Familiar

Overview

Familiar is an offline, local, open-source, free, and model-agnostic screen-history tool designed to capture live work context for AI agents. Based on newsletter mentions, it records screen and clipboard activity every few seconds and stores that context as Markdown, making it easier for local AI systems to reference what a user has been doing without depending on a cloud service or a single model provider.

For AI Product Managers, Familiar matters as an example of infrastructure for context-aware agents: it turns desktop activity into structured memory that can power assistance, automation, and retrieval. It is also noteworthy as the comparison point for OpenAI’s “Computer History,” positioning Familiar as an early open alternative in the emerging category of screen-history and computer-use context tools.

Key Developments

• 2026-03-08: Tal Raviv shared a detailed account of spending significant time “vibe coding” a landing page for Familiar, pushing back on the idea that AI-enabled work is always quick or effortless.
• 2026-03-16: While building a desktop app with Claude Code, Tal Raviv found that Familiar’s workflow used the Markdown `source image` field to interpret and act on UI screenshots, despite image support not being documented.
• 2026-04-28: Tal Raviv launched Familiar as an open-source app that captures screen and clipboard activity every 4 seconds into Markdown so local AI agents can use live work context.
• 2026-08-15: Tal Raviv said OpenAI released “Computer History,” describing it as OpenAI’s version of Familiar and emphasizing Familiar’s offline, local, open-source, free, and model-agnostic approach.

Relevance to AI PMs

• Evaluate context pipelines for agent products: Familiar shows how passive desktop capture and clipboard history can become usable agent context. AI PMs can use this pattern when defining memory, grounding, and retrieval requirements for assistant products.
• Assess privacy and deployment tradeoffs: Because Familiar is local and offline, it offers a strong reference point for products where enterprise buyers care about data residency, privacy, or avoiding cloud dependence.
• Prototype model-agnostic agent experiences: Familiar demonstrates a way to separate context capture from model choice. AI PMs can apply this approach to reduce vendor lock-in and support multiple model backends over time.

Related

• Claude Code: Familiar was used alongside Claude Code, where Tal Raviv observed that screenshots referenced in Markdown were automatically interpreted during app-building workflows.
• Tal Raviv: Creator and primary evangelist of Familiar in the newsletter mentions, documenting both its launch and practical usage.
• Vibe coding: Familiar appeared in discussion around the realities of “vibe coding,” especially the gap between AI hype and the actual effort required to build useful products.
• OpenAI: OpenAI became relevant when its “Computer History” release was framed as analogous to Familiar.
• Computer History: Presented as OpenAI’s version of Familiar, making it the closest comparison point in the newsletter coverage.