Describe an authorised or fictional business, its country context and adoption goal.
Engage now
An adoption review connecting the business priority, readiness gaps and sources needed for the next decision.
Inspect
Development prototype
FastAPI, React, PostgreSQL, Redis/Celery, Qdrant, and PDF-generation source boundaries.
Still to validate
Coverage counts describe source catalogues and rules, not a promise that every record is current or complete.
Establish source-date governance and expert validation for country, funding, GDPR, and EU AI Act mappings and generated reports.
源码审查日期
The decision behind the system
Where should an SME begin with AI?
An assessment prototype combining conversational intake, deterministic scoring, country data and model-assisted action plans and document drafts.
Navigator is an adoption and readiness entry point. Aegis focuses on system evidence; Sovereign Fabric is a workspace implementation reference.
Structured intake
Business profiling, country intake and staged assessment keep the readiness record separate from generated narrative.
From assessment to action
Readiness, project, funding and sandbox workflows help organise questions and next steps for review.
Drafts for specialists
Policy, DPIA, processing-record and declaration drafts sit alongside literacy, incident and data-audit workflows. They require current-source and expert review.
Read the complete system walkthrough
Describe
Describe an authorised or fictional business, its country context and adoption goal.
Assess
Inspect the staged intake and readiness scoring separately from generated narrative.
Prioritise
Connect use cases to project, funding and regulatory questions that require review.
Verify
Check draft plans against current sources and specialist advice. No funding eligibility or legal status is established here.
访客及客户数据的 AI 处理限于通过受治理的欧盟端点使用 Mistral。所有者授权的 Helio、搜索及 Product Council 流程已启用。独立的公开来源编辑流程使用 Anthropic，并以受治理的 Mistral 作为备用；该流程不接收网站咨询、客户文档或访客对话。远程 AI 不可用时，确定性工具仍可使用。
源码证据
源码实际能够证明什么
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.
详细技术快照以英文维护。
成熟度
Development prototype
证据审查
源码快照截至
产品与架构决策
Keep deterministic readiness and regulatory rules distinct from generated narrative.
Ground funding and country guidance in curated records and retrieval instead of prompt memory alone.
Run profiling, compliance, use-case, and report work as a staged pipeline with persisted assessment state.
可核验的证据
FastAPI, React, PostgreSQL, Redis/Celery, Qdrant, and PDF-generation source boundaries.
Country-program modules, assessment APIs, and Mistral model and embedding adapters.
Backend tests for selected scoring, funding, reporting, authentication, country-data, and document paths.
声明边界
Coverage counts describe source catalogues and rules, not a promise that every record is current or complete.
No legal advice, compliance status, funding eligibility, report turnaround, or adoption outcome is asserted.
Country data, generated guidance, and regulatory mappings require source-date checks and expert review.
代码仓库证据不等同于客户、部署、认证或成果证据。
架构
一次对话，4 个智能体，一份实战手册草稿
三个源代码层：一个面向中小企业的对话式信息采集层、一条 4 智能体的合规 + 资金流水线，以及一条 PDF 实战手册交付路径。不声称任何报告交付时间。
仅为源代码图：一层可能代表已实现代码、适配器边界或文档设计，并不证明已配置生产服务。
01
第 1 层 — 对话式信息采集
Conversational intake records company context, current AI use and priorities. Browser i18n represents English, French, German, Italian, Spanish, Dutch, Polish and Greek; this is not a translation-completeness assessment.
Configurable Mistral model and embedding routes, staged Celery workflows, Qdrant retrieval and slowapi rate limiting; provider model aliases are not pinned performance guarantees