# Does Your Computer Belong To Codex? I Went To OpenAI To Ask.

## Executive summary

This discussion explores the shift toward AI-native workplaces, arguing that AI adoption is primarily limited by access to context and data, rather than human aptitude. Speakers detail how tools like Codex and Astra are moving beyond simple chatbots to become universal agents capable of performing complex, multi-step tasks across various domains (e.g., legal, finance, development). The core message for engineers and leaders is that the future lies in building robust integrations, enabling agents to operate seamlessly across disparate systems, and shifting focus from individual output to team-level, automated process improvement.

## Key takeaways

- AI Adoption is Context-Driven, Not Aptitude-Driven: The biggest hurdle for AI is not the user's skill level, but the agent's access to necessary context and data. The ability to connect to and process information from diverse, non-standard sources (like government paperwork or internal documents) is the key differentiator.
- The Shift from Individual to Team Productivity: AI's greatest impact is seen when individual power users build automated processes (e.g., creating a 'chat site' for a financial model) that can then be adopted by the entire team, moving beyond personal productivity gains.
- The Importance of Iterative and Expressive Artifacts: AI is enabling the creation of new 'AI-native artifacts' (like interactive sites) that allow non-technical users to express complex needs that traditional formats (PDFs, spreadsheets) cannot support, filling a gap in knowledge work.
- The Future Requires 'Mid-Agent Work': As agents become more powerful, the user experience must evolve to support fluid, bidirectional interaction—allowing users to interrupt, interject, or guide the agent's run in real-time, rather than just relying on a single question-response cycle.

## Technical details

- Codex/Desktop App: Andrew leads the team behind the Codex desktop app, which is evolving into a universal agent capable of performing complex tasks across different operating systems and applications.
- Astra and 3D/Multimodal Capabilities: Astra is highlighted for its multimodal capabilities, including 3D environment simulation, allowing users to create and interact with photorealistic models (e.g., simulating home renovations from blueprints).
- Computer Use vs. MCP/API: The transition from Machine-Controlled Protocols (MCP) and APIs to direct 'computer use' capability represents a massive tipping point, allowing agents to interact with systems that were previously inaccessible or un-API-ified (e.g., government paperwork).
- Token Efficiency and Orchestration: Advanced techniques for token efficiency include using a multi-agent setup where one model (e.g., Luna) acts as the orchestrator (high context window) while another (e.g., Astra) acts as the executor, minimizing token usage while maintaining context.
- Connectors and Plugins: The power of AI is multiplied by deep integrations. OpenAI is heavily investing in making connectors reliable, fast, and token-efficient to ensure the model has access to all necessary third-party services.

## Practical implications

- Focus development efforts on deep, reliable integrations (connectors/plugins) rather than solely on model intelligence.
- Design user interfaces and workflows that support 'mid-agent work'—allowing users to fluidly interrupt, guide, and refine the agent's process in real-time.
- Shift product development thinking from 'getting code to production' to 'iterative, continuous improvement' (like internal tools).
- Measure the Return on Investment (ROI) of AI by the unit of work accomplished, not just by token count or cost.

## Topics

Generative AI, Agentic Workflows, Productivity Engineering, System Integration, Knowledge Management, Newsletter

Source: https://www.youtube.com/watch?v=TR8RDUzQaMo