AI Agents
Submit a startup, programmatically
If you're an AI agent (or a human who scripts things), you don't need the web form. One JSON POST creates a submission that enters the same human-review queue.
The request
POST https://neuronfeed.com/api/submit
Content-Type: application/json
{
"name": "Example AI",
"website": "https://example.ai",
"tagline": "Voice agents for clinical intake",
"description": "Example AI builds voice agents that handle patient intake calls for clinics. The product transcribes, triages, and books appointments across EHR systems...",
"hqCountry": "US",
"foundedYear": 2025,
"contactEmail": "[email protected]"
} Field rules
| Field | Required | Constraints |
|---|---|---|
name | yes | 2–100 chars |
website | yes | valid URL, must resolve — dead sites are rejected in review |
tagline | yes | 5–160 chars, one plain sentence, no marketing superlatives |
description | yes | 50–2000 chars of factual product description |
hqCountry | no | ISO-2 code (US, GB…) |
foundedYear | no | 1990–2030 |
contactEmail | yes | reachable inbox — used for review questions only, never published |
Responses
200— accepted into the review queue:{ "message": "..." }400— validation failed; the body lists each failing field underissues[]. Fix and retry.429— more than 3 submissions from one contact email in 5 minutes. Back off.
Rules that keep the queue clean
- Check for duplicates first. Query
GET /api/v1/startups(or search) for the company name before submitting — duplicate submissions are rejected and burn reviewer time. - Facts only. Descriptions are edited to house style; invented funding numbers or fake traction get the whole submission binned.
- One product per submission. Batch imports should be individual POSTs, spaced out.
What happens after
Same pipeline as the web form: human review within 48 hours, then enrichment (categories, funding history, agent-readiness scan) and publication. Approval is editorial — listing is free, and there's no way to pay for placement.
Reading this as an agent?
This page is summarized in llms.txt. The directory is also queryable over MCP at https://neuronfeed.com/api/mcp.