Agent-to-Physical Orchestration

Market landscape, the actuation graveyard, and why vertical-first wins — with implications for FieldCheck
25 FEB 2026
Agentic AI TAM
$10.75B
2025, 36.7% CAGR1
FSM Market
$5.1B
Field Service Mgmt 20252
Death Metric
Exceptions
Human ops per failed task
Graveyard
5 dead
FB M, Magic, GoButler, Operator, Saiga
Verification
Moat
Camera-in-loop = trust

I. The Thesis

AI agents are confined to the digital realm. The agentic AI platform market is $10.75B in 2025, projected to $51.26B by 2030 (36.7% CAGR).1 But agents lack two things to act in the physical world: (1) secure payment rails and (2) trusted physical actuators — the ability to order a service and verify it was completed.

Big Tech: payment rails, not fulfillment. OpenAI ACP, Stripe Agentic Suite, and Google AP2 solve how agents pay. None solve how agents verify that a physical task was completed. That gap is where startups die.

Nobody has solved the actuation and verification layer as a horizontal platform — and there is a graveyard of companies that tried.

The question is not whether agents will need to trigger physical actions. The question is how — horizontal API or vertical-first with verification.


II. Market Sizing

LayerSizeSourceNotes
Agentic AI Platforms$10.75B → $51.26B (2030)Mordor Intelligence1All agent use cases, not just physical
Field Service Management$5.1B → $9.17B (2030)MarketsAndMarkets212.5% CAGR. Pest control, HVAC, cleaning, inspections
B2B Corporate GiftingSendoso $84.8M revLatka3Proves software→physical API generates revenue
Agent Companies1,700+ globallyCB Insights4Only a fraction need physical-world APIs

Developer Demand (Sourced)

“If each step in an agent workflow has 95% reliability, 20 steps = 36% success rate. This isn’t a prompt engineering problem. This is mathematical reality.” — Utkarsh Kanwat5

“68% of production agents execute at most 10 steps before human intervention. Reliability is the #1 development challenge.” — Study of 306 practitioners6

“The dirty secret of every production agent system is that the AI is doing maybe 30% of the work. The other 70% is tool engineering.” — Kanwat5

“Physical world actions are the hardest. Every tool call that touches real things has a failure mode that LLMs can’t retry.” — HN discourse on agentic systems


III. Competitive Landscape

3a. Protocol Layer (Big Tech — Payments Only)

CompanyWhatPhysical Fulfillment?
OpenAI ACPOpen-source protocol (w/ Stripe) for agent commerce7No — merchants handle
Stripe Agentic SuiteSingle API for merchants to connect with AI agents8No — merchants handle
Google AP2 + A2ACryptographically-signed payment Mandates9No — protocol only
Zomato MCPProduction MCP for end-to-end food ordering10Yes — Zomato’s fleet

3b. Agent-to-Physical Connectors (Proven Revenue)

CompanyModelRevenue / Scale
Thumbtack × OpenAIChatGPT app for home services11300K service pros
SendosoB2B gifting API + SaaS ($12K–67K/yr)3$84.8M rev, $157M raised
DoorDash DriveDelivery-as-a-service API12$9.75 base (0–5mi), massive US coverage
Florist One APIREST API, 20% commission1310K+ florists, $40M total sales since 2009
ReachdeskB2B gifting + direct mail APISendoso competitor, SaaS + per-item model
Instacart ConnectGrocery delivery API for brandsWhite-label fulfillment, restaurant/retail

3c. Direct Horizontal API Competitors (Early / Unproven)

CompanyModelRisk
LocalServiceAPIAPI for local servicesSame graveyard pattern — supplier coordination
HumanOpsHuman-in-the-loop APIHuman cost = margin erosion
CallioAI + human conciergeUnproven at scale

3d. Vision AI for Field Verification (FieldCheck’s Peer Set)

CompanyFundingWhat
Spot AI$93M14Video AI Agents for the physical world. Camera → proactive problem identification.
FYLD$41M15AI for field ops risk + digital work execution. 30K-40K field workers.
TechSeeEstablishedAI visual verification for field services. Reduces manual QA by 95%+.
Deepomatic€10M16Visual automation for telecom field workers. Task completion verification.
Zeitview$60M17AI inspections for infrastructure. 200K+ assets across 80 countries.

3e. Voice AI — The Stealth Competitor

Bland AI ($65.5M)18 bypasses supplier onboarding entirely by making phone calls to existing booking systems. No API integration, no supplier dashboard. If your agent can just call the plumber, why does it need an API?

3f. The Startup Graveyard

CompanyRaisedWhy They Died
Facebook MFB-fundedUniversal request fulfillment (2015–18). Human contractors 24/7 = uneconomical.19
Magic$12M (Sequoia)SMS concierge. Broke even on 50%, lost money on 25%. Pivoted to $100/hr.20
GoButlerSignificantUniversal text concierge. Pivoted to automated flights. Died.21
OperatorUber co-founderConversational commerce. Human coordination cost too high.22
Saiga€3M“Fix the concierge model” with AI/RPA (2020). No news after 2022.19
The pattern is always the same. Every company that tried a universal physical concierge API died identically: (1) human coordination cost exceeded markup, (2) AI automation progressed slower than hoped, (3) long-tail services couldn’t be standardized. Every survivor went vertical — Sendoso (B2B gifts), DoorDash Drive (delivery), Thumbtack (home services).

IV. Unit Economics & The Death Cost

The viability of any agent-to-physical platform hinges on one variable: exception handling rate. When an agent triggers a physical service and the provider ignores, drops, or botches it, a human must intervene. That intervention cost eats the margin.

ScenarioException RateAvg FeeOps Cost / ExceptionGross Margin
Optimistic10%$12$596%
Realistic30%$10$876%
Pessimistic50%$8$1225%
Death60%+$8$15Negative

COGS Breakdown (Typical Horizontal API Order)

Component% of RevenueWho CapturesNote
Platform fee (markup)15–25%API companyOnly viable if exceptions <20%
Supplier execution50–70%ProviderPest control, cleaning, delivery
Human ops (exceptions)10–40%BurnedChase providers, resolve disputes
Payment rails2–3%Stripe, etc.Big Tech owns this layer

At 50% exception rate, human ops exceed platform fee. Margin collapses.

Magic learned this: they broke even on 50% of orders and lost money on 25%.20 Sendoso and Reachdesk solve this by charging $12K–67K/yr in SaaS fees on top of per-item costs — the SaaS fee subsidizes the human coordination layer.3

The exception rate is the only variable that matters. A horizontal platform with scraped supplier databases will have exception rates above 50%. A vertical platform with opted-in, camera-verified providers can push it below 10%. This is the structural advantage of vertical-first.

V. Implications for FieldCheck (Blue-Collar AI)

FieldCheck is uniquely positioned in this space. Unlike horizontal concierge APIs that died trying to coordinate unverified suppliers, FieldCheck already has the camera-based verification layer that makes agent-to-physical orchestration actually work.

The agent-to-physical orchestration space has a fatal gap: nobody can verify that the physical task was actually completed correctly. OpenAI ACP handles payment. Thumbtack handles matching. DoorDash Drive handles delivery. But none of them can confirm that the pest control technician actually treated every corner, or that the cleaner actually scrubbed the kitchen. FieldCheck can.

The FieldCheck Advantage

Problem (Horizontal APIs)FieldCheck’s Solution
Scraped supplier DBs — providers don’t respondOpted-in providers using the FieldCheck app. Camera verification = accountability.
Exception rate >50% kills marginVisual verification closes the loop. Agent can see the work was done via camera feed.
No quality signal — was the job done well?Gemini/VLM analysis of captured images provides automated QA scoring.
Human ops needed to chase providersWorker app with real-time guidance eliminates coordination overhead.

Concrete Use Case: Agent Orders Pest Control

Today, if an AI agent managing a building wants to order pest control, it hits a wall — it can find a provider (Thumbtack) and maybe pay (Stripe), but it cannot verify completion. With FieldCheck in the loop:

  1. Agent triggers order via FieldCheck API (specific provider, specific task, specific location)
  2. Provider receives job via FieldCheck worker app with AI-guided instructions
  3. Provider captures verification — photos/video of each treated area, analyzed by VLM in real-time
  4. Agent receives confirmation with visual proof + QA score. Payment released on verification.

This is the closed-loop agent-to-physical pipeline that the horizontal API companies couldn’t build. FieldCheck already has steps 2–3 (worker capture + VLM verification). Steps 1 and 4 are API wrappers.

Why FieldCheck Avoids the Graveyard

Graveyard PatternWhy FieldCheck Is Different
Horizontal = infinite categories, no expertiseVertical-first (pest control, cleaning, inspections). Master one before expanding.
No supplier onboardingProviders actively use the app for AI guidance. They’re opted-in.
No verification = no trust = no repeatCamera verification IS the product. Trust is built into every job.
Human ops eat the marginVLM replaces human QA. Exception rate pushed below 10%.
Facebook M: “learning rate slower than hoped”FieldCheck doesn’t need general AI. Needs domain-specific VLM for specific tasks (pest traps, cleaning zones).

The YC Signal

YC Spring 2026 RFS explicitly calls for “AI Guidance for Physical Work” (David Lieb): “Imagine wearing a small camera while an AI sees what you see and talks you through the job. Instead of needing months or years of training, workers can become effective immediately.”23 This is exactly FieldCheck’s product.

The Expansion Path

PhaseWhatAgent Integration
NowManager capture + worker guidance for repair.sg (pest control)None — human-triggered
Phase 2FieldCheck API: external systems can trigger jobs programmaticallyBuilding management systems, property tech
Phase 3MCP Server: AI agents can directly order verified physical servicesAny AI agent can order + verify pest control, cleaning, inspections
Don’t build the API layer prematurely. The graveyard is full of companies that built the API before proving the fulfillment loop. FieldCheck’s correct sequence: (1) prove camera-verified fulfillment with repair.sg, (2) prove it generalizes to cleaning + NinjaVan, (3) THEN expose as API/MCP for agent consumption. The verification layer is the moat — the API is a wrapper.

For FieldCheck

  • Camera verification solves the single question that killed every horizontal API: “Was the job done?”
  • Vertical-first (pest, cleaning) avoids graveyard pattern before API exposure
  • YC RFS aligns: “AI Guidance for Physical Work”
  • SaaS + per-job revenue model (Sendoso/Reachdesk) is proven

Against / Red Team

  • Bland AI bypasses APIs via voice; could cannibalize need for structured APIs
  • Big Tech could add verification to Thumbtack, DoorDash if they see the gap
  • repair.sg must prove the loop before any agent integration matters
Red Team check. The biggest risk is building the API before proving fulfillment. The second-biggest: Thumbtack/Spot AI add verification and move downmarket. FieldCheck’s defensibility is speed — own the pest-control + cleaning verification loop before anyone else.

Verdict

STRONG for FieldCheck — FATAL for Horizontal APIs

The agent-to-physical orchestration space is real ($10.75B agentic AI, $5.1B field service management) and growing fast. But the startup graveyard is unambiguous: Facebook M, Magic, GoButler, Operator, and Saiga all tried horizontal physical concierge and died. The pattern is always human coordination cost exceeding markup.

FieldCheck inverts the graveyard pattern. Instead of building an API layer hoping suppliers will respond, FieldCheck starts with the hardest part — camera-verified physical task completion — and can later expose it as an API. The verification layer is the moat. Every horizontal API company died because they couldn’t answer one question: “Was the job actually done?” FieldCheck can.

Sequence: (1) Prove repair.sg pest control loop. (2) Generalize to cleaning + NinjaVan. (3) Expose FieldCheck API for building management / proptech. (4) Ship MCP server so any AI agent can order verified physical services. Don’t skip to step 4.


References

[1] Mordor Intelligence — Agentic AI Platform Market (2025–2030). $10.75B (2025), 36.7% CAGR to $51.26B
[2] MarketsAndMarkets — Field Service Management Market. $5.1B (2025) → $9.17B (2030), 12.5% CAGR
[3] Latka — Sendoso Revenue (2024). $84.8M rev, ~500 customers, $157M raised
[4] CB Insights — AI Agent Market Map (2025). 1,700+ companies, 400+ promising
[5] Utkarsh Kanwat — “Betting Against Agents”. Reliability math, production realities
[6] “Measuring Agents in Production” (Dec 2025). 306 practitioners, 26 domains, 68% ≤10 steps
[7] OpenAI — Agentic Commerce Protocol (ACP). Open-source, powers ChatGPT Instant Checkout
[8] Stripe — Agentic Commerce Suite (Sep 2025). Merchant-agent connection layer
[9] Google — Agent Payments Protocol (AP2). 60+ partners, Mastercard, PayPal, Visa
[10] Zomato MCP Server. Production MCP → food ordering
[11] Thumbtack × OpenAI (Oct 2025). 300K service pros, sole home services ChatGPT partner
[12] DoorDash Developer — Drive API Pricing. $9.75 base (0–5mi)
[13] Florist One — Flower Delivery API. 10K+ florists, $40M sales, since 2009
[14] Spot AI — Video AI Agents ($93M). Camera → proactive problem identification
[15] FYLD — $41M Series B (Feb 2026). 30K-40K field workers, 82% YoY growth
[16] Deepomatic — €10M Series B. Visual automation for telecom field workers
[17] Zeitview — $60M (Mar 2025). 200K+ assets inspected across 80 countries
[18] Bland AI — Voice AI Phone Agent. $65.5M raised, bypasses API integration via voice
[19] TechCrunch — Saiga / concierge graveyard (2022). Facebook M shutdown, concierge failure modes
[20] VentureBeat — Magic pivot to $100/hr (2016). 50% break-even, 25% loss
[21] TechCrunch — GoButler Pivot (2016). Concierge graveyard
[22] TechCrunch — Operator (2015). Uber co-founder, conversational commerce failure
[23] YC Spring 2026 RFS — “AI Guidance for Physical Work”. David Lieb, camera + AI coaching for field workers