Knowiki provides a schema‑driven knowledge‑base platform that centralizes and structures raw content from code repositories, Slack exports, meeting transcripts, PRs, and architecture documents.
Funding
Funding not disclosed
Founders
Product
Problem
Software teams often struggle with scattered documentation across repositories, chat logs, and meeting transcripts, making it difficult to provide AI agents with consistent, up-to-date knowledge. This fragmentation leads to divergent agent behavior, duplicated effort, and loss of provenance for decisions and constraints.
Solution
Knowiki offers a schema-driven knowledge‑base platform that centralizes raw content from code repositories, Slack exports, transcripts, PR descriptions, and architecture documents. Users define a KB template that categorizes knowledge into sections such as decisions, constraints, architectural patterns, and agent skills. The system ingests the source material—automatically, under supervision, or manually—indexes it, creates cross‑references and citations, and stores it as a version‑controlled collection of typed knowledge files. From the current KB state, Knowiki generates a structured context file that AI agents can consume, ensuring every agent operates under the same behavioral contract. Agents can also write findings back to the KB, with full provenance, allowing the knowledge base to evolve continuously as work progresses.
Target Audience
Primary customers are development teams and enterprises building LLM‑powered assistants, code‑analysis tools, or automated decision‑support systems that require a single source of truth for technical knowledge.
Features
- Schema templates that define sections, knowledge unit types, and governance rules
- Multi‑source ingestion layer supporting repositories, Slack exports, meeting transcripts, PRs, and architecture docs
- Automated, supervised, and manual ingestion modes selectable per source type
- Indexed, cross‑referenced knowledge base with citations, content hashes, and version control
- Agent‑readable context file that aggregates active constraints, patterns, skills, and open questions for consistent consumption
- Write‑back capability enabling agents to log new constraints, decisions, and discoveries with provenance metadata
- Human review workflow for LLM‑generated entries to maintain accuracy and trust