# Claude Haiku Projects at AI Tinkerers

> Canonical HTML: https://aitinkerers.org/technologies/claude-haiku
> Markdown URL: https://aitinkerers.org/technologies/claude-haiku.md
> Technology record last updated: 2026-09-18T15:13:59Z
> Generated: 2026-09-21T07:48:03Z

Claude Haiku is Anthropic's fastest, most cost-efficient large language model, engineered for near-instant responsiveness and high-throughput AI applications.

Haiku is the compact, high-speed model within the Claude family, optimized for latency-sensitive tasks like real-time customer service and agentic sub-agent orchestration. It delivers near-frontier coding quality, scoring 73.3% on SWE-bench Verified (Haiku 4.5), matching performance previously seen in larger models like Sonnet 4. Developers utilize Haiku for its exceptional value: pricing starts at $1 per million input tokens and $5 per million output tokens. This model supports a 200,000-token context window and includes multimodal vision capabilities, making it ideal for scalable, budget-conscious deployments that demand speed and accuracy.

- Official technology site: https://claude.ai
- Public AI Tinkerers demos and talks: 10
- Result page: 1 of 1

## Recent Public Talks and Demos

### [A Self-Improving AI OS for the Chronically Self-Employed](https://barcelona.aitinkerers.org/talks/rsvp_AA_yQ6WeHtc)

A personal AI operating system, built as a plain-text git repo, that connects to my entire freelance business in real time: live contracts, open invoices, job feed, inbox, calendar, and meeting notes. It uses that context to find work, draft proposals, track clients, and run a morning brief every day without me asking. This system is built upon the foundation of Nate Herk's AIS-OS combined with Andrej Karpathy's LLM-wiki guidelines. For the demo I'll show it live: open a terminal, pull the job feed from Contra, trigger a skill, watch the wiki update, then walk through the self-audit loop that finds gaps in the system and ships one improvement per week. The repo stays visible throughout. No slides. We may jump into live client / prospects on Contra.

- Event context: AI Tinkerers Barcelona - September Demo Night — 2026-09-17 — Barcelona
- Public talk page: https://barcelona.aitinkerers.org/talks/rsvp_AA_yQ6WeHtc

### [Teaching an Agent to Behave Like a Data Scientist: A Statistical Harness for Tabular ML](https://amsterdam.aitinkerers.org/talks/rsvp_dYE4vQbTMCM)

We built QAgent, an agent that plans and acts like a human data scientist. Coding harnesses such as Claude Code treat data science as a coding problem and miss the nuances: checking for bias, treating data leakage, and ultimately producing results that look correct but rest on flawed methodology. QAgent fixes that with a harness built for data science: the right interface to the data, and constraints that unlock principled data science capabilities in LLMs.

- Event context: AI Tinkerers Demo Night: Agents in Various Forms — 2026-09-03 — Amsterdam
- Public talk page: https://amsterdam.aitinkerers.org/talks/rsvp_dYE4vQbTMCM

### [The Interview That Runs Itself](https://seattle.aitinkerers.org/talks/rsvp_HVXP2AFZt-U)

I built a brand-discovery workshop that runs itself: an ElevenLabs voice agent interviews you about your brand, and while you're still talking, Claude mines the live transcript and pops "nuggets" (quotable insights) onto the screen in real time — then turns each section of the conversation into structured deliverables: a brand voice guide, an ICP profile, and a channel strategy. I used to run this workshop by hand — 90 minutes per client plus waiting on transcripts — so I automated the interviewer out of the room, starting with me. Live, I'll talk to the agent on stage so you can watch insights get captured mid-sentence, walk the two-model architecture on an interactive deck that replays a real session using the app's actual pacing and threshold code, and show the working system deployed at workshop.trymarketingmachine.com.

- Event context: AI Tinkerers Seattle Summer Bash — 2026-07-29 — Seattle
- Public talk page: https://seattle.aitinkerers.org/talks/rsvp_HVXP2AFZt-U

### [Selling to Scientists: Sales Intent Identification for Super Technical Buyers](https://seattle.aitinkerers.org/talks/rsvp_zsBLWVQ1iE4)

Sales intent tools watch behavior — they tell you an account is researching your category. They can't tell you what a researcher is working on, what they're stuck on, or what would make them reply. I sold to AI researchers for a decade by reading their papers and asking questions only someone who read them would ask. I built the system I wish I'd had the whole time: a seller defines their product, target accounts, and ideal buying signals in plain English; the engine mines papers, linked repos, and adjacent sources, then surfaces named opportunities with openers grounded in real pain quotes from the buyer's own writing. The novel part isn't the retrieval. It's that the seller's domain expertise becomes the system's input — captured as natural language signals, not keywords or trained models — and the LLM applies that expertise across a corpus the seller could never read on their own.

- Event context: GTM / Growth Track - Seattle — 2026-05-27 — Seattle
- Public talk page: https://seattle.aitinkerers.org/talks/rsvp_zsBLWVQ1iE4

### [Merging Multi Agent Council Delibrations with Dynamic Analytics to Redesign Transit Infrastructure](https://toronto.aitinkerers.org/talks/rsvp_zWQT1Z51PV8)

A sandbox planning tool where you draw new Toronto transit lines and a 6-agent AI council critiques it; with the ability to see real-time changes and analytics for routes and ridership as the line is developed.

- Event context: AI Tinkerers Toronto - April 2026 - hosted by Shopify — 2026-04-29 — Toronto
- Public talk page: https://toronto.aitinkerers.org/talks/rsvp_zWQT1Z51PV8

### [Equipo Openclaw - Startup](https://bogota.aitinkerers.org/talks/rsvp_CvGKNTJNn5Q)

Es un equipo de 4 agentes construidos para fines específicos, cada uno de ellos tiene skills únicos, tiene tareas específicas según sus roles, y aporta en la construcción y potenciación de la startup.

- Event context: AI Tinkerers Bogotá: Demos for builders — 2026-04-23 — Bogotá
- Public talk page: https://bogota.aitinkerers.org/talks/rsvp_CvGKNTJNn5Q

### [Wisdom Layer: Agent Cognitive Architecture](https://manchester-nh.aitinkerers.org/talks/rsvp_hAZ9fGsxhsI)

The Wisdom Layer is a Python SDK that wraps any LLM with persistent memory that matures, autonomous reflection cycles, self-authored behavioral rules with a lifecycle, and an internal critic that evaluates output against active rules before it ships. The demo at AI Tinkerers ran on Claude Haiku — Anthropic's cheapest model — and showed how a commodity model becomes substantially more reliable when scaffolded with memory, reflection, and enforcement, rather than fine-tuned. Early benchmark showed fabrication rate dropping from 22% to 2% on the same model (n=45 single corpus, broader eval in progress). The same architecture also powers loom-code (AI-assisted coding across 20+ repos) and a computational pharmacogenomics research platform in active collaboration discussions with academic cancer centers.

- Event context: AI Tinkerers Manchester (Bedford), NH - April 2026 Meetup — 2026-04-15 — Manchester NH
- Public talk page: https://manchester-nh.aitinkerers.org/talks/rsvp_hAZ9fGsxhsI

### [From Abacus to AI: Small Potatoes, Big Results](https://hong-kong.aitinkerers.org/talks/rsvp_3XIaboEtPiM)

I'll share my real, month-long journey as a complete non-coder who used today's AI tools (like Grok, Copilot, Claude Haiku, and especially Claude Sonnet) to build actual working things: two World of Warcraft in-game addons for my uncles to track character stats and items automatically, a "Next Bus" app for our island's private bus routes (pulling schedules from PDFs and images), and a simple dinner-bell phone app that rings my phone when someone hits bell button their phone. The talk walks through the messy reality — starting with total confusion (SQL commas feeling like ancient abacus work, AWS looking like a blurry PS2 game, pasting code line-by-line and debugging parentheses I didn't understand), failing a lot, switching models when one got stuck, and eventually getting functional apps and addons. I'll show how I "failed faster" by iterating quickly, using my own low-tech version control (100+ numbered folders), and leaning on different AIs for different strengths. It's not about becoming a pro developer overnight — it's about an ordinary person getting useful results with zero prior experience.

- Event context: AI Tinkerers Hong Kong GBA at the Hive: Creative AI Demos & Technical Show-and-Tell — 2026-03-26 — Hong Kong
- Public talk page: https://hong-kong.aitinkerers.org/talks/rsvp_3XIaboEtPiM

### [PRESENT: Voice + Steward Architecture for Real-Time Multi-Agent Meeting Assistant](https://seattle.aitinkerers.org/talks/rsvp_zHytRNk1_tA)

Present is a custom Voice AI application and real-time meeting producer built with Next.js. In this talk I’ll cover how I refactored the codebase from the legacy three-agent architecture into a simplified voice + steward system. The production pipeline now runs as two lightweight Node processes: a real-time voice agent using LiveKit’s Agents API that listens to room audio, transcribes speech and calls UI tools, and a conductor that delegates to domain stewards running on the server. Stewards read state from Supabase, reason holistically with the latest LLMs (e.g. Claude Haiku via the CANVAS_STEWARD_MODEL env var), and emit structured UI patches. A browser ToolDispatcher executes these actions in TLDraw and React, enabling collaborative canvas editing. I’ll also show how I integrated Model Context Protocol (MCP) to connect to other AI services, added canvas branding and improved the agent tool catalog, and cleaned up the ‘duct-tape’ code from the summer. Live demos will include voice-driven component creation, flowchart steward updates, and multi-agent cooperation.

- Event context: The Holiday Science Fair (Hot Cocoa, Cookies, and Demos!) — 2025-12-19 — Seattle
- Public talk page: https://seattle.aitinkerers.org/talks/rsvp_zHytRNk1_tA

### [Building a Dynamic AI-Powered Trading Card Generation Pipeline](https://chicago.aitinkerers.org/talks/rsvp_MGCTXwzxCM0)

A demo of an automated pipeline that transforms financial market data into a dynamic trading card game. I'll show the complete system: from data extraction to LLM-powered creature generation to final card artwork creation. Includes live generation examples and useful techniques like meta-prompting, where one AI model optimizes prompts for another.

- Event context: AI Tinkerers Chicago February Meetup — 2025-02-18 — Chicago
- Public talk page: https://chicago.aitinkerers.org/talks/rsvp_MGCTXwzxCM0

## Related Technologies

- [Claude Sonnet](https://aitinkerers.org/technologies/claude-sonnet) ([Markdown](https://aitinkerers.org/technologies/claude-sonnet.md)) — 20 public demos
- [Anthropic API](https://aitinkerers.org/technologies/anthropic-api) ([Markdown](https://aitinkerers.org/technologies/anthropic-api.md)) — 66 public demos
- [Claude Sonnet 4](https://aitinkerers.org/technologies/claude-sonnet-4) ([Markdown](https://aitinkerers.org/technologies/claude-sonnet-4.md)) — 18 public demos
- [MCP](https://aitinkerers.org/technologies/mcp) ([Markdown](https://aitinkerers.org/technologies/mcp.md)) — 129 public demos
- [Next](https://aitinkerers.org/technologies/next) ([Markdown](https://aitinkerers.org/technologies/next.md)) — 186 public demos
- [Anthropic](https://aitinkerers.org/technologies/anthropic) ([Markdown](https://aitinkerers.org/technologies/anthropic.md)) — 36 public demos
- [arXiv API](https://aitinkerers.org/technologies/arxiv-api) ([Markdown](https://aitinkerers.org/technologies/arxiv-api.md)) — 1 public demo
- [AWS](https://aitinkerers.org/technologies/aws) ([Markdown](https://aitinkerers.org/technologies/aws.md)) — 38 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)) — 215 public demos
- [Claude Sonnet 5](https://aitinkerers.org/technologies/claude-sonnet-5) ([Markdown](https://aitinkerers.org/technologies/claude-sonnet-5.md)) — 1 public demo
- [DALL-E 2](https://aitinkerers.org/technologies/dall-e-2) ([Markdown](https://aitinkerers.org/technologies/dall-e-2.md)) — 7 public demos
- [DALL-E 3](https://aitinkerers.org/technologies/dall-e-3) ([Markdown](https://aitinkerers.org/technologies/dall-e-3.md)) — 12 public demos
- [Dream by Wombo](https://aitinkerers.org/technologies/dream-by-wombo) ([Markdown](https://aitinkerers.org/technologies/dream-by-wombo.md)) — 3 public demos
- [ElevenLabs Conversational AI](https://aitinkerers.org/technologies/elevenlabs-conversational-ai) ([Markdown](https://aitinkerers.org/technologies/elevenlabs-conversational-ai.md)) — 3 public demos
- [Flash](https://aitinkerers.org/technologies/flash) ([Markdown](https://aitinkerers.org/technologies/flash.md)) — 14 public demos
- [GCP](https://aitinkerers.org/technologies/gcp) ([Markdown](https://aitinkerers.org/technologies/gcp.md)) — 10 public demos
- [GitHub Actions](https://aitinkerers.org/technologies/github-actions) ([Markdown](https://aitinkerers.org/technologies/github-actions.md)) — 20 public demos