Prototyping as Leadership: How a CTO Ships with AI Agents
AI Engineer World's Fair 2026 · 18:18
AI browsers and workplace software
The Browser Company builds browsers that organize online work and bring AI into everyday tasks. Its focus is Dia, which helps knowledge workers answer questions across tabs and connected tools such as Slack, Notion and Google Workspace. Users can synthesize scattered information into reports, generate presentation slides and prepare for meetings with relevant documents. Dia for Work adds single sign-on and administration tools. Its earlier products include Arc, a Chromium-based browser with separate browsing Spaces, and Arc Search, a mobile browser that gathers information from websites into generated answer pages.
Founded in 2019 by CEO Josh Miller and CTO Hursh Agrawal, the company shifted its development focus from Arc to Dia in 2025 while continuing Arc’s Chromium, security and related bug updates. Both browsers use its proprietary Arc Development Kit, which lets engineers prototype native browser interfaces without directly writing C++. Its engineering contributions also include Swift/WinRT, an open-source bindings generator that enables developers to build native Windows applications in Swift using Windows Runtime APIs.
Atlassian completed its acquisition of the company on October 20, 2025, purchasing all outstanding equity for $488.3 million in purchase consideration. Earlier, a 2024 financing led by Pace Capital raised $50 million at a $550 million valuation, bringing total funding to $68 million.
AI Engineer World's Fair 2026 · 18:18
Affiliations reflect their AIE appearances, not necessarily current employment.
Start with Samir Mody's talk to learn how embedded browser prototyping tools let nonengineers test ideas that use browser context.
Samir ModyAI Engineer Code 2025
Watch Hursh Agrawal's talk for the mechanics of preparing business and repository context, launching an agent, and reviewing pull requests, CI results, and code-review fixes the next morning.
Hursh AgrawalAI Engineer World's Fair 2026
Mody highlights personal-data exfiltration through URL parameters and confirmation steps for consequential actions, making security and user control concrete concerns in AI browser development.
Mody describes collecting evaluation and training data and optimizing prompts with GEPA. Agrawal extends evaluation-driven iteration to agent-directed model training with synthetic data and frontier-model ensembles.
Affiliations reflect each recorded session, not necessarily current employment.