AI Strategy & Data Readiness
We prepare your data infrastructure for the AI era. From EU AI Act compliance to building first-party datasets for custom LLMs and predictive modeling.
- BigQuery
- Vertex AI
- dbt
- Claude API
- Snowflake Cortex
- OneTrust
What we build
Since August 2026 the EU AI Act has been fully in force. Anyone using AI systems, including in marketing and analytics, needs use-case documentation, risk assessment, and in some cases human-oversight processes.
But AI Readiness is more than compliance. It's the question: are your data and processes ready for the next step? Clean first-party data in a structured schema is the foundation for any AI use case that actually works, not just a GDPR checkbox.
The convergence of analytics and AI by 2027: Predictive audiences, AI-powered attribution modelling, LLM-based insight generation aren't future topics. They work today, but only at companies whose analytics foundation is in order.
Who it's for
Marketing teams using AI tools
Run AI-powered ad optimisation, GenAI content tools, or similar. When AI tools process customer data, the EU AI Act applies.
Compliance leads
Need to assess whether company AI usage is compliant, and don't have internal resources to evaluate it rigorously.
Analytics teams
Want to lay the foundation for AI applications: first-party data architecture, structured event data, clean segmentation.
CTO / CDO in EU
Know AI is coming and want to ensure infrastructure is ready, before pressure comes from above.
Compliance context
The EU AI Act, what applies now
As of August 2026, full application of all articles:
Prohibited AI practices (since February 2025): Social scoring, manipulative subliminal techniques, real-time biometrics in public spaces, not relevant for marketing analytics, but worth knowing.
High-risk AI systems: Certain HR AI, credit scoring, biometric categorisation. If you operate such systems you need mandatory documentation, conformity assessment, and registration in the EU database.
General Purpose AI (GPAI) / Foundation Models: Large language models (GPT, Claude, Gemini) used in internal tools or products fall under GPAI rules.
For most marketing AI applications: Transparency obligations, use-case documentation, and internally defined human-oversight processes. No certification required, but clear obligations.
The three pillars of the service
AI use case mapping & risk assessment
Which AI systems do you operate, knowingly and unknowingly? Many companies don't know how many AI components are already in their marketing tech stack.
What we do specifically:
- Inventory of all deployed AI systems including "hidden" ones like Smart Bidding, Advantage+, HubSpot AI
- Risk assessment per EU AI Act for each use case
- Gap analysis between compliance and current state
- Prioritised action plan
Tools: Custom Assessment Framework, EU AI Act checklists
First-party data architecture for AI
AI models are only as good as the data they run on. Fragmented or poorly structured data produces poor AI outputs, regardless of how good the model is.
What we do specifically:
- Audit of existing data architecture for AI readiness
- Event schema optimisation for ML inputs
- Consent architecture check. AI may only train on consented data
- Recommendation for data lake setup if not in place
Tools: BigQuery, Snowflake, dbt, OneTrust
AI governance & human oversight
Who decides what? Who reviews AI outputs before they inform decisions? These aren't technical questions, they're process and organisational questions.
What we do specifically:
- AI policy document use-case-specific to applicable EU jurisdiction
- Human-oversight process design per risk level
- AI literacy workshop for marketing and analytics teams (half-day)
- Documentation template for EU AI Act compliance
Tools: Custom policy templates, EU AI Act compliance framework
Deliverables
Joint engagement with clear ownership split, datascale delivers strategy and compliance documentation, Saloid the technical data infrastructure.
AI use case assessment
- Inventory of all deployed and planned AI systems
- Risk assessment per EU AI Act (which level, which obligations)
- Gap analysis: what's missing for compliance?
- Prioritised action plan
First-party data readiness
- Audit of existing data architecture for AI readiness
- Event schema optimisation for ML inputs
- Consent architecture check
- Recommendation for data lake setup if not in place
AI governance
- AI policy document use-case-specific to applicable EU jurisdiction
- Human-oversight process design
- AI literacy workshop for marketing and analytics teams
- Documentation template for EU AI Act compliance
Tools & stack
Compliance & risk assessment
- EU AI Act risk matrix (our template, EU-law reviewed)
- AI use case inventory template
- Gap-analysis framework: compliance vs. current state
- Audits for Smart Bidding, Meta Advantage+, HubSpot AI
First-party data architecture
- BigQuery (EU region europe-west3) or Snowflake (EU)
- dbt for schema modelling and ML-input pipelines
- OneTrust or Usercentrics for consent-compliant AI training data
- Event-schema optimisation for predictive modelling
AI governance & tooling
- AI policy templates, use-case specific under EU law
- Human-oversight process designs per risk level
- MLflow / Model Registry for auditability (with Saloid)
- EU-compliant LLM providers: Anthropic Claude API (EU DPA), Mistral
Engagement depths
Three depths. Clear scopes.
No retainer trap.
Audit Sprint
We audit what's wrong. Report + prioritised action plan.
plus statutory VAT · fixed price for a clearly bounded scope
Included in the fixed price
- 1 domain
- 1 analytics property
- 1 tag manager / tracking setup
- 1 CMP
- up to 5 core conversions
- 10 working days
- PDF report + 90-min walkthrough
What you get
- Full analysis of your existing setup
- Prioritised report with concrete action items
- Walkthrough call with the team (90 min)
- No follow-up contract, no retainer obligation
When it fits
When the setup works but the numbers are being argued about internally. Or you're unsure what from a UA→GA4 migration still holds.
For e-commerce, multiple domains or App + Web: Audit Sprint Plus, €3,900 net fixed price. Bonus: 50 % of the Audit Sprint credits toward a Build Sprint commissioned within 30 days.
Request an Audit SprintBuild Sprint
Fresh build or restructure of a tracking setup.
plus statutory VAT · final fixed price after scope definition
Typical scope
- 1 domain (multi-domain on request)
- 1 analytics property (GA4 or Piwik PRO)
- server-side container (Stape or own cloud)
- 1 CMP with Consent Mode V2
- up to 15 events / conversions
- 4–8 weeks delivery
- Blueprint, QA sign-off, handover docs
What you get
- Measurement Blueprint for your dev team
- GTM + server-side setup incl. CMP integration
- Full QA against the blueprint with sign-off
- Handover docs + 30-day post-launch support
When it fits
When analytics is structurally broken and fixing it in-flight costs more than a clean rebuild.
Managed Evolution
Ongoing partnership. Analytics as a product, not a one-off project.
plus statutory VAT · monthly cancellation after the minimum term
Included in the monthly price
- up to 3 domains under active care
- GA4 + server-side stack maintenance
- monthly roadmap + sprint planning
- QA on every release deploy
- Slack channel, < 4 h response (Mon–Fri)
- monthly report + executive summary
- 3-month minimum, then monthly
What you get
- Monthly development + feature rollouts
- Ongoing QA on every deploy
- Executive reports + dashboard evolution
- Slack support with guaranteed response times
When it fits
When analytics has to grow with you (new campaigns, new products, new data sources) and you don't want to build that team internally.
All prices net, plus statutory VAT. For companies in Germany, Austria and Switzerland.
Full-Cycle Delivery, who does what
Datascale owns
- Marketing AI use-case inventory (Smart Bidding, Meta Advantage+, HubSpot AI)
- First-party data architecture and consent compliance
- EU AI Act mapping for marketing use cases
- AI literacy workshop for marketing and analytics teams
Saloid delivers
- Technical AI governance framework (MLOps, Model Registry, audit trails)
- Risk classification per EU AI Act for technical systems
- Human-oversight mechanisms in technical systems
- Policy implementation in infrastructure and deployment processes
- EU AI Act Risk Matrix
- MLflow
- EU-compliant LLM providers
- Anthropic Claude API (EU DPA)
E-commerce enterprise. AI Act compliance assessment for Google Ads AI features and an internal recommendation engine. Gap analysis + action plan in 3 weeks.
Saloid: AI Agents & Automation-
Does the EU AI Act apply to my company if we only use Google Ads Smart Bidding?
Yes. Smart Bidding is an AI system that makes bidding decisions. Under current guidance it doesn't fall into the high-risk category, but transparency obligations and internal documentation are still recommended. We clarify in the assessment exactly which use cases trigger which requirements.
-
What does human oversight mean in a marketing context?
When an AI system makes a decision, for example which segment sees which offer, a human has to be able to review that decision and override it when needed. The process doesn't have to be heavy; often a documented review routine is enough.
-
What is first-party data and why does it matter for AI?
First-party data is data you collect directly from your own users, with consent, on your own infrastructure. Unlike third-party data (cookies from outside providers), it's available long-term, usable under GDPR, and qualitatively more reliable. AI models trained on first-party data are more stable and produce better results than those trained on aggregated third-party data.
-
When does the EU AI Act take full effect?
The core requirements have been in force since August 2026. The full roll-out with all delegated acts runs through August 2027. Companies that move now avoid last-minute compliance pressure.
Next step
EU AI Act in force since May 2026: where does your setup stand?
Strategy call about compliance, first-party data, and custom-LLM foundations. Full-cycle implementation together with Saloid.