AI assurance platform

Test AI models on the fly.

incompli registers every AI system you operate, runs adversarial test batteries against it, and holds the release gate until the evidence says otherwise — with an audit trail regulators can read.

6
test harnesses
4
frameworks mapped
0
keys stored by us
claims-triage-assistant · run 4f2a

Composite score

Gate: fail

78.2

  • Prompt injection18p92 / floor 80
  • Jailbreak16p88 / floor 80
  • Data & PII leakage14p61 / floor 90
  • Toxicity & harm12p85 / floor 85
  • Bias & fairness14p64 / floor 70
  • Factual grounding10p79 / floor 70

system blocked · 2 findings raised · audit entry written

Evidence mapped to

EU AI ActNIST AI RMFISO/IEC 42001OWASP LLM Top 10

Most AI governance is a document that nobody tested.

Policies get written, committees get formed, and the model still goes live without anyone proving it resists a jailbreak or refuses to leak training data.

incompli closes the loop: the same platform that tests the system is the one that decides whether it may ship, and the one that keeps the receipt.

Product

Everything between "we built a model" and "we can defend it"

Adversarial testing

Prompt injection, jailbreak, data leakage, toxicity, bias and hallucination batteries run against your live endpoints — not a spreadsheet of intentions.

Governance gates

Every harness has a floor you control. Miss it and the system is blocked from release, with the decision written to the audit log.

Framework mapping

Results map to the EU AI Act, NIST AI RMF, ISO/IEC 42001 and the OWASP LLM Top 10, so evidence lands where auditors look for it.

Assurance artefacts

Model cards, FRIA records, risk tiering and sign-off gates assembled from real test evidence rather than written from memory.

Stop-the-line control

A kill-switch record for every system: who can pull it, how it is triggered, and what happened the last time it was used.

Board-ready reporting

Compliance evidence packs and board retrospectives you can print, or share on a link that expires after seven days.

How it works

Four steps from registered to released

  1. 01

    Register the system

    Describe the AI system, set its risk tier and purpose, and name the people accountable for it.

  2. 02

    Attach the model

    Point incompli at the endpoint under test. Credentials stay in your own secret store — incompli only holds a reference.

  3. 03

    Run the battery

    Pick a test suite and start a run. The harness scores each category and streams results back as they land.

  4. 04

    Let the gate decide

    Scores are judged against your floors. Pass and it ships. Fail and the system is blocked until it is remediated and signed off.

Product tour

Walk the whole journey before you sign up

Four steps, start to finish. Pick one to see what it actually looks like inside incompli.

Register the model you want tested

Point incompli at any hosted or private model, tag what it is used for, and set its risk tier under the EU AI Act. Takes about a minute.

  • OpenAI, Anthropic, Gemini, Bedrock, or your own endpoint
  • Risk tier and intended use recorded up front
  • API keys never stored by incompli

OpenAI

gpt-4o

High risk

Anthropic

claude-3-5-sonnet

Limited risk

Google Gemini

gemini-1.5-pro

High risk

AWS Bedrock

titan-text-premier

Minimal risk

Keys stay in your own environment · incompli stores a reference only

Find your package

Answer four questions, see what you need

No form, no sales call. The recommendation updates as you choose.

Which sounds like you?

How many AI models?

What do you need out of it?

Any security requirements?

Recommended for you

Individual

$100

per month · first 15 days free

Test your models properly and read exactly where they failed.

One model, testing and results. Start free for 15 days and move up whenever your needs grow.

  • 3 AI systems, 1 seat
  • All six attack batteries
  • Probe-level failure transcripts
  • Scores against the regulatory floor
Start 15 days free Compare every package

Frameworks

Evidence that lands where auditors look

Every test definition is mapped to the controls it satisfies, so a compliance pack is assembled from runs you already did rather than written the week before an audit.

EU AI Act
NIST AI RMF
ISO/IEC 42001
OWASP LLM Top 10

Your credentials stay yours

  • Model API keys are never stored by incompli — only a reference to where they live.
  • Every organisation's data is isolated at the database level, not just in the UI.
  • Four roles, from full control to read-only auditor, with actions hidden accordingly.
  • Shared reports expire on a timer and can be revoked the moment you change your mind.

Built for the whole accountability chain

One platform, four points of view

Chief AI Officer

Portfolio-wide scores, blocked systems and sign-offs awaiting a decision.

AI ethics board

FRIA reviews, risk register, and the gate approvals that carry their name.

Engineers and stewards

Register systems, run batteries, work findings down to zero.

Auditors

Read-only access to evidence, reports and the complete audit trail.

Pricing

Priced per organisation, not per test

Run as many batteries as your systems need. Start free while you evaluate and move up when incompli becomes the record of how you govern AI.

Solo builders evaluating a model

Individual

$100

per month

First 15 days free

For solo builders evaluating a model. Free for 15 days.

  • 15 days free, then $100 per month
  • 3 AI systems, 1 seat
  • AI Test Run — all six attack batteries
  • AI Test Findings with probe-level evidence
Start 15 days free

Auditors, consultants and assessors

Auditor & Consultant

$2,000

per month

For auditors, consultants and assessors working across client systems.

  • Audit runs, observations and score comparison
  • Compliance map and evidence packs
  • Comprehensive model cards
  • Expiring share links
  • Impact assessments
  • Multiple client systems and seats
Talk to us
Most chosen

Companies governing their own AI

Enterprise

$5,000

per month

For companies governing their own AI. The full governance system.

  • Governance gates and sign-offs
  • Custom pass floors and stop-the-line holds
  • Impact assessments and risk register
  • Comprehensive model cards and compliance packs
  • Single sign-on, self-hosted test harness and priority support
  • Unlimited systems, runs and seats
Talk to us
What's includedIndividualAuditor & ConsultantEnterprise
AI Test Run and AI Test Findings
AI Regulations—
Model cards—
AI Compliance dossiers—
Score comparison—
Expiring share links—
Impact assessments—
Governance gates and sign-offs——
Custom pass floors——
Stop-the-line holds and risk register——
Single sign-on——
Self-hosted test harness——
Priority support——

All plans include the full test battery, gate enforcement and the audit trail. Invoiced monthly or annually.

FAQ

Questions worth asking

Prove it before you ship it.

Register your first AI system, run a battery against it, and see exactly where the gate stands — in an afternoon.

Get started free