IDEASBERG_

INDEX / MARKETPLACE

VERDICT: BUILDBERG SCORE 71/100

HOW THIS WAS SCORED → · ANALYZED OCT 26, 2026

Real-World Work API (Agent-to-Human Last Mile)

A vetted specialist marketplace that exposes real-world human workers (permit expediters, custom brokers, handymen) as callable APIs so AI agents like Dots can hire them programmatically.

01 THE IDEA

As personal agent platforms like Dots handle millions of user requests, many will require physical-world execution that software alone cannot complete. This business aggregates vetted, specialized human workers and wraps their services in a clean API layer, making them discoverable and bookable by AI agents in real time. Think of it as an API-first staffing layer for the agentic economy.

The model earns transaction fees on each job dispatched. Defensibility comes from the vetting quality, response SLAs, and the breadth of specialist categories covered. Early focus on a single niche (e.g., permit expediters in a specific city or customs brokers for a specific trade corridor) would allow rapid quality control before expanding.

02 THE NUMBERS

EXPECTED ARR

$150K – $3M

INITIAL INVESTMENT

$25K + 500h

MONTHLY BURN

$8K + 120h

AUTOMATION

5/10

COMPETITORS

13 · GROWING →

SKILLS

API design, marketplace operations, worker vetting and onboarding, vertical domain knowledge, sales to AI platform integrators

By Berg Score this idea ranks #158 of 1,749 (top 10%) among all scored ideas in the catalog and #3 of 98 (top 4%) among Marketplace ideas.

MARKET

Model estimate (no live research yet): No dominant player has yet built a clean API layer specifically for routing AI agent requests to vetted real-world human specialists. The gap between digital AI capability and physical-world execution is a known bottleneck in agentic AI adoption. Over 5 years, this category is likely to grow substantially as agent platforms proliferate and the 'last mile' problem becomes acute for enterprise buyers.

HOW MUCH CAN BE AUTOMATED

The API routing and matching layer is highly automatable, but worker quality assurance, onboarding, and dispute resolution require significant ongoing human oversight.

SKILLS AND DOMAIN KNOWLEDGE

Understanding a specific regulated or specialized trade (permits, customs, inspections) is essential to build trust with both the AI platforms routing work and the end customers receiving results.

03 THE VERDICT

This is a category that barely exists today but will be urgently needed as agentic platforms scale. First-mover in a specific vertical with strong vetting and fast SLAs could become the default supplier for Dots, Muse, and similar platforms. High operational complexity is the main challenge but also the moat — hard to replicate at quality.

SCORE HISTORY

The Berg Score moved from 74 to 71 (-3) between the first analysis on 2026-09-30 and the latest of 4 rechecks on 2026-10-26.

DATEANALYSISVERDICTBERG SCORECOMPETITORS
2026-09-30First analysisBUILD744
2026-10-05Weekly recheckBUILD746
2026-10-12Weekly recheckBUILD738
2026-10-19Weekly recheckBUILD7210
2026-10-26Weekly recheckBUILD7113

WHAT THE 2026-10-26 RECHECK FOUND: Three new entrants added as the human-in-the-loop API space continues attracting founders riding the agentic AI wave; berg score trimmed by 1 point to reflect slightly increased competitive density, though no fundamental shift in market dynamics or ARR range is warranted yet.

Verdict: BUILD. Don't have the ~500 hours it takes? Get matched with a vetted builder who does — we review every brief by hand and intro you to up to 3 builders.

FIND A BUILDER →

INTROS ONLY — NO FEES, NO ESCROW. THE PROJECT IS YOURS.

ALREADY BUILT — BY THE COMMUNITY

I BUILT THIS →

Nobody has claimed this one yet. Shipped it? Tell the story — every submission is hand-reviewed, and approved builds get listed right here with a link to your product.

04 THE FIELD

13 competitors are on file. The field is growing. The first 10 we tracked are listed below, each linked to the company's own site so you can check it.

  • Twilio (for telecom APIs)est. 2008STEADY · ADDED 2026-09-30

    CATEGORY LEADER IN COMMUNICATION APIS

    Analogous model — wrapping human-adjacent infrastructure in APIs; not a direct competitor but an architectural blueprint.

  • Invisible Technologiesest. 2015GROWING · ADDED 2026-09-30

    NICHE PLAYER IN AI-AUGMENTED HUMAN TASK EXECUTION

    Provides managed human-in-the-loop task execution for enterprises, directly competing in the AI agent fulfillment space.

  • Taskusest. 2008STEADY · ADDED 2026-09-30

    MID-MARKET PLAYER IN OUTSOURCED DIGITAL SERVICES

    Provides outsourced human task execution but lacks AI-native agent API integration.

  • Scale AIest. 2016GROWING · ADDED 2026-09-30

    LEADER IN AI DATA LABELING AND HUMAN-IN-THE-LOOP

    Manages human workforce for AI tasks; could pivot toward agent-callable real-world work APIs.

  • Mechanical Turk (Amazon)est. 2005DECLINING · ADDED 2026-10-05

    LEGACY CROWDWORK PLATFORM WITH NO AGENTIC API FOCUS

    Still relevant as a conceptual predecessor but has not pivoted toward AI-agent-callable workflows, making it a weak competitive threat.

  • Toloka AIest. 2014GROWING · ADDED 2026-10-05

    HUMAN-IN-THE-LOOP DATA AND TASK PLATFORM GAINING AI INTEGRATION TRACTION

    Toloka has been pushing programmatic task dispatch and human-in-the-loop APIs, making it a plausible adjacent competitor as agentic demand rises.

  • Relay.appest. 2022GROWING · ADDED 2026-10-12

    HUMAN-IN-THE-LOOP WORKFLOW AUTOMATION LAYER

    Relay explicitly positions humans as callable steps inside automated workflows, directly overlapping the agent-to-human handoff use case.

  • Doers.doest. 2023NEW · ADDED 2026-10-12

    EARLY-STAGE AI AGENT TASK OUTSOURCING STARTUP

    Pitches itself as an execution layer for AI agents needing human action, a near-direct analog to this concept.

  • Saga (fka WorkOS for agents)est. 2024NEW · ADDED 2026-10-19

    EARLY-STAGE ENTRANT IN AGENT-TO-HUMAN TASK ROUTING

    Emerging player attempting to bridge AI agent orchestration with human-in-the-loop execution, directly competitive with the API-first staffing concept.

  • Fiverr Pro APIest. 2010GROWING · ADDED 2026-10-19

    LARGE EXISTING FREELANCER BASE EXPERIMENTING WITH PROGRAMMATIC ACCESS

    Fiverr has signaled API-layer ambitions for enterprise and agent-driven task dispatch, leveraging its existing vetted Pro tier as a ready supply side.

+3 MORE COMPETITORS + HEAD-TO-HEAD BATTLE PLANSSIGN UP / LOGIN →

FROM THE SAME EPISODE

The same episode pitches 4 ideas in total. This idea scores #3 of the 4. The episode average is 72. Source: OpenAI DevDay: Dots, Agents & $100B Opportunities.

05 RELATED IDEAS, COMPARED

IDEAVERDICTSCOREEXPECTED ARRTO START
This ideaBUILD71$150K – $3M$25K
Solo AI Agent Agency (Productized Service)MAYBE34$180K – $720K$3K
AI Web Agent Task Automation Service for SMBsMAYBE26$100K – $2M$15K
AI-Powered Vertical SaaS for Trade/Local Service BusinessesMAYBE50$120K – $1.8M$8K
AI-Powered Freelance Task Marketplace (Fiverr/Upwork 2.0)SKIP22$120K – $1.8M$15K
ChatGPT Plugin for Real-World Task FulfillmentMAYBE65$80K – $1.2M$8K

MORE LIKE THIS, WEEKLY