# Claude API Projects at AI Tinkerers

> Canonical HTML: https://aitinkerers.org/technologies/claude-api
> Markdown URL: https://aitinkerers.org/technologies/claude-api.md
> Technology record last updated: 2026-09-18T15:13:59Z
> Generated: 2026-09-20T16:41:29Z

Access Anthropic's state-of-the-art Claude models (Opus, Sonnet, Haiku) via the RESTful Messages API, integrating advanced AI capabilities directly into your applications.

The Claude API is Anthropic's direct developer interface for integrating their powerful large language models (LLMs) like Claude 3.5 Sonnet and Opus into production applications. It utilizes a robust Messages API for all conversational and generative interactions, supporting a massive 200,000-token context window for deep document analysis and sustained, complex reasoning. Developers leverage its Constitutional AI framework for built-in safety and utilize key features like Tool Use (function calling) and the Message Batches API for cost-efficient, high-volume processing. This is the direct, pay-as-you-go route for full feature control and cutting-edge model access.

- Official technology site: https://docs.anthropic.com/claude
- Public AI Tinkerers demos and talks: 20
- Result page: 1 of 1

## Recent Public Talks and Demos

### [Multi-agent Drawing Analysis](https://orange-county.aitinkerers.org/talks/rsvp_W9mR7pms2vA)

Agent teams to analyze construction drawings of any discipline, and return PDFs and report with findings. Uses the Claude API and deployes many agents. Splits the drawings into many tiles and assigns an agent to each tile. This program is useful to engineers and builders.

- Event context: AI Tinkerers Orange County: Tuesday, July 21, 2026 at Centercode — 2026-07-22 — Orange County
- Public talk page: https://orange-county.aitinkerers.org/talks/rsvp_W9mR7pms2vA

### [Contextual Synthesis: Engineering a Multi-Agent Context Extraction Pipeline for Stakeholder Audio](https://doha.aitinkerers.org/talks/rsvp_X1LkLLGS_0I)

I built a multi-agent workflow engine that ingests long-form, unstructured stakeholder meeting transcripts and dynamically transforms them into targeted, context-aware corporate assets—including domain-specific LinkedIn posts, official press releases, and structured follow-up emails with clear action items.

- Event context: AI Tinkerers Doha: Round 2 — 2026-06-15 — Doha
- Public talk page: https://doha.aitinkerers.org/talks/rsvp_X1LkLLGS_0I

### [Why I gave up on agent voting: hard escalation in a 3-reviewer LLM pipeline](https://nyc.aitinkerers.org/talks/rsvp_gIVCOg1tRfI)

Redline is a production-readiness review tool for AI agents. Three specialized LLM reviewers (Engineer, Risk, Business) read an agent transcript against a company's pasted rulebook, mark up the dossier in three colors, and a synthesizer returns a Deploy, Hold, or Kill verdict. Live demo: pasting a real agent dossier into sanjitkangovi.com/redline, watching the three reviewers annotate in parallel, then walking through the deterministic synthesis layer in the backend that decides the final verdict. I'll show the prompt structure for each reviewer role, the JSON contract between reviewer and synthesizer, and the server-side escalation logic firing when a critical Risk finding contradicts the other two reviewers.

- Event context: NY Tech Week Demo Day ft PostHog, Convex, Veris, & HPE — 2026-06-03 — New York City
- Public talk page: https://nyc.aitinkerers.org/talks/rsvp_gIVCOg1tRfI

### [How To Sell Cards Online (Fast)](https://nurnberg.aitinkerers.org/talks/rsvp_j3cTo9Vft5s)

Automated trading card identification, grading and pricing using local VLMs (Qwen3.5), MLX, prompt caching and visual chain-of-thought - fast, on consumer hardware, without training data.

- Event context: AI Tinkerers Nürnberg: April Meetup — 2026-04-22 — Nürnberg
- Public talk page: https://nurnberg.aitinkerers.org/talks/rsvp_j3cTo9Vft5s

### [Jules: An AI Agent That Thinks, Argues, and Writes](https://austin.aitinkerers.org/talks/rsvp_f0lg_KSKY8s)

I built Jules — a Claude Code-based AI agent with a personality, opinions, and structured decision-making workflows. Jules isn't just a coding assistant — it's a strategic thinking partner that follows multi-step reasoning processes, pushes back on my ideas, runs adversarial reviews before recommending anything, and writes content in my voice. I'll demo the live system: how a single prompt triggers structured thinking workflows with visible reasoning steps, how the agent challenges my framing before agreeing with me, and where this is headed — multi-agent coordination through Paperclip and a conversational gateway called Hermes.

- Event context: AI Tinkerers Austin: April 2026 Demo Night — 2026-04-16 — Austin
- Public talk page: https://austin.aitinkerers.org/talks/rsvp_f0lg_KSKY8s

### [Claude Flow V3: Orquestacion Multi-Agente en Accion — Demo con Auto-Discovery de Skills](https://pereira.aitinkerers.org/talks/rsvp_YXtINN6XyJk)

Claude Flow V3 es un framework open-source que orquesta multiples agentes de IA en swarms jerarquicos con memoria persistente, consenso distribuido y auto-aprendizaje. Para demostrar sus capacidades, presento la integracion de Claude Flow con la metodologia Auto Research: un ecosistema donde las skills de IA se monitorean constantemente via SkillSentinel (EWMA + CUSUM), y ante cualquier degradacion, se lanza automaticamente un proceso de reparacion con enjambres de 8 agentes sin intervencion humana. La charla cubre el framework primero (topologias, AgentDB, hooks, ReasoningBank) y luego lo demuestra en vivo con el pipeline completo: monitoreo, deteccion, mutacion y convergencia estadistica.

- Event context: AI Tinkerers Pereira — The Builders Lab — 2026-03-25 — Pereira
- Public talk page: https://pereira.aitinkerers.org/talks/rsvp_YXtINN6XyJk

### [Niuwn AI: Building an AI Twin That Speaks As Me to Strangers](https://bremen.aitinkerers.org/talks/rsvp_BIo4d4pBN0I)

I built a system where anyone can create a conversational AI version of themselves, not a chatbot that answers questions about you, but one that talks as you in first person to anyone who visits a shareable link. Think of it as replacing your static CV or LinkedIn profile with something people can actually have a conversation with. The core idea is a two-agent architecture. One agent (the Curator) works with you privately; you talk to it, upload your CV, share your blog posts, import your LinkedIn export and GitHub profile. It interviews you like a thoughtful journalist, asks follow-ups, spots gaps, and organizes everything into a structured knowledge base. The second agent (the Twin) is what visitors see. It reads from that knowledge base, speaks in your voice, respects boundaries you've defined, and gracefully handles things it doesn't know instead of hallucinating. What makes the system interesting technically: the entire knowledge base is plain Markdown files, transparent, editable, and version-controllable. Retrieval uses memsearch for hybrid BM25 + vector search over those markdown files, which turns out to work really well for professional data where you need both semantic understanding and exact matching on names, dates, and technologies. The two agents have strict asymmetric access, the Curator writes, the Twin only reads, which keeps the trust boundary clean and makes prompt injection defense on the public-facing side much more manageable. The system is built with GDPR in mind, owners can view, edit, or delete anything their Twin knows, and visitors are informed upfront that conversations may be reviewed by the owner. During the Live demo, I'll teach the Twin something new on stage and let the audience interrogate it.

- Event context: AI Tinkerers Bremen — 2026-03-25 — Bremen
- Public talk page: https://bremen.aitinkerers.org/talks/rsvp_BIo4d4pBN0I

### [What If Humans and Agents Had a Portable Cryptographically-Chained Ledger (Single source of truth)](https://nyc.aitinkerers.org/talks/rsvp_nTFADbAUdbI)

I built an MCP server that captures relevant context (Incidents, Decisions, Product Governance, etc.) into a cryptographically chained ledger — think blockchain-style hashing applied to a product development context window. I'll walk through the actual implementation: how the MCP hooks into Claude Code and Cursor, how context gets hashed and chained across sessions, how a voice agent or slack bot reads against the ledger to route decisions to the right person. The goal is that the MCP is always running in the background, capturing the most important context across meetings, tools, etc.; so building with teams and agents can match the speed of using the tools themselves.

- Event context: March Demo Day, hosted by Flowglad — 2026-03-18 — New York City
- Public talk page: https://nyc.aitinkerers.org/talks/rsvp_nTFADbAUdbI

### [The Zeitgeist Experiment](https://columbus.aitinkerers.org/talks/rsvp_d1Aa3s-VUoc)

A side project I have been working on that I hope to someday commercialize. Its a little bit like reddit but completely different.

- Event context: AI Tinkerers - Columbus March Meetup — 2026-03-02 — Columbus
- Public talk page: https://columbus.aitinkerers.org/talks/rsvp_d1Aa3s-VUoc

### [Claude in the Channel: Building an Autonomous Frontend Agent for No-Code Teams](https://tokyo.aitinkerers.org/talks/rsvp_ILK68XIsX3A)

I’ll demonstrate how I turned a Slack channel into a functional "Frontend Desk" where designers and admin staff ship production code without touching an IDE, or having to pay for a no-code platform or a content delivery network (CDN). The system connects a Slack bot to a Claude-powered agent sitting on a server with a live clone of our React/Vite/Tailwind repository. I’ll walk through the technical plumbing that allows the bot to: * Ingest design specs and SVGs directly from Slack threads. * Contextualize those assets within a complex codebase to perform precise UI updates. * Trigger a password-protected staging environment for instant visual feedback. * Execute a `/approve` flow that handles the merging and deployment to GitHub Pages.

- Event context: AI Tinkerers Tokyo - Toranomon Meetup - February 19, 2026 — 2026-02-19 — Tokyo
- Public talk page: https://tokyo.aitinkerers.org/talks/rsvp_ILK68XIsX3A

### [Building a 32-Character Murder Mystery Game Generator](https://valencia.aitinkerers.org/talks/rsvp_JIVtHp5Lz5M)

I'll walk through building a production murder mystery generator that creates full games with 6-32 characters using Claude's API. The talk covers the real costs and constraints of AI at scale, my automation architecture using Make.com for orchestration, and the specific technical challenges I hit when generating complex JSON structures for game content. I'll show the actual webhook flows, how I handle character generation failures, and my debugging approach when Claude outputs invalid JSON that breaks the entire workflow.

- Event context: AI Tinkerers Valencia February Meetup — 2026-02-17 — Valencia
- Public talk page: https://valencia.aitinkerers.org/talks/rsvp_JIVtHp5Lz5M

### [From 4 Hours to 30 Minutes: Building an AI-Powered Treatment Plan Generator](https://miami.aitinkerers.org/talks/rsvp_SATx8PtKtEE)

I'm a Board Certified Behavior Analyst. Treatment plans in my field take 4+ hours to write — 40-80 pages of highly individualized clinical documentation. I built a web app that lets clinicians upload clinical documents, uses Claude to extract data and generate narratives, behaviors, and goals, then outputs insurance-compliant plans. The BCBA just approves, rejects, or edits — no typing. I'll walk through the architecture: document upload → extraction pipeline → Claude API integration → review interface → formatted output. I'll demo the live system and share the trade-offs: hallucination guardrails for clinical accuracy, insurance-specific formatting, and compliance across different states. Built in December 2025, First plan submitted was approved for over $40K in treatment funding January 2026.

- Event context: AI Tinkerers Miami Meetup — The 2026 Builder Kickoff — 2026-02-11 — Miami
- Public talk page: https://miami.aitinkerers.org/talks/rsvp_SATx8PtKtEE

### [Edge-First AI: Per-Student Agents on Cloudflare](https://singapore.aitinkerers.org/talks/rsvp_2uQx8tjb9FA)

ow we built mcq.sg - an AI tutoring platform where every student gets their own stateful agent. The defining infrastructure choice: Cloudflare Durable Objects. Each student has a dedicated DO with built-in SQLite storing mastery data, attempt history, and recommendation queue. The AI agent runs inside the DO - no centralized database, no bottleneck, no cold starts for returning students. I'll show the code for: - Per-student DO with local SQLite schema - Multi-model extraction pipeline (Moondream for detection, Claude for reasoning) - Async batch processing for 50% cost savings on non-realtime tasks Built by 2 people. Previously led engineering on ParkingSG and RedeemSG at Open Government Products.

- Event context: AI Tinkerers - The Age of AI & Infrastructure (Singapore) — 2026-02-11 — Singapore
- Public talk page: https://singapore.aitinkerers.org/talks/rsvp_2uQx8tjb9FA

### [Transcript-First CRM: Turning Call Recordings into Proposals, Summaries, and Account Intelligence](https://austin.aitinkerers.org/talks/rsvp_I-D1INcTSo0)

A walkthrough of our work-in-progress internal CRM built on Airtable that treats call transcripts as the primary data source rather than manual entry. I'll show the technical pipeline: Granola captures calls → transcripts flow through chained prompts to extract client needs, generate proposal drafts, and update account health scores → GitHub integrations pull in delivery context → weekly summaries get auto-generated for account managers. The interesting technical bits: prompt chaining architecture for going from raw transcript to structured proposal sections, how we handle context windows when stitching together multiple calls with the same client, and the surprisingly simple scoring system that flags accounts needing attention. This is held together with Airtable automations and API calls—I'll show what's working, what's duct-taped, and what's still broken.

- Event context: AI Tinkerers Austin: January 2026 Demo Night — 2026-01-23 — Austin
- Public talk page: https://austin.aitinkerers.org/talks/rsvp_I-D1INcTSo0

### [Active Story v2](https://chicago.aitinkerers.org/talks/rsvp_o9un7ggZcuY)

What it is: Active Story is a bedtime story app where kids and parents create stories together with AI. V1 was simple – you prompt, AI continues, back and forth. It works, but stories feel a bit flat. Characters wander. Endings just... happen. V2 adds: Voice in, voice out – just talk. AI narrates back. No typing, no reading. Actual bedtime vibes. Story craft – agentic setup so stories have real arc, characters stay consistent, endings actually land Memory – AI tracks what happened so your ideas genuinely shape where the story goes What I want to learn: Can agentic architecture make AI a genuinely better storytelling partner? Where does memory help? What makes a co-created story feel good?

- Event context: AI Tinkerers Chicago January Meetup ft. PostHog — 2026-01-20 — Chicago
- Public talk page: https://chicago.aitinkerers.org/talks/rsvp_o9un7ggZcuY

### [Building FacultyFinder.io: An AI-Powered Academic Discovery Platform with Smart CV Matching](https://toronto.aitinkerers.org/talks/rsvp_AA3BkfpGWxY)

FacultyFinder.io is a comprehensive platform that helps researchers discover and connect with faculty members across Canadian universities using AI-powered matching and intelligent web scraping. I'll demo the complete technical pipeline: 1. Comprehensive Data Source: Combining different sources of data (names from the university website, publications from PubMed, and OpenAlex) 2. AI-Powered Profile Analysis: Multi-LLM CV analyzer (Claude, GPT, Gemini, Grok) that extracts research interests, keywords, and skills from uploaded CVs/resumes 3. Smart Matching Engine: Database-driven faculty recommendation system using PostgreSQL with semantic matching 4. Publication Integration: OpenAlex API integration with ORCID-based disambiguation, processing 42,595+ publications with 88.6% file size optimization 5. Real-time RESTful API: FastAPI backend serving 10+ endpoints with advanced filtering and search The demo will show: uploading a CV, the AI extracting research keywords, querying our database of 500+ faculty members, and generating personalized recommendations with match scoring. I'll walk through the code for the web scraper, the CV analysis prompt engineering, and the database schema optimization.

- Event context: AI Tinkerers Toronto - December Meetup sponsored by Auth0 and TribalScale! — 2025-12-03 — Toronto
- Public talk page: https://toronto.aitinkerers.org/talks/rsvp_AA3BkfpGWxY

### [Predicting Human Decisions in Agentic Workflows](https://toronto.aitinkerers.org/talks/rsvp_hWxp80wzOEU)

Building off the human decision prediction model work done by the team behind Centaur. My team has been experimenting with ways to implement this decision prediction model in agentic workflows in order to reduce cognitive load for human reviewers and experts.

- Event context: AI Tinkerers Toronto - November 2025 Meetup at Shopify! — 2025-11-10 — Toronto
- Public talk page: https://toronto.aitinkerers.org/talks/rsvp_hWxp80wzOEU

### [Our in-house analytics agent & Lessons learned from agent-to-agent communication](https://london.aitinkerers.org/talks/rsvp_QpzOxgqxuiQ)

This is a rapid, two-part "show-and-tell" straight from our build logs. The 1-Day BI Platform: We'll show how we built a self-serve internal BI tool for our (very small) team. We'll walk through the implementation of a simple Streamlit app, hosted on EC2, that uses Claude to generate and run data queries on the fly. This lets our non-technical folks "vibe up" dashboards instantly. We'll show the core code and prompt structure. When Agents Talk to Agents: We'll share technical insights from having our two core AI agents, "Jack" and "Jill," in production and communicating with each other for the last three months. This isn't a product pitch; it's a technical review of the architecture and the messy discoveries we made building a product reliant on agent-to-agent communication.

- Event context: AI Tinkerers London Meetup - 29th October 2025 — 2025-10-29 — London
- Public talk page: https://london.aitinkerers.org/talks/rsvp_QpzOxgqxuiQ

### [Privacy-First AI: Building Deeply Personal AI Without Data Exploitation](https://poland.aitinkerers.org/talks/rsvp_EDiSYXTpofs)

Life Navigator is an agentic composer that orchestrates multiple AI tools while keeping your personal data out of permanent storage. I'll demonstrate the technical architecture for building agentic AI systems that can access your daily notes, manage tasks, and provide deep personalization while using explicit context control and smart prompt caching. The core challenge: Building agentic AI that's both deeply personal and privacy-preserving. Most AI apps store your data permanently in their databases. Life Navigator's agentic composer uses explicit context control where users choose exactly what personal data the AI sees, combined with prompt caching that reduces API calls by 85% and makes privacy-preserving architecture economically viable. I'll demonstrate: * Agentic composer architecture: tool orchestration, context assembly, and multi-step workflows * Explicit context control system using link expansion - users control every piece of personal data * Smart prompt caching strategies that reduce costs and data exposure * Tool calling patterns optimized for both privacy and cache efficiency * Real agentic workflows: task management, note editing, research, planning This is production-level agentic AI in action - I'll show my actual personal assistant managing my real daily notes, tasks, and goals, then walk through the technical implementation that enables this without permanent data storage.

- Event context: AI Tinkerers Poland #4 - Meetup in Warsaw (June) — 2025-06-26 — Poland
- Public talk page: https://poland.aitinkerers.org/talks/rsvp_EDiSYXTpofs

### [Creating AI Agents in JavaScript](https://miami.aitinkerers.org/talks/rsvp_m6id4m01LSs)

I’ll demo how I built a multi-agent system in pure JavaScript, where agents collaborate to complete complex tasks using LLMs, tool usage, and a shared memory model. The system is entirely open source, built using KaibanJS, and runs both in Node.js and the browser. I’ll walk through: - How agents are defined using roles and goals - How they plan, act, and react with real-time state updates - The architecture behind teams, tools, and LLM integrations - A live coding walkthrough showing agents solving a real problem together No slides. Just code and agents at work.

- Event context: AI Tinkerers - Miami: May Meetup at The LAB — 2025-05-29 — Miami
- Public talk page: https://miami.aitinkerers.org/talks/rsvp_m6id4m01LSs

## Related Technologies

- [Python](https://aitinkerers.org/technologies/python) ([Markdown](https://aitinkerers.org/technologies/python.md)) — 662 public demos
- [Claude](https://aitinkerers.org/technologies/claude) ([Markdown](https://aitinkerers.org/technologies/claude.md)) — 173 public demos
- [Claude Code](https://aitinkerers.org/technologies/claude-code) ([Markdown](https://aitinkerers.org/technologies/claude-code.md)) — 214 public demos
- [FastAPI](https://aitinkerers.org/technologies/fastapi) ([Markdown](https://aitinkerers.org/technologies/fastapi.md)) — 181 public demos
- [React](https://aitinkerers.org/technologies/react) ([Markdown](https://aitinkerers.org/technologies/react.md)) — 219 public demos
- [Cloudflare Workers](https://aitinkerers.org/technologies/cloudflare-workers) ([Markdown](https://aitinkerers.org/technologies/cloudflare-workers.md)) — 22 public demos
- [Gemini API](https://aitinkerers.org/technologies/gemini-api) ([Markdown](https://aitinkerers.org/technologies/gemini-api.md)) — 20 public demos
- [Next](https://aitinkerers.org/technologies/next) ([Markdown](https://aitinkerers.org/technologies/next.md)) — 186 public demos
- [Node](https://aitinkerers.org/technologies/node) ([Markdown](https://aitinkerers.org/technologies/node.md)) — 99 public demos
- [OpenAI API](https://aitinkerers.org/technologies/openai-api) ([Markdown](https://aitinkerers.org/technologies/openai-api.md)) — 520 public demos
- [SQLite](https://aitinkerers.org/technologies/sqlite) ([Markdown](https://aitinkerers.org/technologies/sqlite.md)) — 28 public demos
- [Supabase](https://aitinkerers.org/technologies/supabase) ([Markdown](https://aitinkerers.org/technologies/supabase.md)) — 90 public demos
- [Tailwind CSS](https://aitinkerers.org/technologies/tailwind-css) ([Markdown](https://aitinkerers.org/technologies/tailwind-css.md)) — 22 public demos
- [TypeScript](https://aitinkerers.org/technologies/typescript) ([Markdown](https://aitinkerers.org/technologies/typescript.md)) — 205 public demos
- [AgentDB](https://aitinkerers.org/technologies/agentdb) ([Markdown](https://aitinkerers.org/technologies/agentdb.md)) — 1 public demo
- [Airtable](https://aitinkerers.org/technologies/airtable) ([Markdown](https://aitinkerers.org/technologies/airtable.md)) — 6 public demos
- [Amazon EC2](https://aitinkerers.org/technologies/amazon-ec2) ([Markdown](https://aitinkerers.org/technologies/amazon-ec2.md)) — 7 public demos
- [Amazon Titan](https://aitinkerers.org/technologies/amazon-titan) ([Markdown](https://aitinkerers.org/technologies/amazon-titan.md)) — 1 public demo