1.7 KiB
| title |
|---|
| Roadmap |
In progress
Sovereign LLM deployment — Open-source models (Llama 3, Mistral) running fully inside your infrastructure. No data ever leaves your network. Currently in active testing with pilot participants.
Document ingestion pipeline — Upload any document (PDF, Word, plain text) and have it indexed, searchable, and available to the model as context. Handling large corpora with proper chunking and metadata.
Traceability layer — Every response cites its sources. The system logs which documents were consulted, with what confidence, and returns a structured audit trail. pilot-only
Next
Contextual memory — The system learns the regulatory and organisational context of your work over time. Ask a question about GDPR compliance and it already knows you operate under Danish law with specific sector obligations.
Agentic query mode — Instead of single-turn Q&A, the system breaks down a complex question into sub-queries, retrieves evidence for each, and synthesises a structured answer with full provenance.
Meeting summary generation — After a meeting is logged, the system can generate a draft summary, link it to the relevant roadmap items, and post it to the hub for review.
Later
Self-service agent workflows — Non-technical users define repeatable tasks (e.g. "summarise all new documents tagged GDPR weekly") that run automatically.
Developer API — A clean REST API for organisations that want to embed Fenja into their own internal tools and workflows.
Multi-organisation knowledge graphs — Separate, permission-controlled knowledge spaces for different departments or organisations within a single deployment.