The AI Vibe Check (Replay Vision) | Los Angeles .

Members-Only

Recent Talks & Demos are for members only

Exclusive feed

You must be an AI Tinkerers active member to view these talks and demos.

June 18, 2026 · Los Angeles

PostHog: Replay Vision AI

See how to create configurable scanners to summarize session replays at scale, then use AI to analyze the results for deeper insights.

Overview
Tech stack
  • Temporal
    Temporal is the Durable Execution platform: It orchestrates complex, long-running workflows as code, guaranteeing fault-tolerant completion at any scale.
    Temporal is your microservices orchestration solution, built for mission-critical reliability. It enables developers to define complex, stateful workflows directly in code (Go, Java, Python, TypeScript, etc.) using the Temporal SDKs. The platform's core feature, Durable Execution, automatically persists application state and handles all retries, timeouts, and failures: your workflow always picks up exactly where it left off. This eliminates writing thousands of lines of custom error-handling logic. Global enterprises like NVIDIA and Salesforce leverage Temporal to ensure their transactions, deployments, and AI agents run invincibly.
  • Gemini 3 Flash
    Gemini 3 Flash delivers Pro-grade reasoning and multimodal capability at Flash-level speed and cost: the optimal balance for high-volume, low-latency agentic applications.
    Gemini 3 Flash is engineered for efficiency, combining the frontier intelligence of the Gemini 3 family with the speed and cost-effectiveness of a Flash model. This model is a high-speed, high-value asset, specifically optimized for agentic workflows, multi-turn chat, and coding assistance: it clocks an impressive 218 tokens per second output speed, dramatically outpacing larger models like GPT-4 and Claude 3. It maintains Pro-level reasoning while supporting a massive 1M token context window and advanced multimodal inputs (text, images, audio, video, PDFs). Leverage its low latency for near-real-time interactive applications, complex video analysis, and high-volume data extraction at scale.
  • OpenAI text-embedding-3-large
    OpenAI's most capable text embedding model, delivering high-dimensional vector representations for precise search and retrieval.
    OpenAI's text-embedding-3-large is a high-performance embedding model designed for advanced semantic search, clustering, and retrieval-augmented generation (RAG). It outputs vectors of up to 3,072 dimensions, offering a major accuracy boost for both English and non-English tasks over previous generations. Built-in support for Matryoshka representation learning allows developers to shorten embeddings (down to 256 or 512 dimensions) to save on storage and compute costs without losing critical semantic accuracy. Priced at $0.13 per million tokens, it provides a highly scalable, precise solution for enterprise-grade vector search databases.
  • PostHog
    PostHog is the all-in-one, open-source platform for product engineers: analyze, test, observe, and deploy features from a single stack.
    PostHog delivers a comprehensive developer platform, eliminating tool sprawl for product teams. It bundles critical tools like Product Analytics, Session Replay, Feature Flags, and A/B Testing into a single, cohesive system. Engineers gain a single source of truth for customer data, allowing them to debug code, ship features faster, and understand user behavior without complex integrations. The platform is open-source, offers a generous free tier, and is trusted by over 190,000 teams (as of 2024) for its transparent, usage-based pricing model.
  • rrweb JSON
    An open-source session replay format that serializes real-time DOM mutations and user interactions into a lightweight, pixel-perfect JSON event stream.
    Engineers use rrweb to record and replay web sessions without the heavy performance overhead of traditional video (like MP4 format). By leveraging the browser's MutationObserver API, the library serializes the initial DOM snapshot and subsequent user interactions (such as mouse movements, scrolls, and keystrokes) into a structured JSON schema. This text-based event stream allows developers to reconstruct exact user sessions programmatically, making it a critical asset for remote debugging, customer support triage, and product analytics platforms.