Your agent gets every newly-funded startup — with hiring signals, contacts, and niches — by pointing `OpenAPIToolkit` at our spec. Ten lines. No scraping. Free at 100 calls/day.
Drop into `OpenAPIToolkit.from_url()` — every endpoint becomes a typed Langchain tool automatically. No wrappers.
Let your agent pass natural-language queries: 'AI infra startups hiring in Europe'. We return structured filters + matching rows in one call.
Every field typed in the spec. Optional fields clearly marked. No HTML, no surprises. Function calling just works.
Your retry logic reads the reset timestamp and sleeps exactly the right amount. No exponential-backoff hacks.
Don't poll. Subscribe once with your filter, deliver to your agent's inbound endpoint. HMAC-signed. Retried on failure.
Copy-paste the full agent. 10 lines. Runs today. Official MIT-licensed TypeScript client published as `@getfundedapi/client` on npm.
POST /v1/search/ai
{ "query": "AI infra startups hiring in Europe that raised Seed this month" }# pip install langchain-openai langchain-community requests
from langchain_community.agent_toolkits.openapi import planner
from langchain_community.tools import OpenAPISpec
from langchain_openai import ChatOpenAI
import requests, os
spec = OpenAPISpec.from_url("https://fundedapi.com/api/openapi")
agent = planner.create_openapi_agent(
spec,
requests.Session(),
ChatOpenAI(model="gpt-4o-mini"),
allow_dangerous_requests=False, # FundedAPI is read-only
headers={"Authorization": f"Bearer {os.environ['FUNDED_API_KEY']}"},
)
print(agent.invoke("AI infra startups hiring in Europe that raised Seed"))Yes — our spec is OpenAPI 3.1 compliant. `OpenAPISpec.from_url('https://fundedapi.com/api/openapi')` auto-generates tools for every endpoint including filters, sorting, and pagination.
Every Langchain tool is a Langgraph tool. Use the same `create_openapi_agent` output as a `tools=[...]` entry in your Langgraph node, or wrap individual endpoints as `@tool` decorators.
Wrap `POST /v1/search/ai` as a single `@tool` that takes a string query. Your agent decides when to use it vs. the structured endpoints. See /recipes/langchain for a working example.
Our responses are tight — 10 startups in ~4KB of JSON. Typical agent turn: ~500-1500 tokens per /v1/startups call. Much cheaper than having the LLM reason about scraped HTML.
Free tier: 100 calls/day. Pro: 10k/day. Scale: 100k/day. For long-running crews, subscribe to webhooks instead of polling — 0 calls from your side, delivery pushed to you.
Yes — we use standard `Authorization: Bearer fapi_...`. Pass it via `headers={}` in the toolkit constructor. The OpenAPI spec's `bearerAuth` scheme is auto-recognized.
The spec is stable and served with `Cache-Control: public, max-age=3600`. Safe to cache in your agent build step — rebuild weekly.
Fresh funding rounds + key hiring signals in your inbox every Monday.
Free forever. No card. No tricks.