Adoption and compliance experiment
An SME assessment and reporting prototype
An SME assessment and reporting prototype combining conversational intake, deterministic scoring, country data, retrieval, and a Mistral-assisted report pipeline. It is guidance software under development, not a legal determination.
Business ↔ technology bridge · Adoption and compliance experiment
Explores how an SME can move from AI ambition to a reviewable readiness and action record.
This is a portfolio and source interpretation—not a customer, deployment, certification, or outcome claim.
SMEs must connect business priorities, readiness, regulatory exposure, funding, and feasible use cases without a large specialist team.
Keep scoring and regulatory rules distinct from narrative generation, and ground country and funding guidance in dated records.
Conversational intake, persisted assessments, staged profiling and compliance workflows, retrieval, country data, and PDF report paths.
Readiness gaps, regulatory uncertainty, funding constraints, and use-case evidence return to roadmap, investment, and accountable review.
Architecture
Three source layers: a conversational intake for SMEs, a 4-agent compliance + funding pipeline, and a PDF playbook delivery path. No report turnaround is asserted.
Source map only: a layer can represent implemented code, an adapter boundary, or documented design. It does not assert a configured production service.
The intake is designed for 10–15 exchanges over Mistral chat. Profile: company size, industry, country, current AI usage, pain points, ambition. French and English are represented in source.
Profiler → Compliance → Use Cases → Report. Rule-based scoring, country-program modules, retrieval, and regional-matching surfaces are separated so their sources and assumptions can be reviewed.
Planned delivery path represented in source: WeasyPrint PDF report with compliance posture, funding shortlist, EDIH match, and a 3-phase implementation roadmap. Email flow with verification + report-ready notification. Token-gated admin panel for assessment statistics and approvals.
The Problem
Why European SMEs struggle to adopt AI responsibly
December 2027 deadline looms but most SMEs don't know which risk tier they fall under, what obligations apply, or what concrete steps to take before penalties kick in.
EU and national funding programs exist across Horizon Europe, Digital Europe, and country-specific grants — but navigating the landscape without specialist support overwhelms most organisations.
AI deployments create new data processing activities that organisations fail to assess before deployment, leaving them exposed to Article 83 fines and supervisory authority scrutiny.
Most companies don't know where they stand: digital maturity gaps, infrastructure deficiencies, and workforce readiness issues remain invisible until they cause expensive project failures.
Core Features
From conversational intake to a draft assessment report that requires expert review
Source scope only: an entry can describe implemented code, a documented adapter, or a configured workflow. It is not a statement that a production service is active.
01
A guided Mistral-powered chatbot is designed to collect a company profile across 7 readiness dimensions in 10–15 exchanges. Optional SIREN (France) and VIES (EU VAT) company lookups are represented in source to pre-fill sector, region, and legal form.
02
A deterministic GDPR risk engine (0–100 scale) is represented in source. Its rules score AI-specific data processing across 18 referenced articles: special category data (+20), cross-border transfers (+15), missing DPIA (+10), no DPO (+10). Flagged operations are tied to mitigations by article. The score is a review aid, not a legal determination.
03
A rule-based workflow maps supplied facts to candidate EU AI Act risk tiers, referenced obligations, and authority records. It is a review aid, not a legal classification or guarantee that a catalogue is current.
04
The source represents a 15-question Digital Maturity Quiz scoring 6 dimensions (strategy, infrastructure, data, workforce, compliance, innovation culture) with a peer-percentile comparison by sector, and an ROI Calculator that computes net investment after assumed grant coverage. Both feed a draft, prioritised 12-month roadmap that requires expert review.
05
Mistral embeddings + Qdrant vector search over a source catalogue of 78 EU and national funding programs, filtered by sector, company size, and technology focus. EDIH regional hub matching is represented for 14 mapped EU countries. A match is not a statement of funding eligibility, and records require source-date checks.
06
A PDF workflow assembles assessment inputs, rule outputs, retrieved records, and a draft roadmap. Source use-case catalogues are illustrative and require provenance and current-source review before external use.
14-country funding coverage
Funding is what turns an EU AI adoption deck into a launched project. The Navigator's RAG is designed to retrieve curated national-program records and fall back to open data for the long tail. Below: the 14 countries with hand-curated catalogues. Records need source-date checks.
Repository surface only: availability, completeness, performance, and operational fitness require separate validation.
FRFrance 2030 PEPR AI envelope, IA Booster, Bpifrance Deeptech + Innovation grants, France Num activator subsidies. Aligned with EDIH regional matching for next-step pilots.
DEGerman federal AI strategy funds, IPCEI-AI cloud + edge funding, BMWK Mittelstand-Digital centres, KfW digital transformation loans. State-level programs surfaced as variants.
ESSpanish national AI strategy under Next Tech 2030, Kit Digital for SMEs, AAPP & EDIH regional kits, Red.es AI adoption grants for digital transformation acceleration.
ITItaly's PNRR digital-transformation pillar, MIMIT AI vouchers, Punto Impresa Digitale (PID) network, CDP venture-debt instruments for deep-tech.
NLAiNed national AI programme, WBSO R&D tax credit, MIT scheme for SME innovation, regional ROM AI portfolios (Brabant, Holland, etc.). EDIH map for hub matching.
BEFederal AI 4 Belgium framework, VLAIO AI grants (Flanders), Innoviris (Brussels), DigitalWallonia AI programs. The catalogue surfaces the regional fragmentation.
SEVinnova AI Innovation calls, AI Sweden partner programs, Sweden's national AI competence centre, EDIH Nordic-Baltic network coordination.
ATFFG AI Mission Austria, AWS digital innovation funding, Austrian Research Promotion Agency basic programme, Vienna Business Agency AI vouchers.
PTAI Portugal 2030 national plan, PRR digital transformation envelope, IAPMEI SME digital vouchers, ANI deep-tech programs.
FIBusiness Finland AI Business funding, Innovation Voucher (Innovaatioseteli), regional ELY Centre AI grants, Sitra AI ethics + capability initiatives.
DKInnovation Fund Denmark Grand Solutions + InnoBooster, Erhvervsstyrelsen SMV:Digital, regional growth-hub AI clinics.
IEEnterprise Ireland AI grants, IDA Ireland strategic AI investments, Skillnet Ireland AI upskilling, Local Enterprise Office vouchers.
CZCzech National AI Strategy under MIT, Technology Agency CR (TAČR) AI programs, CzechInvest AI calls, regional digital-transformation centres.
GRGreek National AI Strategy, Recovery + Resilience Facility digital pillar, Hellinikon megaproject AI initiatives, Elevate Greece deep-tech ecosystem.
Tech Stack
Named technologies identify repository dependencies or adapter surfaces; they do not assert a live integration or deployment.
Source Code
The EU AI Navigator prototype is available for early access evaluation.
To access the source code, please sign the Non-Disclosure Agreement.
Need hands-on EU AI Act compliance support or a strategic AI roadmap? My consulting services complement EU AI Navigator.
Twelve to twenty-four weeks to risk-classify your AI systems, support the provider-led conformity assessment and engineer the evidence, produce the Annex IV technical documentation, and stand up post-market monitoring — with special depth on the high-risk categories that apply to autonomous, industrial, and robotics deployments
Partnership
For teams exploring a reviewable AI readiness assessment, and for advisory practices evaluating the prototype.
From demo to dependable operation. Subscribe for new Physical AI reading.
Unsubscribe anytime. No spam, ever.