AI Lab

From Chatbots to Copilots, Integrations & Agents

A curated lab of practical AI examples that show how everyday teams can move beyond simple chat prompts into useful workflows for support, sales, approvals, documents, and daily operations.

AI AgentsMulti-Agent SystemsRAGEnterprise AIProduction Patterns
Flagship Open-Source Builds

Executive AI systems, built with operator judgment

Batuk and Bhairyam AI are the flagship product builds: one for sovereign team chat, one for sovereign UK legal intelligence.

Batuk sovereign AI chat homepage screenshot
Flagship sovereign AI chatActive build

Builders, founders, operators, AI teams, and people who want model choice

Batuk

Batuk is a flagship AI Lab build for Sovereign AI teams. The application starts with one polished chat surface for Ollama, OpenAI, OpenRouter, Claude, Grok, Sarvam AI, and custom OpenAI-compatible endpoints, then adds user-scoped RAG, shared workspace RAG, personal API keys, admin model routing, OpenAI-compatible API endpoints, token usage analytics, Better Auth enterprise identity, audit evidence, whitelabeling, Docker deployment, SQL product storage, Skills, saved multi-agent workflows, and alpha/PoC MCP integrations for enterprise connector discovery.

Chat with any modelOllama and hosted APIsDocument RAGUser API keysAdmin model routingUser-scoped RAGShared workspace RAGChromaDB and PineconeBetter Auth enterpriseTeams, admins, rolesAudit and complianceDocker deploymentWhitelabel brandingSkills and agentsToken usage analyticsMCP integrationsEnterprise connector directoryAlpha/PoC MCP dashboard
suhasbhairav/sb-chat
Bhairyam AI sovereign legal AI sign-in landing page screenshot
Flagship sovereign legal AIActive build

UK solicitors, legal teams, barristers, in-house counsel, and legal operations leaders

Bhairyam AI

Bhairyam AI is a flagship legal AI build for UK firms that need private document intelligence without handing sensitive matter files to a generic SaaS chatbot. It combines Next.js, FastAPI, Postgres pgvector, local file storage, Ollama Qwen, local embeddings, matter workspaces, document ingestion, citation extraction, authority views, chronology, issues, evidence, voice input, and downloadable Drafts & Agreements DOCX workflows.

UK legal focusMatter-based RAGPDF and Word ingestionPostgres pgvectorOllama QwenLocal embeddingsAuthority extractionCitation linksChronology and issuesEvidence workspaceDrafts and agreementsDOCX exportsVoice inputDocker deploymentSelf-hosted storageMIT licensed
suhasbhairav/sb-legal
Sales Knowledge Engine dashboard showing sales document upload, sales context fields, 10 workflow buttons, workflow output, suggested next actions, and fallback chat
AI LabPrototype
AI Lab Implementation

Sales Knowledge Engine: 10 AI Workflow Buttons for Revenue Teams

A workflow-first Sales Knowledge Engine that demonstrates how revenue teams can move beyond blank RAG chatbots. Built with FastAPI, LlamaIndex, OpenAI, Next.js, React, and Tailwind CSS, the prototype turns uploaded sales PDFs into 10 guided AI workflow buttons for account briefing, pain point discovery, objection handling, proposal drafting, renewal risk review, meeting preparation, CRM notes, and more.

FastAPILlamaIndexOpenAI GPT-4o-miniOpenAI Embedding API
Explore Build
Enterprise Customer Support AI CoPilot dashboard showcasing secure knowledge ingestion, active customer metadata tracking, and multi-tone compliance drafting
AI LabPrototype
AI Lab Implementation

Enterprise Customer Support AI Copilot: Instant Knowledge Intelligence Engine

A high-performance Customer Support AI Copilot demonstrating localized Retrieval-Augmented Generation (RAG). Built with FastAPI, LlamaIndex, and OpenAI to enable real-time compliance checking, SLA resolution drafting, technical troubleshooting verification, and ticketing context injection directly from uploaded policy manuals with local vector persistence.

FastAPILlamaIndexOpenAI GPT-4o-miniOpenAI Embedding API
Explore Build