We’re building the data layer for apps and agents.
PrimaryAPI is one key and one credit balance for every kind of data your apps and AI agents need. You only pay when data comes back.
- Social posts
- X, YouTube, Reddit, Instagram, Facebook, TikTok
- Places
- Google Maps
- Email, phone and IP
- Email checks, IP lookups
- Music and podcasts
- Spotify
- Company and government
- Bank records
- Code
- GitHub
- A call that fails
- 0 credits
Why we built it
Getting data should not mean managing a dozen vendors.
The problem
A vendor for every source. A scraping API for one site, a validation vendor for email, a marketplace scraper for the long tail. Each has its own key, its own plan, its own response shape and its own bad days.
Agents make it sharper
Software can’t sign up for twelve plans. An agent can call any API, but it cannot wait for a sales call or notice that a vendor changed its output overnight.
So we do that work once
One key, one balance, one schema per API. Every API has a measured health record, and you get one line on the bill for each call that answered.
Principles
How we decide.
When a product choice is unclear, these settle it. Each one shows up somewhere you can check.
Evidence over claims
We publish what we measure, and nothing we have not.
- The network page is computed from our request log, and stays empty until there is traffic to measure.
- Every response says exactly what the call cost, in credits.
- Comparisons with other products are dated and sourced, or not made.
Pay for success
Credits are spent when you get an answer, not before.
- No monthly fee and no minimum on Pay as you go: a month without calls costs nothing.
- A failed call costs nothing, and neither does a call we refuse.
- A lookup that finds nothing is free, unless its API page says otherwise.
Agents as first-class users
Software is a customer too, and should not need a sales call.
- Finding APIs over MCP and REST needs no key.
- An agent can mint its own capped trial key and start work.
- A person approves real spending and can cap it. How agents work
Public data, used responsibly
Useful data about the world, with lines we do not cross.
- Public, logged-out and licensed data. We never ask for passwords or cookies.
- APIs that describe people in more detail need our approval first.
- No surveillance, stalking or eligibility decisions. Acceptable use
Where we are
Early, and saying so.
Every API ships the same way, so you can expect the same from each one.
Defined in code. Each API is a typed definition, reviewed like any other code.
Tested on recorded answers. A real hit and a real miss are replayed in our test suite before it ships.
Measured live. Once traffic flows, every call feeds the health figures on the network page.
No logos or badges yet. We do not have customer logos, certifications or a long uptime record to show. When we do, they will be measured and dated, like everything else on this site. Missing an API you need? Ask for it
Get in touch
Talk to the people building it.
Support and billing. Keys, calls, credits and plans. Include a request ID if it is about a call.
[email protected]Request an API. Tell us which platform or data you need. Requests decide what we build next.
Tell us what you needSecurity reports. Found a vulnerability? Report it privately and we will name you publicly if you would like.
How to report
Try it on a real call.
Sign up, get welcome credits, and call any API. Or hand the prompt to your agent and let it set itself up.