New ideaAgent Services

Agent Task Proof

Unknown

3 weeks ago · Scored 21 Oct 2026 · ID HBV-261021-H10

Summary

BriefingHypothesis

BET / HYPOTHESIS. RentAHuman (launched Feb 1 2026, YC Spring 2026) lets agents hire humans for real-world tasks: 518,284 humans in two weeks (WIRED), 700k+ sign-ups, $2.1M+ paid to humans (site). Yet a survey of about 800 offers found 87% were posted by humans managing bots and only 9% by pure agents (SondeAR/MIT researcher), and tasks are mostly photos, walks and in-person checks, which are hard for an agent to verify.…

Read the rest

Agent Task Proof is the missing verification layer: a vision-and-metadata checker that returns a verdict and a signed receipt via one API call, paid per verification. Primary users are agents (MCP/REST); humans are workers and task owners. Benchmark: RentAHuman.

DescriptionVerifiedHypothesis

checks: image quality, location/time consistency, duplicate/AI-generated evidence heuristics, spec match via a vision model; returns pass/fail/needs-more with reasons and a signed receipt the agent can attach to a payment release or dispute. Works with any marketplace (RentAHuman, TaskRabbit-like, custom) through MCP and REST.…

Read the rest

Pricing: per verification ($0.05-$0.50, Hypothesis) plus optional escrow-release automation fee. Faceless growth: MCP directory listing, docs-led SEO/GEO ('verify task completion for AI agents'), integration guides for agent frameworks. Global Application: any agent framework and marketplace. LatAm Version: PT-BR/ES evidence reading (receipts, signage, storefronts) and Pix/local-wallet payouts. This is a bet: the market for agent-hired human tasks is new and measured mostly by sign-ups, not verified completed jobs.

Target customer

Agent builders and marketplaces that let agents hire humans for physical tasks, worldwide

Pain

Agents cannot verify physical-world work; fake or low-quality evidence; payment disputes

Why this score

Why it scored lower: Young market and native verification

MoatHypothesis

Verification corpus + receipts. Accumulated fraud patterns, labeled evidence and a neutral signed-receipt format. Marketplaces can copy; Status: Hypothesis.

Demand signals

X posts (7 days)

n/a

Signal trend

n/a

Web demandSelf-reported

RentAHuman: 518,284 humans (WIRED); $2.1M+ paid (site, Self-reported); 87% human-managed offers (survey).

Market and competitors

TAMHypothesisEstimate

$1B (Hypothesis) — Global verification/escrow fees on agent-hired human tasks. Method: Hypothesis; market is new. Sources: Hark estimate (Hypothesis).

SAMHypothesisEstimate

$20M (Hypothesis) — Agent-builder and marketplace verification spend within 2-3 years. Method: Hypothesis. Sources: Hark estimate (Hypothesis).

SOMHypothesisEstimate

$24K ARR (Hypothesis) — 12 months. Method: ~50 integrators x ~$40/month average. Low confidence. Sources: Hark estimate (Hypothesis).

Competitors

Marketplace-native checks, manual review, generic vision APIs

Business model

Monetization

Per-verification API fee; optional escrow-release automation fee

Revenue hypothesis (monthly)Hypothesis

Base ~$2K/month, conservative ~$300/month at 12 months (Hypothesis).

Jobs to be done

When my agent hires a human for a physical task and cannot tell if it was done, I want an automatic, auditable verdict on the evidence, so I can release payment or dispute without me looking at photos.

HXC

Agent builder · 29 · Lisbon: Indie developer whose personal agent hires humans for errands and checks. Frustration & buying signal: Frustrated that photos come back wrong or faked; would pay cents per check to release payment safely.

Build and go-to-market

Tech stack

Cursor + Vercel + GitHub cesarx2 (one repo); Supabase; Stripe

GTM channels

MCP directories, docs-led SEO/GEO for agent builders, integration guides; no founder face, no sales team, no paid ads

Risks

Main risk

Young market and native verification

All risks

Value Risk: Attention — Marketplaces may verify natively; volumes unknown. Next: Interview 5 builders; check native features. | Usability Risk: Supported — One API call and a receipt. Next: None critical. | Feasibility Risk: Attention — Fake/AI-generated evidence is an arms race. Next: Start with spec-match + metadata checks. | Viability Risk: Elevated — Young market; thin margins. Next: Bet-sized; kill fast.

Next validation step

5 interviews with agent builders; check RentAHuman verification features

Sources

rentahuman.ai https://www.wired.com/story/ai-agent-rentahuman-bots-hire-humans/ https://sonde.ar/article/rentahuman-puts-people-to-work-for-ai-agents https://www.ycombinator.com/companies/rentahuman https://glonce.com/stripe-says-ai-agent-purchases-on-link-grew/

agent-servicebenchmark:RentAHuman, verification, escrow, mcp, bet, hypothesis, idea, new

Want the next one the day it is scored?

The score, the sources, the risks and the next step, by email.

ProductBots.ai

Like this idea? Turn it into a plan with ProductBots.ai.

ProductBots.ai is an Agentic Product OS: agents that do the product work with you, from discovery to plan. Download the briefing for this idea and use it as the starting point.

Open ProductBots.ai

Preview the briefing

# Briefing: Agent Task Proof

Source: Startup Ideas HQ (HB Ventures), found 2026-10-21. ProductBots Opportunity Score: 60/100, evidence confidence 30%.
Evidence labels (Verified, Self-reported, Hypothesis) are kept as written by the researcher.

## Goal for ProductBots.ai
Turn this idea into a validated MVP plan and a first build. Start from the next validation step; do not assume unlabelled numbers are verified.

## What it is
unknown

BET / HYPOTHESIS. RentAHuman (launched Feb 1 2026, YC Spring 2026) lets agents hire humans for real-world tasks: 518,284 humans in two weeks (WIRED), 700k+ sign-ups, $2.1M+ paid to humans (site). Yet a survey of about 800 offers found 87% were posted by humans managing bots and only 9% by pure agents (SondeAR/MIT researcher), and tasks are mostly photos, walks and in-person checks, which are hard for an agent to verify. Agent Task Proof is the missing verification layer: a vision-and-metadata checker that returns a verdict and a signed receipt via one API call, paid per verification. Primary users are agents (MCP/REST); humans are workers and task owners. Benchmark: RentAHuman.

## Customer and pain
Target customer: Agent builders and marketplaces that let agents hire humans for physical tasks, worldwide
Pain: Agents cannot verify physical-world work; fake or low-quality evidence; payment disputes

## Market and competitors
TAM: $1B (Hypothesis) — Global verification/escrow fees on agent-hired human tasks. Method: Hypothesis; market is new. Sources: Hark estimate (Hypothesis).
SAM: $20M (Hypothesis) — Agent-builder and marketplace verification spend within 2-3 years. Method: Hypothesis. Sources: Hark estimate (Hypothesis).
SOM: $24K ARR (Hypothesis) — 12 months. Method: ~50 integrators x ~$40/month average. Low confidence. Sources: Hark estimate (Hypothesis).
Competitors: Marketplace-native checks, manual review, generic vision APIs
Geography: Global except China

## Business model
Monetization: Per-verification API fee; optional escrow-release automation fee
Revenue hypothesis (monthly): Base ~$2K/month, conservative ~$300/month at 12 months (Hypothesis).

## Build and go-to-market
Build effort (agent-weeks): 2
Time to first revenue: 3-4 weeks
Tech stack: Cursor + Vercel + GitHub cesarx2 (one repo); Supabase; Stripe
GTM channels: MCP directories, docs-led SEO/GEO for agent builders, integration guides; no founder face, no sales team, no paid ads

## Risks and next step
Main risk: Young market and native verification
Next validation action: 5 interviews with agent builders; check RentAHuman verification features

## Sources
https://rentahuman.ai/ https://www.wired.com/story/ai-agent-rentahuman-bots-hire-humans/ https://sonde.ar/article/rentahuman-puts-people-to-work-for-ai-agents https://www.ycombinator.com/companies/rentahuman https://glonce.com/stripe-says-ai-agent-purchases-on-link-grew/

Existing businessAgent Services· 3 weeks ago
CreativeClaw-style copycat
CreativeClaw is a remote MCP server that gives Claude (and other agents) one creative studio across the top image, video, audio and 3D models; bootstrapped, one person, it did $100,077 in the last 30 days (Verified, Polar) after $5,254 in August, selling credits ($10 = 1,000) with no subscription.
62/100
confidence 55%
Why it scored lower: Margin and spike risk
Open ProductBots.ai →
New ideaAgent Services· 3 weeks ago
Brazil Data for Agents
A pay-per-call API and MCP server that gives AI agents clean, cached Brazilian data (CEP, CNPJ, holidays, FIPE) with agent-friendly schemas, so agents never scrape or guess.
61/100
confidence 28%
Why it scored lower: Agents may not pay
Open ProductBots.ai →
New ideaAgent Services· 3 weeks ago
Agent Handoff Desk
An agent-facing MCP/API where a personal or business AI agent that gets blocked on a website or needs a phone call hands the task to a vetted remote human operator, who completes it under the user's approval and returns structured proof — the digital 'last mile' counterpart to RentAHuman.
61/100
confidence 45%
Why it scored lower: Trust & safety incidents and RentAHuman/agent vendors moving into digital tasks
Open ProductBots.ai →

Research, not investment advice. Numbers keep the label they were published with.