Try it without a key

Try the Algenta engine with a single POST /v1/playground call — no sign-up and no API key. Run a decision, optimization, simulation, or forecast in one request.

The fastest way to feel what Algenta does is the playground: one HTTP call, no account, no key. POST /v1/playground runs a decision analysis, optimization, simulation, or forecast against the engine and returns a recommended action, a value, a confidence, and a plain-language "why". It is rate limited to 10 requests per minute per IP — enough to explore, not to build on. When you are ready for real work, mint a key (Authentication & API keys) and move to /v1/simulate and the rest of the API.

Before you start

Nothing to install and nothing to authenticate. You only need a base URL:

• Hosted: https://api.algenta.ai

• Self-hosted: http://localhost:8000

Make your first playground call

1

Ask a decision question

Send a decision request — "given these inputs, what should I do?". No statistics knowledge required.

cURL

curl -sS -X POST https://api.algenta.ai/v1/playground \
  -H "Content-Type: application/json" \
  -d '{
    "mode": "decision",
    "inputs": { "revenue": 120000, "cost": 70000 },
    "runs": 1000
  }'

Python

import httpx

resp = httpx.post(
    "https://api.algenta.ai/v1/playground",
    json={"mode": "decision", "inputs": {"revenue": 120000, "cost": 70000}, "runs": 1000},
)
print(resp.json())

2

Read the response

The engine returns a compact result you can act on directly:

{
  "playground_id": "b8f1c0de-1234-4abc-9def-0123456789ab",
  "mode": "decision",
  "action": "proceed",
  "value": 50000.0,
  "confidence": 0.87,
  "why": [
    "Expected net value is positive across the modeled range.",
    "Downside stays bounded at the 5th percentile."
  ],
  "details": { "scenarios_evaluated": 1000 },
  "latency_ms": 4.2,
  "upgrade_hint": "Playground cap: 1,000 scenarios. Upgrade for 1,000,000+."
}

Expected result — a 200 response whose action is one of proceed / pause / reject (for decision), with a value, a confidence between 0 and 1, a why list, and a sub-5ms latency_ms. No key was required.

The four modes

mode selects what the engine runs. All modes take inputs as name → value (or name → {value, low, high} for a range).

mode	Question it answers	Key inputs
decision	What should I do?	inputs (values or ranges)
optimize	What is the best configuration?	inputs as [min, max] or {min, max} ranges, plus objective
simulate	What is the range of outcomes?	inputs
forecast	What happens next?	history (a numeric series, most recent last, at least 3 values)

optimize

curl -sS -X POST https://api.algenta.ai/v1/playground \
  -H "Content-Type: application/json" \
  -d '{
    "mode": "optimize",
    "objective": "maximize profit",
    "inputs": { "price": [10, 200], "quantity": [100, 5000] },
    "runs": 1000
  }'

simulate

curl -sS -X POST https://api.algenta.ai/v1/playground \
  -H "Content-Type: application/json" \
  -d '{
    "mode": "simulate",
    "inputs": { "revenue": 120000, "cost": 70000 },
    "runs": 5000
  }'

forecast

curl -sS -X POST https://api.algenta.ai/v1/playground \
  -H "Content-Type: application/json" \
  -d '{
    "mode": "forecast",
    "history": [80000, 85000, 90000, 95000, 100000]
  }'

Request fields

Field	Type	Default	Notes
mode	string	decision	One of decision, optimize, simulate, forecast.
inputs	object	{}	Your variables as name → value or name → {value, low, high}.
objective	string	maximize_value	Used by optimize, e.g. maximize profit, minimize cost.
runs	integer	1000	Scenarios to evaluate. Range 100–5000 (playground cap).
history	array of numbers	—	Required for forecast; at least 3 values, most recent last.

The playground caps runs at 5,000 and allows 10 requests per minute per IP. Over the limit returns 429 with error.code = "playground_rate_limit". For 1,000,000+ scenarios per request and no rate cap, get an API key and call /v1/simulate.

Next steps

Run the engine for real

Install, mint a key, and call /v1/simulate.

Get an API key

Authenticate and lift the playground caps.

See the full API

Every endpoint, schema, and error shape.