Tessari

AI-powered knowledge-base helper that integrates Confluence, Google Docs, Notion, and more (WIP)

Role: Solo dev — prototyping, retrieval design, UX
Timeframe: 2025 — Concept Phase

Tessari screenshot

Overview

Tessari is an AI knowledge-base helper that unifies search across Confluence, Google Docs, Notion, and other sources. It retrieves answers with citations, giving teams grounded, explainable insight.

Problem

Teams waste time searching fragmented documents. Conventional search lacks semantic understanding and grounded references, leading to repeated questions and lost knowledge.

Goals

Connect multiple knowledge sources into one search interface
Use hybrid retrieval (keyword + vector) for accuracy
Display AI answers with transparent citations

Solution

Tessari unifies data ingestion, hybrid search, and citation display in one minimalist, AI-forward interface. Each result is explainable and linked to original sources for trust.

Architecture / Approach

Ingestion workers normalizing data from Confluence, Notion, etc.
Chunking and vector embedding pipeline (OpenAI + BM25 hybrid)
Answer composer with citation alignment and re-ranking
Sleek dark UI with glowing accent palette for clarity and focus

Key Screens

Unified search across multiple sources

AI answer view with inline citations

Source preview drawer showing original document

Integrations settings page with connected sources

Outcomes

Concept validated through design mockups
Defined architecture for hybrid retrieval with grounding

Next Steps

Implement real ingestion connectors
Integrate RAG pipeline and conversational interface

Tech Stack

Next.jsRAGEmbeddingsMulti-source

← Back to Projects