Decision intelligence API other devs plug into their apps
The decision layer,one endpoint away.
Lumi puts the whole Kauzio platform behind one clean API. Plain JSON in, plain JSON out. Send any choice your users are about to make. Get back the answer, the case for and against, a score for how hard it is to undo, and a signed receipt your users can check. 50 calls free every month, no card needed. Paid plans get higher limits and priority support.
50 calls free · Priority support on paid plans · No card to start
Kauzio LumiSee it in motion
From an API key to a signed verdict in 30 seconds.
The real Lumi UI, the real flow. 6 steps a new developer goes through, exactly as they’ll see it.
Create your Lumi account
No card · 50 calls/mo free forever
Get an API key in 30 seconds. Build, prototype, ship. Upgrade when you outgrow the cap.
See it in action
One request in. A decision out.
Lumi is one REST endpoint. Here is the whole flow, end to end.
Send a decision as plain JSON
POST the question your app needs answered, with whatever context you already have. One endpoint, any language, any framework.
$ curl -X POST https://api.kauzio.com/v1/decide \
-H "Authorization: Bearer kz_live_..." \
-H "Content-Type: application/json" \
-d '{
"question": "Approve this £4,000 loan application?",
"context": {
"credit_score": 612,
"debt_to_income": 0.41,
"prior_defaults": 0
}
}'Get back a verdict you can act on
A clear verdict, a confidence score, the reasoning behind it, and a receipt_id. Drop it straight into your own logic or UI.
{ "verdict": "approve_with_review", "confidence": 0.78, "reasoning": "Score is below threshold but no prior defaults.", "receipt_id": "rcpt_9f3a21c8", "latency_ms": 341 }
Keys, usage and webhooks in one place
Manage API keys, watch live usage and latency, and wire up webhooks so your app reacts the moment a decision lands.
API keys
production
kz_live_4f8a…2c1staging
kz_test_9b3e…7d014,515
Calls this month
361 ms
Avg latency
3
Active keys
Webhook
ACTIVEPOST https://yourapp.com/hooks/kauzioExample request and response. Field values are illustrative.
What teams build on Lumi
One API. Any product that decides for its users.
Lumi is not tied to a sector. If your product makes a choice on a user's behalf, the decision layer drops straight in.
Not on this list? If your product makes a call a user has to live with, Lumi fits.
/v1/decide{ verdict: "…" }Kauzio Lumi
Decision intelligence.
Drop signed verdicts and risk scores into your own product. Six checks behind one endpoint.
/v1/decide{ verdict: "…" }Simple by design
One API call. Full decision analysis.
Send any decision as plain text. Get back a clear breakdown, flagged thinking traps, a suggested action and a score for how sure we are. All in one response.
Get your API keycurl -X POST \
https://api.kauzio.com/api/public/v1/forge/decide \
-H "Authorization: Bearer kz_live_..." \
-H "Content-Type: application/json" \
-d '{
"decision_text": "Should we expand to the US market in Q3?",
"context": "Two of our top three markets are still under-penetrated.",
"domain": "strategy"
}'{
"question": "Should we expand to the US market in Q3?",
"answer": "Not yet. Two of your top three markets are still under-penetrated. Recommend a Q4 pilot with a 90-day proof window.",
"data_context_loaded": true,
"source": "kauzio_brain",
"latency_ms": 312
}API reference
Key endpoints
A selection of the most powerful endpoints in the Kauzio API.
/api/public/v1/forge/decideSend a question. Get back the answer, a risk score, how hard it is to undo, the case for and against, and a signed receipt. All in one call.
/api/public/v1/forge/opposeGet the strongest case against a decision across six checks: sales at risk, profit hit, how customers react, extra work, cost to undo, and how sure we are.
/api/public/v1/forge/calibrationSee how close past answers got to what actually happened, by sector.
/api/public/v1/forge/certificate/issueGet a signed receipt for an existing decision_id. Returns a public verify link.
/public/v1/verify/{certificate_id}Check any signed receipt by its ID. No API key needed.
Platform features
Built for production
Everything you need to ship the decision layer into a real app, from keys to webhooks to signed proof.
API Keys & Rate Limiting
Generate multiple API keys per project, set rate limits, and monitor usage per key, all from the Lumi developer dashboard.
Webhooks
Get notified when decisions, simulations and receipts complete. Every message is signed, so you can confirm it came from Kauzio.
See your usage
Track calls per day and per month against your keys. See your plan cap in the dashboard and a full log for each key.
Signed receipts
Every /decide call comes back with a signed receipt and a public verify link. Anyone can check the call later. No API key needed.
Plain JSON, no SDK
Every endpoint is a simple JSON POST. Hosted on European infrastructure. No client library to install, curl works fine.
Follow-up tracking
Send the result back once a decision plays out. Kauzio learns. Every result makes the next call sharper, for you and every other builder.
Works with any stack
Plain JSON in, plain JSON out. REST API works with any language or framework today. Python, Node.js, and Go SDKs are on the roadmap.
Sign any AI output
Time-stamp and sign the output of any AI model. A signed record of what your AI said and when.
Real answers, not made up
Every claim ties to your real data. When the data is not there, Kauzio says so instead of making something up.
In the terminal
One API key. The whole of Kauzio behind it.
One key. Three signed endpoints. Eight webhook events. Drop a verdict, a risk score and a tamper-proof certificate into anything you build, and close the loop when the outcome comes in.
One API. Every product you build.
Your code. One verdict away.
Lumi sits behind one REST endpoint. Drop a verdict, a risk score and a signed certificate into anything you ship: web, mobile, server, SaaS, AI agent, e-commerce, you name it.
Web app
browser → API
Mobile · iOS · Android
on-device call
Backend service
server-side decision
CLI · ops scripts
curl or SDK
AI agent
tool-call from LLM
E-commerce store
checkout, returns
SaaS product
embedded inside
Slack · Discord bot
chat-driven calls
Analytics pipeline
near-real-time scoring
API · /v1/decide
Web app
browser → API
Mobile · iOS · Android
on-device call
Backend service
server-side decision
CLI · ops scripts
curl or SDK
AI agent
tool-call from LLM
E-commerce store
checkout, returns
SaaS product
embedded inside
Slack · Discord bot
chat-driven calls
Analytics pipeline
near-real-time scoring
One API key. Pay per call, free to start. Every response ships a tamper-proof signed receipt you can verify with /v1/verify.
inside / lumi
One POST in. One signed verdict out.
Calibration on the way back.
Lumi runs the full decision loop server-side. Your app posts the question, gets a signed verdict, then patches the outcome back. Every outcome you report sharpens the coefficients on the next call.
// your app asks Lumi
POST /api/public/v1/
forge/decide
{
"decision": "Raise prices
8% in Q3?",
"context": { ... }
}// signed, verifiable, ~280ms
200 OK
{
"verdict": "go",
"risk": 0.18,
"reversibility": 0.62,
"cert_id":
"KZ-2026-001234",
"verify_url":
"/v1/verify/..."
}PATCH /decisions/
KZ-2026-001234
{ "outcome":
"margin +2.4%" }decision.calibratedcoefficients +0.034
next /decide is sharper, for you and for the platform.
memory is a first-class field, not an afterthought.
Inside Kauzio Lumi
Production decision intelligence at API-call scale.
Four endpoints, seven webhook events, signed certificates. Drop the entire decision tool into anything you ship.
API Keys
kz_live_* keys, scopes, rate limits
7 Webhook Events
HMAC-signed, retried, logged
Signed decision certificates with public verify
Tamper-evident, hashed at signing, public verify endpoint, no auth
/brain/query
Grounded AI Q&A over user data
/consequence/simulate
Full cascade in one API call
/opposition/challenge
Six-axis opposition across revenue risk, margin impact, demand elasticity, operational friction, reversal cost and uncertainty level.
Usage Meter
Daily + monthly limits, updated on each call
Delivery Logs
Every webhook attempt, status, retries
Priority email support
Faster response on paid plans
Lumi Forge API
Build on the same sector causal graph and decision primitives that power Kauzio.
Per-sector calibration
We publish how accurate Kauzio is in your sector. Updated nightly.
Cross-business patterns
Anonymised peer outcomes from other Kauzio businesses in your sector and size band.
Forge API
The four calls that run the engine.
Forge is how you call Kauzio from your code. Decide, push back, check the track record, sign it. Plain JSON in, plain JSON out.
forge.decide
POST /api/public/v1/forge/decideSend a question. Get back the answer, a risk score, how hard it is to undo, the case for and against, and a signed receipt with a public verify link. All in one call.
POST /api/public/v1/forge/decide
{
"question": "Should we raise prices by 8%?",
"sector": "hospitality",
"context": "Margin 0.18, footfall +12% last 30d."
}forge.oppose
POST /api/public/v1/forge/opposeThe strongest case against any decision, across six checks, with evidence for each.
POST /api/public/v1/forge/oppose
{
"decision": "Raise prices by 8% across all venues.",
"context": "Margin trending +4% YoY."
}forge.calibration
GET /api/public/v1/forge/calibrationSee how close past answers got to what actually happened, by sector.
GET /api/public/v1/forge/calibrationforge.certificate / issue
POST /api/public/v1/forge/certificate/issueGet a signed receipt for an existing decision_id. Returns a public verify link.
POST /api/public/v1/forge/certificate/issue
{
"decision_id": "dec_8f3..."
}Every Forge call is signed and has per-key rate limits. Results you send back help Kauzio learn. Patterns stay anonymous across businesses.
Pricing
Pay per call. Free to start.
Free · Indie · Studio · Enterprise.
50 calls free
See API pricingFree tier available · Cancel any time
