Back to work

AI Knowledge Platform

Anderiq

Turn a company's documents into a governed, citable AI assistant, and embed it anywhere

Live MVP

The problem

In most companies the knowledge people need is scattered across manuals, SOPs, HR policies and datasheets, spread over PDFs, shared drives and inboxes. Finding an answer means knowing which document to open and reading to the right line, or interrupting a colleague who does.

Plain search returns documents, not answers. And when a generic AI does answer, you can't trust it without seeing where the answer came from, or knowing whether the person asking was even allowed to see that information.

How it solves it

Answers, not links

Anderiq runs multi-pass agentic retrieval over your indexed documents and returns a written answer with citation pills that open the exact source passage, so every answer is verifiable.

Governed by design

Access is enforced per document by role and department on every retrieval, backed by a full audit log and rate limits, so employees only get answers from what they are cleared to see.

Trustworthy ingestion

A Python quality-scoring sidecar rejects garbled or low-quality files before they reach the knowledge base, and stale-document detection flags content that has drifted out of date.

Cost under control

A token and cost dashboard with spend budgets and alerts sits over switchable AI providers (Claude, Gemini, Azure OpenAI), keeping the platform affordable as usage grows.

Built to embed

An embeddable public widget drops the assistant onto a public website for anonymous visitors, secured with short-lived HMAC tokens and locked-down CORS.

Key capabilities

  • Multi-pass agentic RAG (up to 3 retrieval passes) with coverage scoring and streaming source citations
  • Governance built in: role- and department-based access enforced on every vector search, plus a full audit log and rate limits
  • Quality-gated ingestion: a Python scoring sidecar rejects low-quality or garbled documents before they reach the knowledge base
  • Cost controls: a token and cost dashboard with spend budgets and alerts, over switchable multi-provider AI (Claude, Gemini, Azure OpenAI)
  • Embeddable public widget for anonymous visitors on a public site, secured with short-lived HMAC tokens and locked-down CORS

Built with

Next.js 16React 19TypeScriptNeon Postgres + pgvectorPrismaNextAuth v5Python FastAPIClaude · Gemini · Azure OpenAIVercel