Team
Mötley Crëw
This team is at maximum capacity.
Project Concept
TL;DR
A live podcast enhanced by CrewAI agents that dynamically sketch, diagram, and visually surface ideas onto an infinite TLDraw canvas. Think of it as your podcast’s interactive assistant augmenting conversations in real-time with visuals, fact-checking, and context-aware insights.
What we’re building
- Interactive Casting App – Conversations enriched by real-time AI-generated visuals on TLDraw’s infinite canvas.
- Visual Augmentation Layer – CrewAI agents use transcripts and canvas context to identify interesting moments, automatically adding visual aids such as diagrams, images, quotes, or fact-checking the conversation.
Why it matters
- Dynamic Storytelling – Immediate, contextual visuals deepen listener engagement and clarify complex ideas.
- Real-world CrewAI Showcase – Demonstrates powerful agent orchestration, context-aware dynamic UI and the potential for real-time multimodal interactions.
- Reusable Tech Stack – Applicable to podcasts, educational livestreams, collaborative ideation sessions, or developer tutorials.
Ideal Tech Stack
| Layer | Tooling | Why? |
|---|---|---|
| Audio/Video | LiveKit | Reliable, low-latency WebRTC streaming |
| AI Agents | CrewAI | Multimodal generation and intelligent orchestration |
| Canvas | TLDraw | Infinite, collaborative, flexible drawing environment |
| Generative UI | Tambo | Context-aware AI-driven UI elements |
| Front-end | React (Next.js) | Clean, performant frontend |
Progress so far
- Established LiveKit + TLDraw integration with basic streaming and canvas interactions.
- Implemented Whisper-based speech-to-text pipeline integrated with agent framework.
- AI agents semi-currently working placing basic visuals (notes, images) onto TLDraw canvas during conversations.
How you can help
| Skill | Where to dive in |
|---|---|
| Realtime UI/UX | Improve canvas interactions, dynamic UI elements, intuitive gestures, and custom visual components |
| Agent Prompt Tuning | Extend and optimize CrewAI agent tools, async triggers, and prompts for accurac, responsiveness and overall enhancement |
| DevOps / Backend | Deploy scalable backend infrastructure supporting real-time events in a collaborative environment |
| Canvas Dynamics | Help shape conversation flow, identifying optimal UX for visuals and tool responses |
| Design & Polish | Develop cohesive visual language, UI refinements, and interaction aesthetics |
Stretch goals
- Visual Playback – Users can revisit past conversations via animated canvas replay.
- End-of-Show Recaps – Agents produce concise, interactive summaries ready for sharing and prompt-driven exploration.
Open-source ethos
Currently MIT license. Clone, remix, and adapt freely.
Check the repo.
Working title: PRESENT – alternative ideas (like “AI Jamie”) warmly welcomed.
Entry
Status: Submitted
Last saved: May 18 at 2:27 PM PDT
Team Roster (team is at max capacity)
Message board not available for this team yet.
Ben Steinher Team Lead RSVP Approved
active listener at Thought Partners
Coordinated project direction and integrated LiveKit with the frontend.
Turning conversations into action. 🗣️🤹♂️
Jumping on as many problems as I can. 🧑🍳🫡
Love discussing as niche topics as possible, find me to get an expert opinion on things I've never thought about before.
product, America, projects, events, operational efficiency, problem-solving, mechanical systems, troubleshooting, niche topics, pragmatic solutions
Currently, Thought Partners Consulting.
Previously, Open Interpreter.
Previously, Music.
Eventually, The meaning of life.
Michael Magan RSVP Approved
Founder at Tambo
Built the Next.js interface and Tambo components
I'm a technical product manager, turned founder.
Mostly interested in learning about building AI agents.
I'm mostly building tambo.co. But I have a few side projects!
Ian McCann RSVP Approved
Cto at Steno.ai
Ian developed the real-time audio stream processing and agent backend using LiveKit. He set up the Whisper transcription pipeline and integrated voice activity detection, so the system knows when to trigger the AI response. He also implemented the Conversational Assistant agent’s logic to remain silent until addressed or needed, mirroring good human assistant behavior. Ian’s work resulted in the LiveKit backend working reliably to feed the conversation to the AI.
As the Chief Technology Officer of Steno.ai, a Techstars-backed startup providing AI-driven content universes to help brands and creators scale their reach and revenue. I lead the development and deployment of cutting-edge natural language models and systems. I have over 7 years of experience in data science, programming, and artificial intelligence, with a focus on natural language processing, deep learning, automation, and natural language generation.
Very interested in meeting others who are deep in the weeds of building AI backends including RAG pipelines and prompt workflows.
I am currently building an AI application businesses can deploy on their cloud in one click. This includes a preprocessing pipeline to embed terabytes of data into a postgres database and a backend to create a prompt from a user's query. As well as dashboards to expose insights to the client about their data and chat interactions.
Suma Movva RSVP Approved
Applied Scientist (AI/ML) at Microsoft
Implemented CrewAI agents and the intelligent behaviors of PRESENT – integrating external APIs (OpenAI, Perplexity, YouTube) into the agent workflow and ensuring the outputs are accurate and contextually relevant. She also played a major role in debugging and achieving the “working version” of the project (as evidenced by numerous commits)
Suma Movva is an Applied Scientist (AI/ML) in Microsoft Security. I have been in <5 person founding teams in fintech companies helping raise ~2 million funds as a product lead. Recently, I’ve had the founder itch again and I’m looking to find a cofounder or join an early stage - Series B startup in fintech, social good, security, and general SaaS.
Looking to connect with potential cofounders or join a pre-seed - Series B startup with a strong mission.
Write/Build Journey here: https://open.substack.com/pub/sumamovva/p/the-fifth-vortex-what-sedona-taught?r=97hla&utm_medium=ios
Tinkering:
1. Hospital Contract - Automated pharmacy invoice handling.
2. AI journal - AI driven journalling assistant
3. Bamboo Daily - Make productivity intentional, 1 goal at a time.
Martha Schumann RSVP Approved
Founding Software Engineer at LOGIC
Helped with initial brainstorm.
Martha Schumann is a software engineer with over 16 years of experience in startups and large companies alike. Currently, she serves as a Founding Software Engineer at LOGIC. She previously had a principal engineering role at Highspot, and senior engineering roles at Uber (2013-2018).
Very excited to work with anthropic/copilot kit. I <3 Claude. Claude has 10x'd my productivity
Most of my time is spent building a startup that bridges the gap for non-technical domain experts to integrate with AI systems. But I'm also trying to make a sentient furby with a raspberry Pi zero and tinyllama....ask me about it (and please help me with the soldering)