Feature · Watch · Written by Yobel Tzegai · Last checked 3 November 2026

How often does Complipath check our code and logs, and what does it flag?

In Complipath (complipath.io), Live monitoring is how Growth reads your connections: every connection is checked every hour, and GitLab is read again after each push. On Starter a check runs when you press Check now, at most once a day per connection. A finding from code alone is marked Possible.

What it is for: when an enterprise customer's review asks about your AI, the answers are ready, each with its source, so the review need not hold up the deal.

Get my risk classes, freeBook a 20-min call20 minutes with the founder.

Always on

No integrations needed to start.
Connect GitHub, GitLab, Azure or Datadog when you're ready, and Complipath finds new AI systems, changes to your AI vendors' usage policies and changes in your code on its own.
Every document stays ready to download, and the next questionnaire starts from what is already answered.
How Complipath keeps itself current.

How often it reads

On Growth, every connection is checked every hour.
GitLab is read again after each push through its push webhook; so is code the in-network scanner reads in your own CI.
A GitHub read after each push needs the GitHub App, which is switched off in production today.
Hourly and after-push reading can each be switched off per tool.
On Starter, a check runs when you press Check now, at most once a day per connection; code and the Claude or OpenAI connection are also read when you connect.

What it flags

From code alone, possible problems: a chatbot with no AI notice, personal data sent to a model, a decision about people with no review. Each is marked Possible.
Every open problem becomes a Fix step on your plan.
A system no check reads is shown as Not checked, never as OK.
A problem the next check could see and did not find is marked fixed.
A confirmed audit answer it contradicts is marked Needs review and never rewritten, and the workspace’s admins get a letter from that check.

What it does not do

It does not monitor how a model performs or drifts in production: the only model figures it shows are each AI service’s calls and errors and the models used in the last 30 days.
Connecting Datadog or Azure Monitor tests the keys and saves the connection; the first check runs on Check now or at the next hourly read.

Which plan

Included in Growth. Starter includes the connections and Check now, at most once a day per connection; reading every hour and after each push is Growth.

Where it sits in the law

It is not itself an EU AI Act duty. A Possible finding is a lead for a person to check, not a decision that a duty is broken.

Growth

€1,499 a month

or €14,990 a year. Prices exclude VAT.

Up to 50 AI systems
Everything in Starter
The Free plan is for one AI system and one user, with no end date

Get my risk classes, free

All plans

Questions

How often does it check?

On Growth, every connected tool is checked every hour, and GitLab is read again after each push; each can be switched off per tool. On Starter, a check runs when you press Check now, at most once a day per connection, and code is also read when you connect it.

Does it say we break the law?

No. A finding from code alone is marked Possible, a lead for a person to check. A system no check reads is shown as Not checked, never as OK. A problem the next check could see and did not find is marked fixed.

Which plan includes it?

Growth, at €1,499 a month or €14,990 a year for up to 50 AI systems. Prices exclude VAT. Starter, at €499 a month or €4,990 a year, includes the connections and Check now, at most once a day per connection. Reading every hour and after each push is Growth.

Does it watch how our model performs?

No. It does not monitor how a model performs or drifts in production. The only model figures it shows are each AI service’s calls and errors and the models used in the last 30 days, so performance testing stays with your own tools.