Login AI
AI governance and compliance
We describe which AI systems you use, which risks they create, and which documents are required. We deliver a register, risk assessment, policies, and model cards for an internal audit.
We build an AI system register, risk classification, and a policy pack for the EU AI Act and ISO 42001.
Pricing
from ₽500,000
per project
This is the lower bound of a project. We scope the exact quote after a brief: it depends on volume and the number of integrations.
Capabilities
AI system register
We collect every AI system and scenario in one register: owner, purpose, model, data, users, and status. The register becomes the base for classification.
Risk classification
We assign a risk category under the EU AI Act and an internal matrix. Each system gets prohibited practices, obligations, and evidence.
Policy pack
We write an AI use policy, roles, and an approval process. The policy describes how to launch a system, handle incidents, and update a model.
Model cards
We prepare model cards and an evidence pack: data, limits, metrics, tests, and owners. The documents are ready for an internal audit and legal review.
From request to result
1
Scope
We define jurisdictions, systems, and the documentation level. We fix which requirements apply.
2
Inventory
We collect the AI system register and owners. We check which models and data are in use.
3
Risk assessment
We classify systems and score risk by likelihood and harm. High-risk systems get evidence of controls.
4
Policies and rollout
We agree policies with legal and security. We introduce a review process for new systems.
Check if it fits you
Good fit
AI in several units
Models and scenarios run in sales, support, and operations. The register shows the whole picture and the owners.
Not a fit
A policy template
A ready template describes general rules. The register, risk assessment, and evidence need a separate project.
No system register
We build the register in the first stage: without it, classification and evidence stay incomplete.
What you get
4
4 artifacts
The project delivers four artifacts: a system register, risk assessment, policy pack, and model cards. The register and model cards carry a system ID. Policies describe the shared perimeter.
30 days
30 days
The base program for one jurisdiction takes about 30 days. The timeline covers interviews, classification, and document review.
3
3 jurisdictions
The program covers up to three jurisdictions with different requirements. Each one gets applicable rules and evidence.
What the delivery includes
AI system register
The register lists systems, owners, models, data, and status. It shows which obligations apply to each system.
Risk assessment
The risk assessment shows category, likelihood, harm, and controls for each system. High risk gets a separate evidence plan.
Policy pack
The policy pack describes roles, launch, incidents, and model updates. The documents are ready for legal and security review.
Model cards
Model cards record data, limits, metrics, and owners for each model. The cards join the evidence pack for an audit.
How the system is built
Register
We collect AI systems, owners, and models in one register. Each record has a purpose, data, users, and status, so the register stays a checkable base for assessment.
Classification
Classification maps a system to a risk category and the EU AI Act requirements. For high risk we record obligations, controls, and the evidence needed before launch.
Policies
Policies and model cards link requirements to roles and processes. An internal audit checks the evidence, and a model update triggers a fresh risk assessment.
Where the solution pushes back
- 01
Legal review
Your legal team reviews the policy pack. We provide the structure, wording, and evidence, and the final decision stays with the lawyers.
- 02
Register and reality
The register rests on interviews and technical documentation. With incomplete data a system gets a status and an evidence collection plan.
- 03
Documents and work
Governance documents describe requirements and controls. We implement guardrails, DLP, and monitoring as a separate project.
Technologies
The technologies we use in our work and their role in a project.
Evidence and control
We collect the model inventory through OpenAI API, trace requests in Langfuse, manage roles in Keycloak, and store the evidence pack in PostgreSQL.
Cost calculator
This is an indicative range, not a quote. We scope the exact estimate after a brief: it depends on volume and the number of integrations.
AI systems
Jurisdictions
Document pack
Add-ons
Support
Year one: updates and monitoring, a reply within 2 business days.
Service
AI governance and compliance
The estimate covers our team's work. Hardware, third-party service licences and travel are added on request.
Estimated cost
₽960,000
Timeline
5 wk
Frequently asked questions
How much does a governance project cost?
The base program starts at RUB 500,000 and covers the register, risk assessment, policies, and model cards. The number of systems and jurisdictions affects the document volume.
How long does it take?
Four to six weeks for six systems and one jurisdiction. Several jurisdictions add 1–2 weeks for requirement checks.
Does the EU AI Act apply to us?
The act covers systems placed on the EU market or affecting people in the EU. During scoping we check roles, users, and the risk category.
Is ISO 42001 included?
Yes, the pack can follow ISO/IEC 42001: policy, roles, risk assessment, objectives, and an internal audit.
What teams add to this service
Ready to discuss your task?
Leave a request – we'll get back with a proposal and a preliminary estimate.
Contact us