Warp Open Agentic Development and Oz Orchestration Platform

Warp implements Open Agentic Development with GPT-5.5

Warp is transitioning software development toward a model called Open Agentic Development, where humans define objectives and supervise outcomes while AI agents handle planning, coding, testing, and pull request generation. This approach leverages GPT-5.5 to enable agents to reason across larger problem spaces and improve efficiency, with internal benchmarks showing GPT-5.5 uses 30% fewer tokens per agentic coding task compared to GPT-5.4.

In Warp's own engineering organization, agents now co-create approximately 90% of the company's pull requests. The company currently serves nearly 1 million developers and is used by more than 56% of the Fortune 500.

The Oz Orchestration Platform

Oz serves as the control plane for deploying and coordinating agents across both local and cloud environments. It allows developers to launch agents via a web interface, select specific skills and environments, and monitor long-running workflows centrally.

Key technical capabilities of Oz include:

• Hybrid Execution: Agents can run remotely while developers inspect live sessions and transfer workflows between cloud and local environments without losing context.
• State Management: To maintain reliability during extended workflows, Oz employs context compaction, persistent memory, and dedicated subagents for specialized tasks such as file analysis and code search.
• Automation: The platform supports recurring workflows, allowing agents to function as scheduled cron jobs.

Integration of OpenAI Models

Warp utilizes a tiered model strategy within Oz, routing complex coding and reasoning tasks to stronger model configurations, including GPT-5.5. Additionally, OpenAI models are used as "LLM-as-a-judge" systems within Warp's evaluation pipelines.

CEO Zach Lloyd notes the efficiency of these models:

‘We’ve found that OpenAI models regularly provide frontier-level intelligence while taking fewer tokens and turns to complete the same tasks. The models are especially strong for coding tasks that require reasoning across large problem spaces.’

Business Impact and Future Outlook

Warp's shift toward agentic infrastructure has coincided with significant financial growth, with ARR growing 35x last year and enterprise revenue increasing by more than 500% since Q4 2025.

By open-sourcing its terminal client and building in public, Warp aims to collaborate with the developer community to define the standards for orchestration, supervision, and verification as agents become more autonomous. The company posits that if orchestration is sufficiently advanced, agents may eventually produce more consistent code than loosely coordinated human groups, shifting the human role toward providing product judgment and shared vision.