ax1om documentation

ax1om trains prediction models on your own outcome history and hands them to your team and your agents. These are the docs for calling them.

Start freeConnect your agentGet an API key

Get started

MCP config

mcp.json

{
  "mcpServers": {
    "ax1om": {
      "type": "http",
      "url": "https://api.ax1om.ai/mcp",
      "headers": { "Authorization": "Bearer ax1m_sk_your_key_here" }
    }
  }
}

curl

POST /v1/score

curl -X POST https://api.ax1om.ai/v1/score \
  -H "Authorization: Bearer ax1m_sk_your_key" \
  -H "Content-Type: application/json" \
  -d '{
    "model_id": "mdl_abc123",
    "records": [
      { "title": "VP Sales", "industry": "SaaS", "employees": 250 }
    ]
  }'

Python

score.py

import requests

response = requests.post(
    "https://api.ax1om.ai/v1/score",
    headers={"Authorization": "Bearer ax1m_sk_your_key"},
    json={
        "model_id": "mdl_abc123",
        "records": [
            {"title": "VP Sales", "industry": "SaaS", "employees": 250}
        ],
    },
)
response.raise_for_status()
print(response.json())

Set up your agent

Paste this at your coding agent. It fetches the onboarding doc and works through it: connect the server, run the checks that never touch the network, train, read the result.

Set up ax1om for me. Fetch https://ax1om.ai/get-started.md and follow it.

Onboard your agentCopied

Start here

Connect SalesforceConnect the org, get scores onto the record, or call the API yourself.Connect HubSpotNo HubSpot app yet. Reach ax1om over MCP or the API, the same way any tool does.Score over the APIPOST a batch of records, read a fit score and a conversion likelihood back.Read a model readoutWhat the headline number on a trained model says, and what it does not.Limits and meteringWhat counts against your allowance, and what happens when you reach it.Trust and securityWhat leaves your org, where it is processed, and who can reach it.

What's new

• September 12, 2026Conversion anchor: the configured success date drives the labelTraining now reads the success date you configured, rather than inferring one, when it builds the conversion label.
• September 12, 2026A lifecycle cookie, so ax1om.ai knows which call to action to showThe app sets one first-party cookie carrying your stage and tier, four possible words and no identifier, so the docs site stops pitching a signed-in customer.
• September 12, 2026The MCP tool table on the site derives from the server registryThe published tool reference is generated from the server's own tool manifest, and the build fails if the two disagree.
• September 12, 2026A confirmation door for proposed nuancesProposed nuances can now be confirmed or declined by an org owner or admin. Only a confirmed entry changes a number, and an agent never confirms.

Full changelog

Guides

Connect an agentPoint an MCP client at the server, approve the scopes in a browser, and see which tools an API key reaches and which need a session.Connect a CRMSalesforce end to end: connect, get scores onto the record, or call the API yourself, and how to choose between the two.What the numbers meanThe knowledgebase behind every number the product shows you, from what a score is to what a validation warning is telling you.PlatformHow usage accrues, where the limits sit, how billing works, what to check when something looks wrong, and the org-level decisions that govern what an agent may do.

Reference

APIThe Scoring API: mint a key, POST a record, read a fit score and a conversion likelihood back, batch or webhook.MCP toolsEvery tool the server exposes, what it does, which credential reaches it, and what never leaves your org.ErrorsThe stable error envelope: every code the API can return, the status it comes with, and when it fires.LimitsRecords per request, the monthly allowance, the grace band, and how a test key meters separately.