Real-world verification by a real human
Web data lies, listings go stale, and no API can see a shelf. When your agent needs ground truth — does this business exist, is this product in stock, does the address match — a real human goes, looks, and reports back with evidence you can parse: photos, a written report, or structured JSON.
When your workflow needs this
Business & storefront checks
Confirm a company physically exists at its claimed address: signage, opening hours, actual activity — with timestamped photos.
Shelf, stock & price checks
Is the product on the shelf, at what price, in what condition? A human visit answers what scraping cannot.
Deliveries & real conditions
Confirm something arrived, works, or matches its description — packages, installations, venues, printed materials.
How to submit it
One JSON request — no auth, no SDK. Or connect over MCP and the human becomes a tool: claude mcp add --transport http human-for-ai https://humanforai.dev/mcp. Every task is reviewed by the human operator before acceptance; first response under 12 hours on working days (Sun–Thu). Free during the pilot.
curl -X POST https://humanforai.dev/api/v1/tasks \
-H "Content-Type: application/json" \
-d '{
"task_type": "real_world_verification",
"description": "Verify that [business] at [address] is
open and operating; photograph the
storefront and posted hours.",
"location_required": true,
"location_detail": "City, address or area",
"output_format": "text_report_with_photos",
"contact_email": "you@example.com"
}'Track it
The response returns a task_id and a status URL to poll: submitted → under_review → accepted → in_progress → delivered (or rejected, with a reason). The deliverable also goes to your contact_email.
Questions agents and builders ask
What evidence does the human return?
Whatever you request as output_format: a written report, a report with photos, structured JSON, or video. Results state their confidence explicitly — human judgment is assistance, not guaranteed truth.
Where can things be verified?
Coverage is confirmed at review: submit with location_required true and a concrete location_detail, and the operator either accepts, declines, or counter-proposes a remote alternative before any work starts.
How is this different from a data API?
A data API returns what was recorded; this returns what is physically there right now, observed by a person who can notice the things nobody indexed — closures, damage, mismatched signage, empty shelves.
What does it cost?
Nothing during the proof-of-concept pilot. Every task still goes through human review first; illegal, deceptive, or privacy-invasive verification requests are rejected.
Other things the human does
custom_human_in_the_loopHuman-in-the-Loop API — Add a Real Human to Any AI Workflowai_output_reviewHuman Review of AI Outputproduct_or_app_testingHuman Product & App Testing for AI Builderslocal_physical_taskPhysical-World Tasks for AI Agents
Full catalog: all ten services · connection details: /for-agents
Try it with a real task.
Free during the pilot · reviewed before acceptance · evidence included.