Your agent answers fast.It answers like a rookie.
Ask your agent something you know cold. It answers in two seconds, sounds sharp, and skips the first thing anyone who's done the job would say. It read everything written about the work. The part that makes someone good at it never got written down. PostGrad pipes that in, from people who do the work for a living.
Two sides,
one platform.
Pick a feed in the domain your agent operates in.
Paste an API key or your MCP URL. About two minutes.
Ask the same question that made your agent sound like a rookie.
This time it answers like someone who has done it.
Drop in a transcript, a doc, or connect your meeting tool.
We turn it into entries, structured and confidence-scored.
Approve every entry before it ships. Set your price.
You get paid every month agents keep using it.
What a pro
would do
When your agent hits a knowledge gap, the usual fix is to point it at your docs, a web search, or a pile of scraped pages and let it retrieve an answer. Builders call this RAG. It works when the answer is a fact, a price, a date, a definition.
It breaks the moment the answer depends on judgment, the hard-won operator wisdom that never gets written down because the people who have it are too busy doing the work. PostGrad pipes that judgment in, structured and confidence-scored, from people who do the work for a living.
A named person, not a crawler
Operator feeds come from a verified person who does the work. Compiled feeds are aggregated from public sources and labeled as such. The marketplace badge tells you which is which before you connect.
Transparency over gatekeeping
Every feed names the operator behind it. Every entry carries confidence and freshness. Quality compounds as agents query: strong signal rises, weak signal fades.
Built for agents to consume
Categorized, tagged, and confidence-scored. Ready to use without parsing.
Works with any agent
REST API for Python and TypeScript. MCP connectors for ChatGPT, Cursor, Claude Code, and Windsurf. Webhooks for n8n and Zapier.
Current, not stale
Every entry has a last-updated date. Stale entries rank lower, so your agent gets the current take, not last year's.
{
"mcpServers": {
"postgrad": {
"url": "https://postgrad.io/api/mcp",
"headers": {
"Authorization": "Bearer pg_live_..."
}
}
}
}Add to ChatGPT, Cursor, Claude Code, or any MCP-compatible agent. Two minutes to connect.
PostGrad handles delivery, billing, and the agent audience. Your share starts at 70% and rises to 85% as you scale.
You have judgment that took years to earn
Right now it walks out the door with every call. Turn it into a feed agents pay to query, every month it's live.
You control everything
What gets published, what it costs, who can access it. Review and approve every entry before it goes live.
Automated extraction
Connect your meeting tool or upload documents. PostGrad handles structuring and categorization.
Recurring revenue
Subscribers pay monthly. You earn every month your feed is active.
Zero infrastructure
No API to build, no server to maintain, no billing to manage. PostGrad handles delivery.
PostGrad handles delivery, billing, and the agent audience. Your share starts at 70% and rises to 85% as you scale.
Pick the domains your agent operates in
Each feed is curated by people who have done the work, not a model that has read about it. Subscribe to one or stack several. Your agent answers the same way an experienced operator in that domain would.
Live from the knowledge base
Every entry below was published by a domain operator and approved through review before any agent could query it. Counts update as new entries land.
Simple, transparent pricing
Start free. Several feeds are free to subscribe, and the stats and categories endpoints need no API key. Connect, watch your agent change, then pick a plan.
Starter
For wiring one agent to a few feeds and seeing it work.
$10/mo
- 1,000 queries/month
- 20 requests/min burst rate
- All search modes: keyword, semantic, hybrid
- Unlimited feed subscriptions
- REST API + MCP
Pro
For an agent in production that queries all day.
$19/mo
- 10,000 queries/month
- 60 requests/min burst rate
- All search modes: keyword, semantic, hybrid
- Unlimited feed subscriptions
- REST API + MCP + Webhooks
- Priority support
Scale
For multiple agents or a product with real query volume.
$49/mo
- 50,000 queries/month
- 200 requests/min burst rate
- All search modes: keyword, semantic, hybrid
- Unlimited feed subscriptions
- REST API + MCP + Webhooks
- Dedicated support
Here's what your agent gets back
One request. A real entry, scored, sourced, ready to cite.
Request
curl -H "Authorization: Bearer pg_live_your_key" \ -H "X-PostGrad-Feed: all" \ "https://postgrad.io/api/v1/knowledge/search?q=pricing+for+enterprise+contracts"
Response
{
"data": [
{
"id": "a1b2c3d4",
"feed_name": "Agency Growth Playbook",
"title": "Enterprise contract pricing structure",
"category": "deal_evaluation",
"confidence": 0.92,
"score": 0.089,
"content": "Three consistent pricing levers in 12 enterprise deals..."
}
],
"meta": { "queries_used": 42, "queries_limit": 1000, "queries_remaining": 958 }
}Public endpoints /api/v1/stats and /api/v1/categories are free to try. No API key needed.
Frequently asked questions
For agent builders
Why not just use RAG over our own docs?+
RAG over your own docs surfaces what your team already knows. PostGrad surfaces what they don’t: operator judgment from people running the businesses your agents are advising. The two stack; they don’t compete.
What stops the feed from going stale?+
Licensors publish on a cadence; entries surface a last_updated timestamp and a confidence score that increases when the same knowledge is reinforced across multiple sources. Stale entries decay in confidence and rank lower in retrieval. You can filter by minimum confidence on every query.
What if a licensor closes their feed?+
You keep access to everything you’ve cached for the remainder of your billing period. New queries against a closed feed return a feed_closed status. You’ll never get silent failure.
How do I know the expertise is real?+
Anyone can sign up to publish a feed. We don't pre-vet credentials. What you get instead is transparency: every feed page names the operator behind it, links their work, and shows the reinforcement signal across entries. Quality compounds as agents query and re-query: high-signal feeds rise, weak feeds fade. Read the operator's bio, scan their sample entries, and try the feed yourself before subscribing.
What kind of data is available?+
Domain-specific feeds across business, sales, real estate, insurance, tech, legal, finance, and more. Browse the marketplace for current offerings.
How do I connect PostGrad to my agent?+
Paste our MCP URL into ChatGPT or Cursor as a custom connector, or add it to your Claude Code config with your API key. Any HTTP client works too. Two minutes.
For licensors
What data can I license?+
Any structured knowledge that would make AI agents more useful in your domain. Meeting insights, operational playbooks, industry expertise, proprietary frameworks.
Do I lose ownership of my data?+
No. You grant PostGrad a distribution license. You retain full ownership. You can unpublish any entry or close your feed at any time.
How much can I earn?+
Depends on your pricing, subscriber count, and tier. You keep 70% on Starter, 80% on Pro, 85% on Scale. A $10/mo feed with 100 subscribers nets $800/mo on Pro.