Members-Only
Recent Talks & Demos are for members only
You must be an AI Tinkerers active member to view these talks and demos.
Passing the Taste Test: Creating a Fully Agentic Newsroom That Writes Well
See how a fully agentic newsroom creates useful, readable briefings daily without human input. This talk details the system's lifecycle, decision points, and critical evaluations for high-quality autonomous content.
A continuous research system that acts like a personal newsroom. It monitors one reader’s world, commissions its own overnight research, plans what deserves their attention, writes, reviews, and emails a briefing every morning with no human in the loop. The aim is output that’s genuinely useful and readable.
I’ll demo the lifecycle of one piece of content and the decision points in the architecture that make it anti-slop. I’ll start with the briefing that landed in my inbox that morning, then trace it back to where it began and tell the story. I’ll also cover how critical evals are in a product with this many variables.
Autonomous agents monitor external data sources to surface strategic business intelligence.
- ClaudeClaude is Anthropic's flagship family of large language models (LLMs): a high-performance, Constitutional AI system built for safety, complex reasoning, and expert-level collaboration.Claude is a next-generation AI assistant developed by Anthropic, a research firm prioritizing AI safety. The models (including Opus, Sonnet, and Haiku) leverage Constitutional AI to ensure helpful, honest, and harmless outputs, a key differentiator from competitors. Claude excels at complex enterprise tasks: processing massive context windows for in-depth data analysis, generating and reviewing code, and providing expert-level summarization for documents up to 200,000 tokens. It is deployed as a conversational chatbot and via API, offering scalable AI solutions for developers and businesses.
- Voyage AI embeddingsVoyage AI provides high-performance, domain-specific text and multimodal embedding models optimized for retrieval-augmented generation (RAG) and semantic search.Voyage AI delivers state-of-the-art embedding models designed to maximize retrieval accuracy in RAG pipelines and search systems. As Anthropic's preferred embeddings partner, Voyage offers a versatile suite of models, including the high-capacity voyage-3-large, the efficient voyage-3.5-lite, and specialized options for code, finance, and legal sectors (such as voyage-law-2). Supporting context windows up to 32,000 tokens and advanced multimodal inputs (interleaving text and images), Voyage enables developers to build highly precise semantic search engines and context-aware applications with minimal latency.
- Supabase pgvectorSupabase pgvector is the open-source PostgreSQL extension for storing, indexing, and querying vector embeddings at scale.Supabase pgvector transforms your PostgreSQL database into a high-performance vector store, eliminating the need for a separate vector service. It leverages the robust `vector` extension to enable fast, accurate similarity searches (e.g., cosine distance) on high-dimensional embeddings (up to 1536 dimensions for models like OpenAI's `text-embedding-ada-002`). This unified architecture simplifies your stack: you can run complex SQL joins between transactional data and vector data, all within a single, proven, and scalable Postgres instance.
- NextNext.js is the full-stack React framework: it delivers high-performance web applications via hybrid rendering and powerful, Rust-based tooling.This is the React Framework for production: Next.js enables you to build full-stack web applications with zero configuration and maximum efficiency. It supports a hybrid rendering approach (Server-Side Rendering, Static Site Generation, and Incremental Static Regeneration) for optimal speed and SEO performance. Key features include React Server Components, Server Actions for running server code directly, and the App Router for advanced routing and nested layouts. Developed by Vercel, it leverages Rust-based tools like Turbopack and the Speedy Web Compiler for the fastest possible builds and a superior developer experience.
- ResendResend is the email infrastructure platform for developers, built to deliver transactional and marketing emails with React-based templates and high-deliverability APIs.Resend replaces clunky legacy SMTP providers with a modern developer experience. It integrates React via the React Email library: allowing teams to build responsive templates using code instead of brittle drag-and-drop editors. The platform handles the heavy lifting of deliverability (SPF, DKIM, and DMARC) while providing a clean dashboard for real-time tracking. With a global edge network and a 99.9% uptime track record, Resend ensures critical notifications reach the inbox without the typical configuration friction.
Compose Email
Loading recent emails...