# WebSocket Projects at AI Tinkerers

> Canonical HTML: https://aitinkerers.org/technologies/websocket
> Markdown URL: https://aitinkerers.org/technologies/websocket.md
> Technology record last updated: 2026-09-18T15:13:51Z
> Generated: 2026-09-20T14:48:27Z

WebSocket is the full-duplex communication protocol: it establishes a persistent, low-latency, bidirectional channel over a single TCP connection.

WebSocket fundamentally changes client-server interaction from HTTP's request/response model by delivering persistent, full-duplex data exchange. The connection initiates via an HTTP/1.1 handshake (using the `Upgrade` header), then switches to the dedicated, stateful WebSocket protocol. This single, long-lived TCP connection (often on Port 443 for secure `wss://`) minimizes per-message overhead, eliminating the need for constant polling. The result is a low-latency, real-time data flow: essential for applications like live chat, multiplayer gaming, and financial trading dashboards.

- Official technology site: https://websockets.spec.whatwg.org/
- Public AI Tinkerers demos and talks: 10
- Result page: 1 of 1

## Recent Public Talks and Demos

### [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

### [Six Claude Agents and a Trust Boundary: A Clinical Co-Pilot](https://montreal.aitinkerers.org/talks/rsvp_4qzrWmd9MiM)

A multi-agent clinical co-pilot that runs six specialist Claude agents in parallel on top of a PHI-safe pipeline : every LLM call sees only de-identified text, and re-identification happens server-side after the model returns. Live, I'll load a synthetic 68F new-AFib case where the cardiology plan proposes amiodarone for a patient already on warfarin. The Triage, Differential, Pharmacy, Guidelines, Bias-Check, and Communication agents stream into six panels via asyncio.gather. The Pharmacy agent flags the warfarin–amiodarone CYP2C9/3A4 interaction; Bias-Check independently flags anchoring on rhythm control; the orchestrator detects the cross-agent convergence and elevates it as a high-severity flag. I'll then toggle the Safety Gate off so the audience sees raw PHI hit the LLM and get echoed back — then flip it on and show the audit log assertion (zero raw PHI, only counts + SHA-256 hashes). I'll show the FastAPI/WebSocket code that fans out the agents, the trust-boundary diagram, the structured-field redactor, the live audit log tail, and the keystone pytest (test_pharmacy_catches_warfarin_amiodarone). Repo, logs, and 35/35 passing tests are all open.

- Event context: AI Tinkerers Montreal - May Demo Meetup @ Ubisoft — 2026-05-26 — Montreal
- Public talk page: https://montreal.aitinkerers.org/talks/rsvp_4qzrWmd9MiM

### [The Full Playbook Journey: Form-to-AI Pipeline](https://prague.aitinkerers.org/talks/rsvp_fV-_hHjy-jk)

A complete walkthrough of how minimal form inputs transform into comprehensive sales playbooks through AI generation. The demo follows this flow: 1. Profile Step - Basic information (name, job title, company, target segments) 2. Sales Info Step - Methodology selection (BANT, MEDDIC, SPIN, etc.), sales cycle length 3. Playbook Upload (Optional) - Drag-and-drop existing playbook for AI enhancement 4. Gap Fill Chat - AI interviews user to fill missing information naturally 5. Methodology Details - Value proposition, pricing strategy, sales channels 6. AI Processing - Real-time progress as AI generates the 6-phase playbook 7. Review & Edit - Markdown preview with inline editing 8. Export - Download final playbook as Markdown

- Event context: ❄️ AI Tinkerers Prague – February Meetup — 2026-02-19 — Prague
- Public talk page: https://prague.aitinkerers.org/talks/rsvp_fV-_hHjy-jk

### [Investment Research Assistant](https://seattle.aitinkerers.org/talks/rsvp_2hF-NS5iNhE)

I have built an MVP of an AI agentic assistant along with MCP servers for real-time market data, to help equity investors research about their exiting portfolios and new stock research for US and Indian markets. I will present a working scenarios where I will fundamentally research a stock using my tool and display how this can save retail investors time and simplify their investment process.

- Event context: AI Tinkerers Seattle: January Meetup — 2026-01-31 — Seattle
- Public talk page: https://seattle.aitinkerers.org/talks/rsvp_2hF-NS5iNhE

### [Cost-Effective Real-Time Voice AI for Robotics](https://waterloo.aitinkerers.org/talks/rsvp_3YU21hFfQYg)

Live demo of a working conversational robot (Reachy Mini) using ElevenLabs AI instead of gpt-realtime, while demonstrating the cost-prohibitive reality of real-time voice models of today I'll demonstrate: - The robot having natural voice conversations in real-time (small physical robot will be present) - How I integrated ElevenLabs Conversational AI with Reachy Mini hardware - Why ElevenLabs is dramatically cheaper than OpenAI's realtime API for applications - Head movement synchronization with speech audio (limited) This is a working open-source app on HuggingFace that anyone can install. I'll show the live robot, explain the architecture, and discuss the practical integration decisions that made it work.

- Event context: AI Tinkerers Waterloo – January Meetup — 2026-01-28 — Waterloo
- Public talk page: https://waterloo.aitinkerers.org/talks/rsvp_3YU21hFfQYg

### [Building Voice-First AI Agents: Real-Time Orchestration with WebSockets and LiveKit](https://sf.aitinkerers.org/talks/rsvp_QACgZtoPuiQ)

I'll demo a voice-controlled AI coding agent that delegates work to parallel sub-agents while maintaining bidirectional sync across terminal, web, and mobile. The core challenge: how do you coordinate multiple autonomous agents in real-time while enabling voice interaction? I'll walk through the technical architecture: - Agent orchestration: How the main agent spawns workers for parallel execution and aggregates results - Real-time sync: WebSocket relay architecture that broadcasts messages via Redis PubSub to all connected clients - Voice integration: Connecting LiveKit's WebRTC audio to the agent via separate channels (audio through LiveKit, commands through WebSocket relay) - Async voice notifications: Using debounced triggers to make the voice agent proactively speak when background tasks complete The interesting bits: handling the dual-connection model (audio vs. text), preventing message echo loops in pubsub, and designing the database schema to support hierarchical agent sessions with parent/child relationships. Code walkthrough will cover TypeScript agent orchestration, Python voice agent with STT/TTS, and the WebSocket relay protocol.

- Event context: AI Tinkerers x Google Cloud SF Meetup — 2025-11-21 — San Francisco
- Public talk page: https://sf.aitinkerers.org/talks/rsvp_QACgZtoPuiQ

### [DIY Wedding Translator for Three Languages](https://tokyo.aitinkerers.org/talks/rsvp_YG0dq12PBoM)

I recently hosted a wedding ceremony in Tokyo and needed live captions in English, Mandarin, and Japanese so every guest could laugh or "cry" together. Hiring a professional translator who speaks all three languages was almost unrealistic and too expensive, so I hacked together a browser-based translator that listens, slices audio on a steady beat, and beams subtitles to a shared screen. I’ll demo the heartbeat timer, the playful operator dashboard, and the safety checks that kept the captions flowing while I emceed.

- Event context: AI Tinkerers Tokyo Kickoff – October 10, 2025 — 2025-10-10 — Tokyo
- Public talk page: https://tokyo.aitinkerers.org/talks/rsvp_YG0dq12PBoM

### [Building a Market Intelligence Assistant for cord.com](https://kuala-lumpur.aitinkerers.org/talks/rsvp_N7nRQmsV2aE)

As a jobs platform, our users are always trying to better understand the jobs landscape, eg: how much to pay for a role or what is the demand for a skill. In the past, these requests would be handled manually by a data analyst. Using OpenAI's assistant APIs and code interpreter, we are able to let users explore proprietary data on their own while maintaining security. The demo will show how this was built, and lessons learnt that were not in the documentation.

- Event context: AITKL Demo Day — 2025-07-31 — Kuala Lumpur
- Public talk page: https://kuala-lumpur.aitinkerers.org/talks/rsvp_N7nRQmsV2aE

### [EZ Bomberman: a vybe-coded, multiplayer online game that scales](https://montreal.aitinkerers.org/talks/rsvp_mBPKmNCmMCs)

During this demo, the audience will participate in a live, 5-minute tournament of EZ Bomberman. We will have a leaderboard that shows the winners, and we can stream the games live on the big screen. EZ Bomberman is a multiplayer game inspired from Bomberman that I built on GCP. The architecture allows the game to scale from 0 to thousands of concurrent players seamlessly.

- Event context: AI Tinkerers Montreal - May 2025 Meetup at LightSpeed — 2025-05-07 — Montreal
- Public talk page: https://montreal.aitinkerers.org/talks/rsvp_mBPKmNCmMCs

### [AI agents to conduct recruiting interviews](https://bengaluru.aitinkerers.org/talks/rsvp_3VqlZJ1Qt0A)

Interviews are critical because they determine the people you hire. They are, however, also messy, unstructured, filled with human bias. The interviewer's mood determined by their emotional state, time of the day and even how hungry they are, unfairly affect the outcome for candidates. Additionally, measuring technical capability requires domain expertise. This is a broken system, organizations are missing out on great talent, and candidates barely ever get feedback. At Audition, we are building generalized AI agents to conduct interviews. We envision conducting all the interviews including technical, simulation and even panel interviews to evaluate the ability of the candidate. The human speaks to highly vetted candidates, where they can evaluate them for human qualities like empathy, likeability, optimisms, etc. The key that makes this possible is: - an LLM is the average of human domain knowledge. So using it as a judge allows you to evaluate human's ability too. - AI agent with cognitive architecture that allows them to conduct dynamic interviews, and respond on the basis of the user's input. Thus the no two interviews are ever the same For this demo I will be showing the first steps of this, where an agent is gathering the information to construct itself with a human's help.

- Event context: AI Tinkerers Bangalore - November - RSVP REQUIRED — 2024-11-12 — Bengaluru
- Public talk page: https://bengaluru.aitinkerers.org/talks/rsvp_3VqlZJ1Qt0A

## Related Technologies

- [Python](https://aitinkerers.org/technologies/python) ([Markdown](https://aitinkerers.org/technologies/python.md)) — 662 public demos
- [FastAPI](https://aitinkerers.org/technologies/fastapi) ([Markdown](https://aitinkerers.org/technologies/fastapi.md)) — 181 public demos
- [Gemini](https://aitinkerers.org/technologies/gemini) ([Markdown](https://aitinkerers.org/technologies/gemini.md)) — 188 public demos
- [Node](https://aitinkerers.org/technologies/node) ([Markdown](https://aitinkerers.org/technologies/node.md)) — 99 public demos
- [TypeScript](https://aitinkerers.org/technologies/typescript) ([Markdown](https://aitinkerers.org/technologies/typescript.md)) — 205 public demos
- [AGENTS](https://aitinkerers.org/technologies/agents) ([Markdown](https://aitinkerers.org/technologies/agents.md)) — 20 public demos
- [Amazon Kinesis](https://aitinkerers.org/technologies/amazon-kinesis) ([Markdown](https://aitinkerers.org/technologies/amazon-kinesis.md)) — 5 public demos
- [Anthropic](https://aitinkerers.org/technologies/anthropic) ([Markdown](https://aitinkerers.org/technologies/anthropic.md)) — 36 public demos
- [Anthropic API](https://aitinkerers.org/technologies/anthropic-api) ([Markdown](https://aitinkerers.org/technologies/anthropic-api.md)) — 66 public demos
- [Apache Kafka](https://aitinkerers.org/technologies/apache-kafka) ([Markdown](https://aitinkerers.org/technologies/apache-kafka.md)) — 8 public demos
- [Azure](https://aitinkerers.org/technologies/azure) ([Markdown](https://aitinkerers.org/technologies/azure.md)) — 10 public demos
- [Azure OpenAI](https://aitinkerers.org/technologies/azure-openai) ([Markdown](https://aitinkerers.org/technologies/azure-openai.md)) — 7 public demos
- [Bloomberg API](https://aitinkerers.org/technologies/bloomberg-api) ([Markdown](https://aitinkerers.org/technologies/bloomberg-api.md)) — 1 public demo
- [Claude Agent SDK](https://aitinkerers.org/technologies/claude-agent-sdk) ([Markdown](https://aitinkerers.org/technologies/claude-agent-sdk.md)) — 16 public demos
- [Claude Code](https://aitinkerers.org/technologies/claude-code) ([Markdown](https://aitinkerers.org/technologies/claude-code.md)) — 214 public demos
- [Claude Haiku](https://aitinkerers.org/technologies/claude-haiku) ([Markdown](https://aitinkerers.org/technologies/claude-haiku.md)) — 10 public demos
- [Claude Sonnet](https://aitinkerers.org/technologies/claude-sonnet) ([Markdown](https://aitinkerers.org/technologies/claude-sonnet.md)) — 20 public demos
- [Cloud Run](https://aitinkerers.org/technologies/cloud-run) ([Markdown](https://aitinkerers.org/technologies/cloud-run.md)) — 12 public demos