HomeEngagementsEU AI Act and GDPR engineering-readiness review
Agentic applications · 01 · The problem and the decision

EU AI Act and GDPR engineering-readiness reviewWhich decisions does your model take without a record?

Your AI system is approaching production, but its decisions, data flows and responsibilities are hard to explain. Give your DPO, counsel and engineering team an evidence pack they can review together.

For CTOs, Heads of AI, DPOs and ISVs who need technical facts about one AI system.

2–3 weeks elapsedOne agreed AI system; selected data flows and controlsA named architect accountable
02 · What we check

Agree the scope. Inspect the evidence.

We agree the data flows, controls and review depth, then prepare the engineering evidence for those questions. Your DPO and counsel determine the legal position, applicable duties and timing.

System and data

Read the agreed architecture, code or configuration, prompt examples and provider documentation for the selected data flows. Establish where inference runs and what is retained.

Decisions and disclosure

Inspect where the system informs or acts, what the user sees, who can intervene and which records exist. Identify evidence gaps for legal review.

Runtime and control

Review tool permissions, confirmations, evaluation, tracing and failure handling. Connect gaps to concrete design decisions and owners.

What we need from you

  • Read access to the agreed architecture, configuration, traces and logs; use redacted or representative data where suitable.
  • An owner for the AI system and availability of your DPO or counsel for the read-out.
  • The integrations, data flows and provider information agreed in the proposal.

Scope and boundaries

  • Legal advice, certification and a guarantee of regulatory compliance are outside this engineering review.
  • Additional AI systems, integrations or data flows, remediation and detailed implementation-ticket engineering require separately agreed scope.
03 · What you get

A decision you can act on.

  • An engineering map connecting the selected decisions and controls to evidence for your DPO and counsel.
  • A record of the agreed data flows and runtime controls, with assumptions and missing evidence identified.
  • A prioritised gap list with suggested owners and next actions.
  • Questions and dependencies for legal review, including responsibility, applicability and timing.
04 · Relevant work and a sample

See the thinking behind the review.

Atsky demonstration

Engineering choices you can inspect

Atsky Voice is an Atsky demonstration developed in partnership with Telnyx. It provides an example of session boundaries, tool use and disclosure in an agent application. It is not evidence that a customer bought this review or achieved a compliance outcome.

Client architecture and engineering

A bounded flow for network operations

For Nokia, working with a Tier-1 Nordic operator, Atsky designed the situation-enrichment agent as a four-step flow: retrieve, annotate, recommend and format. The NOC retains the decision; tracing and in-tenancy inference form part of the design.

05 · Duration and price

A defined engagement. An agreed scope.

From €7,500Starting price, excluding VAT

From €7,500, excluding VAT. We quote a fixed fee after agreeing the AI system, integration and data-flow inventory, access and review depth. The elapsed delivery window depends on the agreed evidence and stakeholder availability. Remediation, detailed implementation-ticket engineering and any extension are scoped separately.

We agree the included scope, deliverables and fixed fee in the proposal before work starts.

Duration
2–3 weeks elapsed
Base scope
One agreed AI system; selected data flows and controls
Your data
Read-only access, in a workspace that holds only your engagement. If your policy requires it, we work with EU-only processing and zero data retention. Any extra cost is named in the proposal.

Our own AI-assisted tooling reads configuration and logs and drafts findings; a named architect checks and signs them. How we work and handle data →

06 · Request a working session

Start with the question you need to resolve.

In 45 minutes, discuss the decisions your system makes, the evidence your stakeholders need and an example of disclosure and records in our own runtime. Agree whether this engineering review fits.

This requests a 45-minute conversation. We confirm a suitable time with you.

Does the question start with the cloud platform beneath the AI system? Explore the landing-zone assessment.