Applied AI Showcase
Strategy grounded in earned knowledge — AI tools I design, build, and run.
Built, not theorized
The most credible perspective on AI is one grounded in direct experience with the tools. Each project here was designed, built, and deployed by Andy Kannurpatti — using the same modern AI stack he recommends to clients.
These tools are live, functional, and in ongoing development — proof that the economics of building have fundamentally changed, and a window into what that means for your organization.
Revive
Community-owned AI repair platform for home appliances & electronics
A RAG-powered knowledge platform where community repair guides are embedded into a vector database and surfaced through a conversational AI. Users ask repair questions and receive step-by-step guidance grounded in real, moderated content. Built on n8n, Qdrant, Claude API, Supabase, and Netlify.
Market Condition Tracker
Real-time market signal dashboard — running at $0/month
Pulls data from public sources, runs eight deterministic signals — RSI, moving averages, credit spreads, yield curves — and produces a transparent Bullish, Neutral, or Bearish view. Updates hourly on a serverless architecture with no subscriptions or infrastructure cost.
Price Management Dashboard
B2B pricing intelligence tool for a specialty polymer supplier
A working implementation of the McKinsey Price Advantage framework — built end-to-end using synthetic industry data. Pocket margin is calculated for every customer-application pair, share of wallet estimated via Claude AI, and a hardcoded rule engine surfaces prioritized pricing recommendations with break-even analysis and sales action notes. Role-specific dashboards give pricing managers, sales reps, and leadership exactly what they need to act. Built on n8n, Supabase, Claude API, and Netlify.
Connection Explorer
Privacy-first LinkedIn network analysis — runs entirely in your browser
Most people have thousands of LinkedIn connections and no real way to work with them. This tool turns your LinkedIn data export into a searchable, filterable network — segmented by industry, function, seniority, connection age, and relationship strength derived from message history. An AI-powered Project Matcher lets you describe a project or activity and surfaces the most relevant people from your network, using either a locally running Ollama model or the Claude API. No data leaves your browser — everything runs client-side from a single HTML file. Open source on GitHub.
ON EARNED KNOWLEDGE
"Building even a simple system creates a level of understanding that cannot be replicated through secondhand exposure. It clarifies where systems fail, where costs emerge, and where expectations diverge from reality."
If you're exploring how AI can be applied to your business — not in theory, but in practice — Andy works with leadership teams to design and implement AI strategy grounded in direct experience with the tools.