Meet Guido
Guido is the AI assistant inside ComplyCore, elearning.edu.pl and EasyACC. Managers ask in plain language — “Who in the warehouse has an expired forklift training?” or “Assign the workplace induction to everyone new on shift A” — and Guido answers questions about training and prepares assignments, groups and changes to staff records.
Guido does not replace anyone. It takes repetitive work off people’s plates so they can focus on decisions, and it makes everyday work faster and more comfortable.
Assign the workplace induction to everyone new on shift A.
I have prepared an assignment for 6 new employees on shift A (warehouse, receiving zone). Nothing is saved until you approve it.
- Workplace induction → 6 people, shift A
- Due within 7 days of each start date
- Group: New starters · shift A
Boundaries by design
An assistant is only as trustworthy as its limits. Guido is closed inside clear boundaries that the business, not the model, defines.
- It acts with the user’s own permissions — it can only do what that person is allowed to do
- Every change waits on a card for the user’s “Approve”; nothing is saved without it
- Every call is recorded in an audit log
- It is limited to the tasks and data of the platform it lives in
No personal data sent to the model
Before a request reaches the AI model, we strip personal data from it. The model works on the task, not on people’s identities — which keeps the assistant on the right side of GDPR by design, not by policy.
Multiple models, one assistant
Guido runs on Amazon Bedrock, which lets us use different AI models for different tasks and switch them as better models appear, without rebuilding the assistant. Models run in an EU region, and the cost of each answer is visible next to it.
It learns your routine
Guido learns which actions a user performs most often and offers to repeat them, so recurring tasks take one sentence instead of a dozen clicks. The user stays in control: repeated actions still go through approval.
Compliant from the start
We designed Guido for GDPR and the EU AI Act: human oversight of every change, traceable decisions, a record of every call and no personal data in model requests. We provide the documentation our clients need for their own assessments.
Your own assistant, inside your systems
Most businesses do not need another general-purpose chatbot. They need an assistant that knows their processes, respects their permissions and stays inside their systems. We build assistants like Guido for client platforms: we define which tasks it may perform, connect it to your data through the permissions you already have, and add approval and audit from day one.
- An assistant inside your existing web platform or internal tools
- Plain-language questions over your own data
- Prepared actions with human approval
- Model choice through Amazon Bedrock, running in the EU
Built by CYAP
- ComplyCore — Mandatory training, onboarding and renewal tracking for operational teams.
- elearning.edu.pl — Business learning through decision scenarios, AI conversations and structured learning paths.
Frequently asked questions
Does the AI model see our personal data?
No. We strip personal data from requests before they reach the model.
Can the assistant change data on its own?
No. Every change is prepared as a proposal and saved only after a person approves it.
Which AI models do you use?
Models available through Amazon Bedrock, chosen per task. Because the assistant is not tied to one model, it can switch as better ones appear.
Is it compliant with the EU AI Act?
We design for it: human oversight, traceability and record-keeping are built in, and we provide documentation for your assessment. For legal interpretation we recommend your legal adviser.
Can you add an assistant to our existing system?
Usually, yes, if the system has an API we can work with. We start by defining which tasks the assistant may perform and with whose permissions.