, 2026 Demo Day report, editor Todd Bishop identified co-founder and CEO Arsalan Mousavian, a former NVIDIA robotics research manager—connecting the startup directly to Seattle's active NVIDIA Robotics Research Lab talent pool. Names of Mousavian's co-founders, an official founding year, and total employee headcount have not been publicly disclosed on reflectionrobotics.co or in GeekWire coverage.

    Seattle waterfront startup incubator loft overlooking Puget Sound ferry lanes and the Olympic Mountains
    Washington AI ecosystem context: Seattle waterfront startup incubator loft overlooking Puget Sound ferry lanes and the Olympic Mountains.

    Products, Users, and Public Traction

    Reflection Robotics targets manufacturing facilities and industrial operations teams confronting acute factory labor shortages and frequent product changeovers. On reflectionrobotics.co, the company cites a U.S. manufacturing labor gap of 400,000 workers projected to reach 2,100,000 by 2030, pitching same-day task retraining as a way to expand the share of factory work with viable automation payback.

    Public commercial proof points remain undisclosed as of November 19, 2026. Reflection Robotics has not publicly named pilot factories, commercial manufacturing customers, supported robot arm vendors, or software pricing tiers. Its website provides direct Partnership, Careers, and Contact navigation paths for prospective industrial partners and prospective hires.

    • Target buyers: manufacturing plants and industrial operations teams needing sub-millimeter precision and frequent line changeovers.
    • Public site pathways: Mission, Careers, Contact, and Partnership sections live on reflectionrobotics.co.
    • Undisclosed commercial details: pilot customer names, factory deployment counts, and contract pricing have not been published.

    Funding, Stage, and Place in Washington's AI Landscape

    Reflection Robotics completed Plug and Play Seattle Batch 3 in mid-2026 and pitched investors from Seattle, the Bay Area, Boston, New York, and Colorado at AI House Demo Day in September 2026. No institutional seed round, funding dollar figure, or venture investor syndicate beyond Plug and Play accelerator participation has been publicly announced through November 2026.

    Within Washington's growing physical-AI and robotics ecosystem, Reflection Robotics occupies the pure foundation-model end of the industrial automation spectrum. Alongside Seattle physical-intelligence stealth venture Noosphere Labs and industrial workstation builder Outrun—and drawing directly from Seattle's NVIDIA robotics research community—Reflection Robotics tests whether single-demonstration post-training can hold tight sub-millimeter factory tolerances quickly enough to compete with traditional systems integrators.

    Related Washington AI Profiles & Tygart Media Context

    Verified Primary Sources

  • Outrun: Seattle AI Profile — Product, Footprint & Verified Record

    Outrun: Seattle AI Profile — Product, Footprint & Verified Record

    Washington AI Companies Census · Verified Record (November 19, 2026)
    Sourced Baseline: As of November 19, 2026, Outrun (Outrun Robotics) is an active, independent Seattle industrial robotics startup backed by AI2 Incubator and NVIDIA Inception, having graduated from Creative Destruction Lab (CDL) Seattle's fifth cohort in July 2026. Founding team details, founding year, headcount, and institutional funding amounts are not publicly disclosed.
    PROFILE BRIEF · HOUSE PDF & MARKDOWN TWIN
    Outrun — Seattle AI Profile
    Download the formatted 2-page profile brief (PDF) or open the single-source Markdown twin for Outrun. Every claim below links to its verified primary source.

    What Outrun Builds

    Outrun (operating as Outrun Robotics) builds multi-purpose robotic workstations designed to bring industrial automation to small- and medium-scale production lines. On outrun.bot, the company targets stationary, repetitive factory tasks—specifically case packing, kitting, machine tending, and inspection—where conventional custom-engineered robotic cells are often too slow or expensive to justify.

    The product combines a packaged Hardware Platform with a managed Software Platform. On the factory floor, each workstation bundles industrial collaborative robots (cobots), 3D cameras, compute hardware, a tablet user interface, and secure networking. Behind the hardware sits Mission Control, Outrun's 24/7 remote engineering and teaching workforce: when a factory introduces a new SKU or needs to adjust robot behavior, Outrun's remote engineers update the software and deliver tailored tablet interfaces so on-site floor operators can switch recipes or make routine adjustments without writing code.

    Readers evaluating Outrun's technical positioning should note a clear framing distinction across sources. Seattle's CDL Cohort 5 graduation report in GeekWire describes Outrun Robotics as building 'capable, flexible, intelligent robotic workstations' alongside Seattle's AI2 Incubator portfolio, whereas outrun.bot focuses on 3D vision hardware, compute, and human-in-the-loop remote engineering via Mission Control rather than foregrounding explicit AI model branding on its landing page.

    • Target factory tasks: case packing, kitting, machine tending, and inspection on small-to-medium production lines.
    • Hardware Platform: industrial cobots, 3D cameras, edge compute, tablet operator UI, and secure remote networking.
    • Software Platform (Mission Control): 24/7 remote engineering team managing SKU additions, behavior updates, tailored operator UIs, and uptime recovery.
    • Four-step rollout: (1) process review, (2) on-site production pilot, (3) promotion to live production, and (4) ongoing multi-year operation.
    Outrun — Verified Washington AI Snapshot
    Verified profile snapshot for Outrun (Seattle, WA) as of November 19, 2026.
    FieldVerified Record (As of Nov 19, 2026)
    CompanyOutrun (Outrun Robotics)
    Official Websitehttps://outrun.bot/
    Washington HQSeattle, WA
    Other WA FootprintAI2 Incubator / AI House (Seattle); Creative Destruction Lab Seattle at UW Foster School of Business
    FoundedNot publicly disclosed
    Founders / LeadershipNot publicly disclosed
    StageIncubator / Accelerator (AI2 Incubator; CDL Seattle Cohort 5)
    Announced FundingNot publicly disclosed (backed by AI2 Incubator; non-equity CDL Seattle graduate)
    Team SizeNot publicly disclosed
    Primary Ecosystem TiesAI2 Incubator / AI House; Creative Destruction Lab Seattle (Cohort 5, July 2026); NVIDIA Inception

    Washington Footprint and Leadership

    Outrun's Washington footprint is anchored by two prominent Seattle startup institutions. The company is listed as an active, non-acquired portfolio company on the AI House / AI2 Incubator directory and displays 'Backed by AI2 Incubator & NVIDIA Inception' in the footer of outrun.bot. In July 2026, GeekWire reported that Outrun Robotics graduated in the fifth cohort of Creative Destruction Lab (CDL) Seattle, a nine-month, non-equity nonprofit accelerator hosted at the University of Washington's Foster School of Business and backed by founding members including UW and Microsoft.

    Public executive disclosure is notably sparse compared to peer startups in the region. Neither outrun.bot, the AI House directory, nor GeekWire's CDL Seattle cohort announcement publishes founder names, leadership biographies, a founding year, or total employee headcount. One minor site-maintenance signal is also visible: outrun.bot still displays a 'Copyright Outrun Robotics 2025' footer and placeholder '#' links for its Terms of Service and Privacy Policy, though its July 2026 CDL Seattle graduation confirms active operations in 2026.

    Seattle waterfront startup incubator loft overlooking Puget Sound ferry lanes and the Olympic Mountains
    Washington AI ecosystem context: Seattle waterfront startup incubator loft overlooking Puget Sound ferry lanes and the Olympic Mountains.

    Products, Users, and Public Traction

    Outrun builds for small- and mid-sized manufacturing operators running high-mix or medium-volume lines that struggle to amortize traditional systems-integrator fees every time packaging or part SKUs change. Rather than selling bare robot arms and leaving programming to plant engineers, Outrun's hybrid model pairs off-the-shelf industrial hardware with ongoing remote engineering support.

    While Outrun has not publicly named paying manufacturing customers or disclosed installed workstation counts, outrun.bot lists an established roster of industrial hardware, gripper, framing, and compute partners integrated into its workstations: FANUC, Universal Robots (UR), Yaskawa, Piab, Coval, Vention, 80-20, and NVIDIA. Pricing, leasing versus capital-purchase terms, and customer pilot metrics are not publicly disclosed; prospective buyers engage via an on-site intake form and email contact (info@outrun.bot).

    • Named hardware and compute brand partners: FANUC, Universal Robots (UR), Yaskawa, Piab, Coval, Vention, 80-20, and NVIDIA.
    • Deployment model emphasizes an on-site production pilot before full promotion to live production.
    • Commercial customers, unit counts, and pricing structure are not publicly disclosed.

    Funding, Stage, and Place in Washington's AI Landscape

    Outrun's publicly verified backing consists of AI2 Incubator support, membership in NVIDIA Inception, and graduation from the non-equity CDL Seattle advanced manufacturing track in July 2026. No institutional venture funding dollar figure, lead VC firm, or headcount milestone has been publicly announced through November 2026.

    Within Washington's emerging physical-AI and industrial robotics cluster, Outrun offers an instructive contrast in technical strategy. Whereas Seattle peer Reflection Robotics focuses on training robot foundation models to learn sub-millimeter tasks from a single demonstration in software, Outrun pairs standard industrial cobots and 3D cameras with a 24/7 remote engineering workforce (Mission Control) to solve the SKU-changeover and uptime gap pragmatically for smaller factories.

    Related Washington AI Profiles & Tygart Media Context

    Verified Primary Sources

  • Aria: Seattle AI Profile — Product, Footprint & Verified Record

    Aria: Seattle AI Profile — Product, Footprint & Verified Record

    Washington AI Companies Census · Verified Record (November 19, 2026)
    Sourced Baseline: Aria (operated by Alpyne Labs at heyaria.com) is an active, revenue-generating Seattle behavioral health practice automation startup listed in Seattle's AI2 Incubator / AI House portfolio. Originally launched in October 2025 as a bootstrapped startup under the product name Juno, the company automates insurance billing workflows for mental health and ABA group practices and pitched venture investors at AI House Demo Day in September 2026.
    PROFILE BRIEF · HOUSE PDF & MARKDOWN TWIN
    Aria — Seattle AI Profile
    Download the formatted 2-page profile brief (PDF) or open the single-source Markdown twin for Aria. Every claim below links to its verified primary source.

    What Aria Builds

    Aria (built by Alpyne Labs) develops a HIPAA-compliant practice automation platform for insurance-based behavioral health, mental health therapy, and Applied Behavior Analysis (ABA) group practices. When GeekWire covered the startup's debut on October 2, 2025, Alpyne Labs introduced its first product as Juno, a Chrome-extension digital assistant floating over electronic health records (EHRs) such as TherapyNotes, SimplePractice, and TheraNest to handle benefits checks, billing, payment posting, and scheduling.

    Under its current brand at heyaria.com, the company focuses squarely on four recurring back-office workflows that bottleneck growing group practices: Eligibility and Benefits Verification (EBV), claim scrubbing, provider credentialing and roster management, and nightly practice performance reporting. As detailed on its Technology page, Aria layers directly onto a practice's existing EHR (including SimplePractice) with no migration required, routing verification checks through clearinghouses or automated payer portal logins and writing admin notes and benefits PDFs straight back into client charts. Aria explicitly limits automation to administrative operations without touching clinical judgment, halting and flagging exceptions—such as lapsed coverage or mismatched member IDs—for human staff.

    • Eligibility & Benefits Verification (EBV): Checks active status, deductibles, copays, authorizations, and dual-coverage coordination of benefits before scheduled sessions.
    • Payer-Specific Claim Scrubbing: Validates claims before submission to catch modifier errors, coding mismatches, and documentation gaps.
    • Credentialing & Roster Management: Tracks provider applications across insurers such as Aetna, BCBS, UHC, and Cigna and alerts staff when payer applications stall.
    • Nightly Reporting (Ask Aria): Aggregates EBV results, credentialing progress, payment history, and collection gaps in one dashboard refreshed nightly.
    Aria — Verified Washington AI Snapshot
    Verified profile snapshot for Aria (Seattle, WA) as of November 19, 2026.
    FieldVerified Record (As of Nov 19, 2026)
    CompanyAria (Alpyne Labs; formerly launched under product name Juno)
    Official Websiteheyaria.com
    Washington HQSeattle, WA (per GeekWire and AI House; heyaria.com/about also notes a distributed U.S. team)
    Other WA FootprintAI House / AI2 Incubator at Pier 70, Seattle
    FoundedNot publicly disclosed (public product launch as Alpyne Labs / Juno reported Oct 2, 2025)
    Founders / LeadershipJustin Ith (Co-founder & CEO), Michelle Arendas (Co-founder & VP of Customer Success), Ronen Reouveni (Co-founder & Head of AI)
    StageIncubator / Bootstrapped
    Announced FundingNot publicly disclosed (bootstrapped at Oct 2025 launch per GeekWire)
    Team SizeNot publicly disclosed
    Primary Ecosystem TiesAI2 Incubator / AI House (Pier 70 Demo Day, Sep 2026); UW and Madrona Venture Labs alumni roots

    Washington Footprint and Leadership

    Aria originated in Seattle and maintains active ties to Seattle's core startup institutions. GeekWire profiled Alpyne Labs as a Seattle health tech startup working with design-partner practices mostly in the Seattle area, and Seattle's AI House / AI2 Incubator companies directory lists Aria as an active incubator company. On September 25, 2026, GeekWire reported that Aria (formerly Alpyne Labs) was among 25 startups pitching national investors at AI House's inaugural Demo Day on Pier 70. While its incubator and launch record are rooted in Seattle, Aria's About page notes that the company operates with a distributed U.S. team and does not list a public street office.

    Aria's three co-founders combine local venture-studio and revenue-cycle backgrounds. Co-founder and CEO Justin Ith is a University of Washington graduate who began his career at Madrona Venture Labs, launched a medical device startup during the COVID-19 pandemic, and returned to Madrona to co-found healthcare billing and claims automation spinout Outbound AI in 2023. Co-founder and VP of Customer Success Michelle Arendas brings roughly 20 years of healthcare administration and automation experience, and Co-founder Ronen Reouveni serves as Head of AI (per heyaria.com/about).

    Seattle waterfront startup incubator loft overlooking Puget Sound ferry lanes and the Olympic Mountains
    Washington AI ecosystem context: Seattle waterfront startup incubator loft overlooking Puget Sound ferry lanes and the Olympic Mountains.

    Products, Users, and Public Traction

    Aria sells on a predictable flat monthly pricing model positioned against traditional percentage-of-collections billing services, alongside free lead-generation tools on heyaria.com including a three-minute practice audit and local ZIP-code Area Insights market benchmarks. At its October 2025 launch under the Juno name, GeekWire reported that Alpyne Labs was already generating revenue from mental health practices and had opened a public waitlist after beta testing with Seattle-area design partners.

    On its live site, Aria claims adoption across more than 1,000 clinicians nationwide and publishes named testimonials from four behavioral health practices: Talk. Heal. Thrive., reVIBE Mental Health, KZTA Behavioral Health Agency, and Malaya Counseling Mind. In its published Talk. Heal. Thrive. case study—profiling a Washington statewide telehealth therapy group with 30+ clinicians using SimplePractice—Aria reports 2,021 verified appointments, approximately $4,920 in protected revenue, and 339% month-one ROI after replacing manual Medicaid and coordination-of-benefits portal lookups. CEO Justin Ith has also discussed practice revenue leakage on the Wise Practice Podcast (Episode 201).

    • Named Customer Testimonials: Talk. Heal. Thrive., reVIBE Mental Health, KZTA Behavioral Health Agency, and Malaya Counseling Mind (published on heyaria.com).
    • Washington Telehealth Case Study: Company-reported 2,021 verified appointments and ~$4,920 protected revenue in month one at Talk. Heal. Thrive. (30+ clinicians on SimplePractice).
    • Commercial Model: Flat monthly software fee rather than a percentage cut of practice collections.

    Funding, Stage, and Place in Washington's AI Landscape

    Public financial disclosures remain intentionally minimal. GeekWire reported in October 2025 that Alpyne Labs was bootstrapped and generating revenue, and neither Aria nor AI House has publicly announced a formal venture financing dollar amount, investor list beyond incubator participation, incorporation date, or current team headcount as of November 19, 2026. Participating in AI House Demo Day at Pier 70 in September 2026 positioned the company directly in front of visiting and local institutional investors.

    Inside Seattle's applied AI ecosystem, Aria sits within a growing cluster of AI2 Incubator companies tackling U.S. healthcare insurance friction from complementary angles. Where fellow incubator startup Decipher focuses on employee-facing benefits navigation and Seattle peer MediScan AI addresses med-legal record review, Aria concentrates specifically on the revenue-cycle operations of outpatient behavioral health and ABA practices—a fragmented segment where payer rules and credentialing delays directly cap clinical capacity.

    Related Washington AI Profiles & Tygart Media Context

    Verified Primary Sources

  • Your Site’s Next Visitor Sends an Agent. Tobira Wants Yours to Meet It.

    Your Site’s Next Visitor Sends an Agent. Tobira Wants Yours to Meet It.

    Ask your AI to do a simple errand: book a call with ten real companies you might buy from. Not research them. Book the call.

    Tobira’s founder, Vlad, ran that exact test before he built anything. He asked his Claude to book a call with 10 real companies. The result, in his launch account, was zero bookings and zero payments. The agent didn’t fail at reading. It failed at the last step — account walls it couldn’t pass, booking pages that block AI, and forms built for fingers and eyes.

    That zero-for-ten is the failure every small business is about to meet. More buyers research inside ChatGPT or Claude, and increasingly ask the same chat to take the next step: book the call, start the trial, place the order. Your site was built for the researcher, not the errand. If the errand dies at your form, you never learn the buyer was there.

    Tobira Site Agent is one vendor’s answer to that break. We have not tested it on our own sites — this is a desk read, built from Dirty Static’s guide to Tobira Site Agent, itself built from Tobira’s maker site and Product Hunt launch post, facts checked November 7, 2026. Where a number comes from Tobira, we say so.

    The clerk, not the brochure

    Most of what businesses have done for AI visitors so far is brochure work: clearer pages, structured content, files an agent can read. Necessary — and it’s the layer we already run. Tobira bets the brochure isn’t enough. A brochure can’t close. A clerk can.

    Tobira puts a site agent of your own on your website, one that lives there, answers questions, and sells — not just to people, but to the AI agents people send. Your site should not wait to be read by a visiting agent. It should send an agent of its own to meet it, answer it, and do the transaction.

    The job list is sales work, not trivia. Per the maker, the agent asks about the visitor’s case, suggests the option that fits, books the call, starts the trial, captures the lead, and sends a payment link — right inside the chat the visitor started in. Booking, trial starts, leads, and payment happen in that chat, not on a separate page. It can talk by voice, with a transcript, and answers in the visitor’s language. When a case gets hard, a person steps in with the chat context attached, so the human isn’t starting from zero.

    Two visitors, one site

    The mechanism is a split. People see your normal website. Visiting agents get a version made for them, built by your agent from your own content. Per the maker, your agent publishes the files agents look for — llms.txt, agent.json, and AGENTS.md — so a visiting agent reads the offer directly and talks to your agent, instead of scraping page after page.

    The token math is the quiet part of the pitch, and it’s a maker example, not a measurement. In Tobira’s worked example, a normal pricing page runs to about 120,000 tokens of page weight. The structured answer lands at about 1,300 tokens, which the company presents as 92 times fewer tokens. Treat that ratio as a company example, not a measured result for your site. Your pages are not their example page. The direction still makes sense: an agent handed the plan, the price, and the next open slot as clean data finishes faster than one guessing its way through your navigation.

    The guardrails are the real product

    A salesperson who improvises prices is worse than none, and a site agent is a salesperson. The important part of Tobira isn’t the booking trick — it’s three guardrails the maker describes.

    First, answers only from your own materials: your site, price lists, PDFs, and FAQs. If the answer isn’t there, the agent doesn’t guess. It offers a call instead. Second, it shares only what you made public. Pricing and plans you chose to publish can be public; internal notes and deal margins stay private. Third, a person steps in on hard cases, with context — the chat history goes to your team, so the buyer never starts over.

    Rules are set in plain words, per the maker — “price ranges only,” “no promised discounts.” If it works as described, the guardrails are the product. The chat is just where they show up.

    Setup and cost

    Setup, per the maker, starts with a scan: Tobira reads your site and builds the knowledge base from what it finds. You add documents, set tone and rules, and test it on the real questions buyers ask. Then it goes live as a badge or widget.

    How long? The founder says 10 minutes on Product Hunt; the product page says 2 minutes. Both are maker claims. The honest reading is minutes, not days — with the real work in checking what the scan learned before you trust it with prices.

    Pricing, maker site checked November 7, 2026, USD monthly before taxes: Free is $0 with your own AI key — Tobira’s fee is zero, the model bill is yours. Lite is $49, about 300 conversations. Start is $169. Business is $399, adding payments, your own branding with no Tobira badge, and voice up to about 1,500 minutes. Premium is $799, up to 3 sites, and on that tier Tobira keeps 15% of AI-subscription revenue. Paid plans carry a 14-day money-back guarantee, and the maker says a data processing agreement comes with paid plans, with data handled under GDPR.

    Free is a genuine test, not a stripped demo. Paid tiers climb fast.

    The network bet

    A site agent alone is a better shop window. Tobira also wants it to be a doorway into a market. Per the maker, your agent gets a public address in the Tobira network — an email-shaped handle, but for AI. When another company’s agent looks for what you offer, the two agents talk first and check fit. You get a qualified deal, or silence.

    The founder says the network holds more than 800 agents. That is a company figure, not audited, and a network is worth what its members are worth — from outside today you can’t tell a market from a list.

    Where this sits in our own house

    We’ve written this pattern before, from the other side. In October, after Kin’s WebMCP launch, we published “The Agent-Ready Site”: welcome the machine, structure the content, give the agent a way to act, measure what it does. Our fleet runs three of those four layers today. Our llms.txt has been live since October 2026, roughly 62% of our traffic is bots rather than people, and we measure the machine side with Bing’s citation exports.

    Tobira is what the unbuilt layer looks like when a vendor builds it for you. The missing layer was the third: give the agent a way to act. An llms.txt is the brochure on the door — it tells a visiting agent what you offer, in its own language. A site agent is the clerk behind it — it answers back, books, and takes payment. Same argument, next step. The pattern said structure your content for agents. Tobira’s bet is that the structure should be a salesperson.

    For the record: Tygart Media has not installed Tobira or run it on a client site. It’s a live commercial answer to a layer we know is missing.

    The two questions that decide it

    One: do your buyers send agents yet? The whole model depends on buyers asking their AI to book and buy, still a young habit. Every speed figure Tobira publishes — including its “2.2 times faster” claim — is founder-tested, not from an outside lab. If your customers still call and complete forms without trouble, a good booking page beats an agent you have to feed.

    Two: are your materials clean enough to be spoken aloud? An agent answering only from your materials is only as good as those materials. A stale price on a page is an embarrassment. A stale price spoken with confidence by your own agent, with a payment link attached, is worse than no agent at all.

    If your buyers already arrive through AI and die at the booking step, the free route is worth a real test. If they don’t, watch the habit and the network count, and keep your booking page in good shape. Guessing wrong about which one you’re in is the expensive option.

    The practice

    Run the founder’s test on yourself. Open the AI you actually use and ask it to book a call with your business — your real site, no help from you. Watch where it dies. If it books cleanly, your forms were never the wall; file this category under “later.” If it stalls at an account wall, a blocked page, or a form it can’t complete, you’ve just watched the exact failure Tobira is selling against — free, in ten minutes, on your own storefront.


    Want this thinking applied to your own business? Book a call with Will Tygart — 30 minutes, and we start from your receipts, not a pitch deck.

  • How We Qualify Contractor Leads With AI: The Intake Matrix

    How We Qualify Contractor Leads With AI: The Intake Matrix

    How can AI automatically qualify contractor leads?

    Short answer, and it is the whole article: the AI does not qualify the lead. It collects the facts qualification runs on — the same questions, in the same order, at 2 a.m. exactly the way they run at 10 a.m. — and it stops where judgment starts. Qualification questions are not qualification judgment, and the difference is the entire ballgame. The questions collect facts. The judgment decides what the facts mean: is this a job we want — at this size, this distance, for this payer, with the crews already buried?

    That distinction is what makes the automation worth having. A flooded kitchen ten minutes away and a “come look at a stain” forty minutes out both answer the questions perfectly. Only one of them should get your Saturday. The matrix below is how we draw that line before a human ever picks up the phone.

    The intake matrix: six facts, in order

    Every inbound call and form runs through the same six gates. Each gate is a fact, not an opinion, and each one exists because it changes what happens next.

    1. Territory. Where is the job, exactly? Distance is the first filter because it is the one an unsupervised agent gets wrong first — nothing about a polite conversation tells a model that forty minutes each way eats the margin. The address gets captured and checked against the service area before anything is booked.

    2. Trade and scope. What happened, in the caller’s words? Water, fire, mold, storm — and what they can already see: standing water, a burst line, a smell, a stain. The intake records the problem as described; it does not diagnose. Scope facts decide who the lead routes to and how fast.

    3. Size signals. How big is it likely to be — one room or the whole floor, home or commercial space, occupied or empty? Size is a fact the caller usually knows and an agent can ask without pretending to estimate. Nobody quotes anything at intake. Size just tells the human what kind of job this probably is.

    4. Payer path. Is this going through insurance, or is the customer paying directly? Asked plainly, recorded as stated. The answer changes the paperwork, the pace, and who needs to be in the conversation — it never becomes a coverage opinion. The agent asks; it does not advise.

    5. Urgency. Is water still moving? Is the home livable tonight? Urgency is what separates the 2 a.m. emergency from the Tuesday estimate, and it drives the one promise-shaped thing the system is allowed to do: escalate to a human now, not at 8 a.m.

    6. A complete record. Name, callback number, address, and the facts above — logged on every call, whether it books or not. In our own operation the intake pipeline logs calls into a database the follow-up system can query, so no communication ever begins from scratch. A lead that doesn’t book today is still a record the shop owns tomorrow.

    What happens to each outcome

    The matrix sorts; it does not decide. A lead inside territory, in scope, and urgent gets booked or escalated immediately. A lead that is real but not urgent gets logged and followed up by the automated sequence — the same pipeline discipline we run on our own 3 a.m. shift, where follow-ups compile into a daily briefing sorted red, yellow, green for human review. A lead outside the lines — wrong territory, wrong trade — gets a polite, honest close and a record of why. Nothing disappears, and nothing gets promised that a person hasn’t approved.

    What the human keeps

    Pricing. Promises. The handoff. What a job costs is a person’s call, made with the board in front of them. “We’ll have a crew there by nine” is a promise only someone who knows the crew should make. And the moment a prospect stops being a lead and starts being a customer — when they need to hear that a real person owns their problem — a person takes the conversation. The agent’s job is to make that moment arrive fast, facts collected and appointment real. It is not to be that moment.

    The failure shape, so you can recognize it

    Hand an agent the whole seat — the questions and the judgment both — and it will qualify everybody. It will book the estimate outside your service area because nothing told it distance matters. It will fill the calendar with meetings a person would have filtered in thirty seconds, and once your techs learn to distrust anything the system books, you have automated the exact problem you bought it to solve. The bad version doesn’t fail loudly. It fails as a slow leak — wasted drive time, stood-up appointments — until somebody turns it off and announces that AI doesn’t work.

    The seat works when its boundary is drawn, not when the model is smart. A boundary is a list: what it can ask, what it can book, what it can never promise, and where it stops and hands over. Answer, qualify, schedule — first response, not the close. That is not a limitation of the setup. That is the product, correctly shaped.

    This is the intake standard we publish and work from. It describes our method; it is not a guarantee of lead volume or outcomes for any individual business.

  • AI Governance Is the Gate That Caught the Thing

    AI Governance Is the Gate That Caught the Thing

    Most AI governance is written by people who have never had to govern an agent that could actually break something. It reads like insurance paperwork: frameworks, maturity models, principles laminated on a wall. Then a real agent, on a real Tuesday, publishes your drafting notes as if they were the article — and the framework has nothing to say about it, because the framework was never in the room.

    We run a small agency where AI agents do real production work every day: research, drafting, publishing, measurement, and an unattended night queue that runs while we sleep. Governance, for us, is not a document. It is the set of gates that have caught real things. Every rule below exists because something specific got past us first, and the rule is the scar.

    Governance starts with who is allowed to remember

    The agents are stateless; the memory lives in one place. Every agent we dispatch starts cold. It knows nothing about our clients, our sites, or yesterday’s decisions until its work order tells it. That sounds like a limitation. It is the control surface. Because the agent holds no standing memory, it holds no standing permissions either — every job is scoped, priced, and bounded by the order that created it, and when the job ends, the access ends with it.

    The work order itself is a governance document. It states the whole price up front, names the points where a human must tap before work proceeds, and lists what the executor may ship directly and what it must hold. An agent that has to ask a question the order should have answered is not a governance failure by the agent. It is a drafting failure by us, and we treat it that way.

    Draw the doors as lines, not policies

    Four things always need a human tap: money, credentials, outbound sends, and publishing. Everything else, the fleet runs on its own. That split is the entire access policy, and it fits in one sentence because it has to survive contact with a tired operator at the end of a long day. Policies that need interpretation get interpreted — usually at the worst moment, in the most generous direction.

    Credentials get the hardest line. Our agents never hold passwords, never see them, never pass them along. Work that reaches a login the agent cannot complete through an already-signed-in session does not get worked around. It stops, logs a hold, and waits for the human whose door that is. One night this fall, a run reached a registry login at three in the morning. The operator could not face the tap, said so, and went to bed. The work was built to survive stopping, so stopping cost nothing. The credential never moved. That is what a line is for: it has to hold when nobody has the energy to argue with it.

    The risk is not a rogue agent. The risk is a report you believe.

    Every serious failure we have had was a verification failure, not a rebellion. A judge model, asked to grade drafts against a length limit, invented word counts that were off by a quarter to a third — confidently, in both directions. A publish report stated a post ID from memory instead of reading it back from the result file, and the number was wrong. A page was verified from its HTML source and declared fine while its signup form rendered as blank white cards. Drafting notes were once published as article content because the check stopped at “the words are there.”

    So the rule is mechanical: proof lives at the receiving end. Numbers get read back by value from the file that recorded them. Pages get verified as rendered pages, in a browser, the way a reader sees them — not from source. A finished article must show visible structure, or it is a hold, not a publish. None of this requires trusting the agent less. It requires trusting the report exactly as far as the evidence under it, which is where trust should have been calibrated all along.

    We also measure trust per model, the way you would per employee. In scored head-to-head runs on identical briefs, exactly one of our models graded perfectly — every defect real, no invented ones — so that model holds the verifier seat. Another writes fast and clean but invents defects when it judges, so it drafts and never judges. The cheapest tier once certified work it had not actually checked, so its output is never trusted without a pass behind it. “The AI said so” is not a control. A scorecard is.

    Failures become regressions, not folklore

    Every break gets an ID, an entry in the failure ledger, and a gate. When the night queue failed in a new way, the failure was written down with its cause and its receipt, and a regression check was built that fails the next run if the same shape reappears. The ledger is unglamorous. It is also the difference between an operation that learns and an operation that just has stories. Governance that lives in people’s memories leaves with their mood, their weekend, and eventually their employment. Governance that lives in a gate runs every time.

    Each work order also carries one pre-decided instruction for the moment something breaks: if a step’s end state cannot be verified, stop that step, log the hold, and move to the next independent step. Never improvise past a failed check. The decision is made in advance, in calm, because the middle of a failing run is the worst possible place to invent policy.

    The 3 a.m. test

    Here is the only governance question we ask anymore: what can this system not do at 3 a.m.? Our night queue runs unattended with hard caps — a maximum number of jobs, a maximum number of attempts, a hard stop before dawn, and a pause switch that halts everything if a single file exists. Agents working overnight cannot spend money, cannot send anything to anyone, cannot touch a credential, and cannot publish. They can draft, measure, and report. Morning is a review: a human reads what the night produced, approves what is good, and the approvals are the fuel for the next day’s work.

    That is the whole philosophy. Governance is not the set of things your agents are allowed to do. It is the set of things they structurally cannot do while you are asleep, plus the evidence trail that lets you check everything else over coffee.

    What this is not

    This will not satisfy a compliance department, and it is not meant to. If you handle medical records or other people’s money under regulation, you need the paperwork kind too. But the paperwork kind alone will not save you from the ordinary failures — the believed report, the generous interpretation, the confident invented number — because those failures are not compliance failures. They are operating failures, and they happen at every scale, in shops with two people and in companies with two thousand.

    Start where the risk actually is. Give the agents no memory and no keys. Draw four doors and hold them. Verify at the receiving end. Write every failure down and make it a gate. Then run the 3 a.m. test on your own setup, honestly, and fix what it shows you. The framework can come later. The gate that catches the thing comes first.

  • The First Agent You Hire Answers the Phone

    The First Agent You Hire Answers the Phone

    Updated November 2, 2026 — added direct answers on how AI handles after-hours calls for a service business, from our own published work.

    Nobody’s first AI hire is a strategist. Ask a room full of operators what they’d hand off first and nobody says the thinking — they say the phone. The call that comes in at 6:40, after the office went dark at five. The form filled out at midnight, water coming through the kitchen ceiling. The lead that goes to whichever company answers first — and in home services, the first answer is very often the whole sale.

    That seat — first response — is where small businesses are actually hiring agents, and the money just said so out loud. On October 13, Dapta, an AI sales agent company out of Scottsdale and Miami, announced a $6 million seed round led by Chingona Ventures. Its agent, Dapti, answers, qualifies, and books the meeting the moment a prospect reaches out — no human touching it. Dapta says 200,000-plus users are on it, signing up at better than 4,000 a week, and that revenue will pass $4 million this year. Home services is one of the verticals where it took hold first.

    The seat, in a real shop

    First response in a small shop is a person-shaped hole. The owner answers between jobs, one glove off. The office manager answers until five. After five, a service answers, or nobody does, and the prospect — wet floor, search results page in hand — calls the next name on the list. It’s a staffing problem, not a marketing problem — one that shows up at the exact hours nobody wants to staff.

    The money makes it worse. On Google Local Services, a connected call that runs past twenty seconds can bill you whether a person picked up or not — you pay for the ring. And the caller isn’t comparison-shopping at leisure. They’re standing in water, or sleeping in a 90-degree house, and they hire the first competent voice that treats them like a person.

    What the agent can own outright

    The front half of the seat is script work, and script work is exactly what an agent is for. Answer instantly, at any hour, in the customer’s own language — Dapta got its start with Spanish-speaking owners in the U.S., and first response is meeting the customer where they are. Ask the basic qualifying set: what happened, where is it, when did it start, who is paying, how do we reach you. Then do what most shops fumble even in business hours: put a real appointment on a real calendar, confirmed where the customer can see it — not a callback promise the morning forgets.

    None of that requires judgment — just the script, the calendar, and the discipline to run both at 2 a.m. the way they run at 10 a.m., on the nights the office sits empty.

    What the human keeps

    Qualification questions are not qualification judgment, and the difference is the entire ballgame. The questions collect facts. The judgment decides what the facts mean: Is this a job we want — at this size, this distance, for this payer, with the crews already buried? A flooded kitchen ten minutes away and a “come look at a stain” forty minutes out both answer the questions perfectly. Only one of them should get your Saturday.

    The human also keeps pricing, promises, and the handoff. What a job costs is a person’s call, made with the board in front of them. “We’ll have a crew there by nine” is a promise only someone who knows the crew should make. And the moment a prospect stops being a lead and starts being a customer — when they need to hear a real person owns their problem — a person takes the conversation. The agent’s job is to make that moment arrive fast, facts collected and appointment real. It is not to be that moment.

    The failure shape

    Hand an agent the whole seat — the questions and the judgment both — and it will qualify everybody. It will book the estimate outside your service area because nothing told it distance matters. It will fill the calendar with meetings a person would have filtered in thirty seconds — and once your techs learn to distrust anything the system books, you have automated the exact problem you bought it to solve. The bad version doesn’t fail loudly. It fails as a slow leak — wasted drive time, stood-up appointments — until somebody turns it off and announces that AI doesn’t work.

    The seat works when its boundary is drawn, not when the model is smart. A boundary is a list: what it can ask, what it can book, what it can never promise, and where it stops and hands over. Dapta’s own framing stops at answer, qualify, schedule — first response, not the close. That is not a limitation. That is the product, correctly shaped.

    How can AI handle after-hours calls for a service business?

    For a service business, handling after-hours calls with AI is a matter of strict boundary design rather than conversational cleverness. When a pipe bursts or an AC fails after the office closes, an AI voice agent can own the front half of the intake seat outright: answering instantly at 2 a.m. in the caller’s language, running the structured qualifying questions (what happened, where is the property, when did the damage start, who is paying, and how to reach them), and booking a confirmed appointment on a live calendar instead of leaving an empty callback promise.

    What the AI agent cannot own is the judgment. As detailed in What the human keeps, human dispatchers must retain pricing, arrival promises (“we’ll have a crew there by nine”), and the decision of whether a job fits current crew capacity and service territory. The danger, as shown in The failure shape, is handing an agent both the script and the judgment: without hard boundaries, it books calls outside your territory, clogs the schedule with low-margin jobs, and creates a slow leak of wasted drive time until crews distrust the system. An effective after-hours setup treats the agent as first response—capturing verified facts and locking in the calendar slot—and hands off the data to an automated operational workflow so morning crews arrive with complete context.

    Hire it like a person

    You would never hand a new receptionist the phone on day one with no script, no boundary, and no “if it’s this, come get me.” An agent seat deserves the same three things, because it is the same hire: a script it runs word for word, a boundary it does not cross on its best day, and an escalation path that puts a human in the conversation the moment judgment enters.

    The $6 million is a bet that every small business staffs this seat the way it eventually got a website. The operators who get it right will not be the ones with the smartest agent. They will be the ones who wrote the seat down — job, boundary, and the name of the human who owns the judgment — before they let anything answer their phone.

    Sources: “Dapta Raises $6M to Accelerate the Adoption of AI Sales Agents for SMBs across the U.S. and Latin America,” Dapta LLC via GlobeNewswire, October 13, 2026 — globenewswire.com. User, signup-rate, and revenue figures are Dapta’s own statements in that release.

  • Desktop, Cloud, or API: One Cursor Harness, Three Seats

    Desktop, Cloud, or API: One Cursor Harness, Three Seats

    The seat is a routing decision

    Cursor is not one tool in this shop. It is three seats that share a harness — a Desktop app on a laptop, cloud agents behind an API, and a phone — and picking the wrong one wastes the day in a characteristic way. Desktop waits on a human. The cloud knows nothing about your logged-in life. The phone is a steering wheel, not a workbench. The routing rule we run: the seat is decided by where the work’s context lives.

    Desktop wins when the unlock is a logged-in browser

    A Desktop sitting is a supervised run: a written work order, an executor (Composer 2.5 in our shop) that makes an unaided first pass in its own words before any templating, and a human in the room for the real gates. Its superpower is everything the laptop already is — signed-in dashboards, the WordPress admin, the GA4 Advisor, the files on disk. When a fleet site needs hands inside a web UI that only exists behind Will’s login, no API lane substitutes.

    We learned the boundary the hard way. On October 18, an executor stopped cold at a site’s login redirect and left the step undone — which cost that sitting its unlock path. The standing rule now: in a sitting on Will’s own laptop, the executor clicks through sign-ins using the browser’s saved session. Saved-login only. No saved login means the step logs a HOLD — nobody hunts a password, ever.

    Desktop’s honest cost: it waits on a human, and a queued sitting is not a running job. Our TM pricing diagnosis (WO-353) has sat untouched since October 18 at 6:16 PM — steps written, folder staged — because a sitting starts when Will sits down. Anyone who tells you their Desktop pipeline is autonomous has not watched a folder sit still for a day.

    The cloud wins when the work fans out

    Cloud agents, launched from this desk over api.cursor.com, are the opposite shape: no login, no laptop, no waiting room — and they multiply. On October 16 we ran a census through that API: 44 models, one identical brief, five comparison rounds, graded on their execution trajectories. That is a shape of work a single Desktop seat cannot do before lunch, and it set our model routing for everything since.

    The scale shows in the bookkeeping: our completion watcher has seen 1,431 cloud agents come and go. The API runs two launch shapes for us — repo tasks that come back as pull requests, and no-repo research and evaluation runs with the model pinned in the launch body. The API is beta, and we treat it that way: when an endpoint changes shape, the error gets surfaced plainly, never papered over with invented fields.

    Mobile wins when the human is the bottleneck

    The phone’s job is judgment at human speed. Will reviews runs, reads results, and steers from his phone while the machines work elsewhere — and because a review that evaporates is a review wasted, a distiller folds his mobile and Desktop runs into the shop’s memory and Feed every 20 minutes. Mobile is not a production surface here. It is where the approvals live, and approvals are the scarcest resource in the building.

    What each seat’s failure taught the kit

    • Desktop: permission prompts were taxing the one human in the loop — fixed by a Step 0 permissions pre-flight in every work order (October 18): the executor declares the commands it expects up front, the human approves once, and prompts are reserved for sends, deletions, money, and credentials — which are never pre-approved.
    • Cloud: a finished summary is not evidence — fixed by trajectory grading and a completion watcher: results are verified at the receiving end, or they are not results.
    • Mobile: steering from the phone only counts if it lands in the record — fixed by the 20-minute distiller.

    One harness, three seats, and a routing rule a shop can steal: if the context is a login, sit down at Desktop. If the work fans out, launch it in the cloud. If the blocker is a decision, put it on the phone.

    Method & settings — this piece

    Evidence: This shop’s work-order record, October 2026: WO-198 standard, the WO-343 and WO-348 sittings, the WO-353 stall, the October 16 cloud census

    Desktop: Cursor 3.25.9 on Will’s laptop (read from the install, October 19, 2026); executor seat Composer 2.5

    Cloud: api.cursor.com/v1 (beta); completion watcher count 1,431 agents seen as of October 18, 2026

    Mobile: Cursor mobile app in a review/steer role; runs distilled into memory + Feed every 20 minutes

    Filed under the Cursor desk — where the catalog and the changelog are watched daily.

  • The Operator Is the Harness

    Somebody I follow posted a testing result this week that most people scrolled past. His benchmark harness plants defects in a codebase, lets an agent fix them, and then checks something subtler than whether the defect went away: whether the agent broke something else on the way through. A changed defect signature, he pointed out, is not proof of a correct fix. The bug can change shape and still be a bug. What catches that isn’t a better model. It’s the harness — the checks around the work.

    He’s right, and he’s also only halfway to the point. The harness matters. But who built the harness? Who decided what gets checked, what gets automated, and what never gets automated at all? That’s not the model. That’s the operator.

    I run a small fleet of websites and a stable of AI agents, and I test models the way you’d test trucks: same load, same road, see what happens. This month I ran one identical production brief through 44 different models — flagships, mid-tiers, cheap ones, open ones. Five rounds of comparisons. The results rearranged everybody’s assumptions, including mine. A mid-priced model I’d never heard anyone brag about did the best first-pass work on the board, at a fraction of the flagship cost. Models with famous names flopped on judgment calls. The leaderboard the internet argues about had almost nothing to do with the leaderboard in my shop.

    But here’s the part that actually matters: the brief was the same for all 44. Same instructions, same constraints, same definition of done. The spread in results wasn’t really a spread in models. It was a spread in how much each model needed the operator to have thought clearly. The ones that did well did well because the work was set up so that doing it well was the path of least resistance — the mission was written down completely, the boundaries were structural instead of hopeful, and the verification at the end didn’t care how confident the answer sounded.

    I grade my agents the way that benchmark grades code: not by the summary they hand me, but by the steps they actually took. I built a scorer that reads the execution record — what the agent went and looked at, how many sources it actually consulted, whether the answer was built on evidence or generated from memory with good posture. It has caught polished, lengthy, completely confident answers that were built on zero consulted sources. Zero. The answer read beautifully. It was worthless. The model didn’t fail there; the operation almost did, because a summary is the easiest thing in the world to fake and I was nearly set up to accept summaries.

    That’s the work nobody posts about. Deciding what the machine is allowed to do alone and what it is never allowed to do alone. In my shop, agents draft, research, extract, and stage. They do not touch money, they do not send anything with a commitment in it, and nothing publishes on an agent’s say-so — there’s a verification pass that reads the actual result at the receiving end, because “the API said 200” has lied to me more than any model ever has. Those rules aren’t prompts. Prompts are suggestions. The rules live in the structure: separate accounts, staged drafts, checkpoints that halt the batch when a check fails instead of improvising past it.

    My dad had a saying: a locked door keeps an honest man honest. You don’t make work reliable by telling the worker “don’t mess up.” You make it reliable by building the room so the mess-up is hard to reach.

    So when people argue about which model is best, I hear a question about drivers asked by people watching cars. If you’re a good driver, you can drive anything. Give me a mid-tier model, a cheap model, a local model running on a beat-up laptop — with a clear brief, structural boundaries, and honest verification, I’ll out-produce a careless operator running the flagship. Not because the model doesn’t matter at all. It matters the way a truck matters: payload, turning radius, fuel. But the truck has never once decided where the load needed to go.

    The model is rented. It changes every few weeks and the prices change faster. The operation — the judgment about what to automate, the discipline of verifying at the receiving end, the memory of every way it has gone wrong before — that’s the asset, and it compounds. Everybody I know is shopping for a better engine. The race is being won by the people building better drivers.