Products
Intelligence is grown. Put it to work.
Six products on one Kongen key: three you call directly, three that grow in the Garden. The first 1,000 KT are free, with no credit card.
$ pip install kongenlabs
Click to copy
# the request curl https://api.kongenlabs.life/v1/logic/score \ -H "X-API-Key: $KONGEN_API_KEY" \ -H "Content-Type: application/json" \ -d '{"text": "Given these three failing tests, find the shared root cause and propose a fix."}'
# the response { "regime": "deep", "confidence": 0.78, "confidence_adj": 0.0, "recommended_tokens": 1500, "tokens_used": 1, "request_id": "req_..." }
Real output from the shipped scorer. Same input, same answer, for a given scorer version.
01What you can use, and what is coming
Six products. One Kongen key.
Three you call directly. Three grow in the Garden.
Call Kongen directly
Logic SDK Available now
One call returns the regime, the confidence and the token budget for a step: the same answer for a given scorer version, with no model call to wait for. Python and TypeScript SDKs on the same key.
- Size the token budget for a step before you spend it.
- Send a step that does not need deep reasoning to a small, fast model.
- Gate a low-confidence step to a reviewer or a person.
- Audit model spend by the kind of step it went on.
Pattern Transfer Available now
Score a structural signature against the organisms’ cross-domain reference patterns, and get back what it matches and how closely. The reference patterns are research outputs, some from work we have since withdrawn. Unverified: to be re-derived on our rebuilt research platform.
- Check a signal against reference patterns from other domains.
- Classify a structural signature against the universal archetypes.
- Adjust a confidence you already carry with a cross-domain match score.
MCP Server Available now
The same key spoken as Model Context Protocol, so an agent calls Kongen natively instead of through glue you maintain. Claude Code, Cursor and any MCP-compatible client can discover the tools and call them.
- Let an agent size its own token budget, step by step.
- Put pattern scoring inside any MCP-compatible system.
- Chain scoring with the other MCP tools an agent already uses.
Sending each step to the tier its regime earns takes an estimated 21 to 43% off the cost of running every step on a frontier model, at the estimator’s defaults; shorter requests save less. See how it is calculated.
In the Garden
Flow Available now
Chat without choosing the model. Flow tells you which one answered and what it cost, and its client is Apache-2.0, so you can read exactly what it sends and what it shows. Your chats stay on your device with no server copy, and you bring your own provider keys.
Loop Not shipped yet
A harness between your agents and the models they call: route each step, check what comes back, gate what gets through. Nothing to install yet, and no release date.
The data-centre app Research stage
A research-stage app on machine telemetry. Its first figures came from synthetic data and are withdrawn; nothing about it is claimed yet.
Open source on GitHub, MIT: Python SDK, TypeScript SDK, MCP server, examples. Flow is Apache‑2.0.
02The decision, in full
Everything the API returns.
Four fields, typed, so your code branches on them instead of parsing prose.
regimeenum- One of five, and typed: trivial, fast, moderate, deep, exhaustive. An enum, not a paragraph you have to parse, so a switch statement reads it.
confidence0 to 1- How decisively the step maps to that regime. This is the field to gate on: act on the answer when it is high, and treat a low one as the signal that the step is ambiguous.
recommended_tokensinteger- The budget this step is worth. It goes straight to your provider’s max-tokens, or to whatever ceiling your own harness enforces.
confidence_adjsigned- A signed adjustment computed from the prompt’s own structure, for callers that already carry a confidence of their own and want to nudge it rather than replace it.
What it does not do is answer arbitrary questions. It answers two: how much reasoning an input is worth, and which universal pattern a signal matches. That is narrow, and the narrowness is the point: an agent asks both at every step, thousands of times, where today they are either hardcoded or handed to another model call, which adds a wait and a cost and can answer differently on each run.
Spend the budget
recommended_tokens goes to your provider, or to whatever ceiling your own harness enforces.
Pick the tier yourself
The SDK ships a plain regime-to-model map and you are meant to replace it: which model you trust with a deep step is your decision, and it stays yours.
Gate on the confidence
A low one says the step did not commit to a regime. That is the step to widen, to retry with more context, or to put in front of a person.
Your agent does not need a client library to reach any of this. The MCP server is live on the same key: POST /v1/mcp/tools/list returns logic_score and transfer_score, and POST /v1/mcp/tools/call runs them, so an agent can discover the decision and use it at runtime. Each MCP call is metered the same as the operation it wraps.
03Pricing
1,000 KT free. No credit card.
Pay-as-you-go after that at $0.0007 per KT. A scoring call is 1 KT.