Worktrace AI emerges from stealth with $9M seed to automate enterprise workflow discovery

Summary

Worktrace AI emerged from stealth on December 11, 2025, closing a $9 million seed round backed by Conviction, the OpenAI Fund, Logan Kilpatrick, Mira Murati, and Genius Ventures. The company was co-founded by Angela, a former OpenAI product manager with a deep learning background spanning multiple roles before her three years at OpenAI.

The core problem Worktrace targets is the persistent gap between model capability and enterprise adoption. Even as frontier models improve, most organizations struggle to identify which internal workflows are actually automatable, a discovery process that currently depends on consultants or vendor go-to-market teams and can take months.

Worktrace's approach is a desktop application that observes how employees work in real time, flags repetitive or flow-breaking tasks, and matches them to current model capabilities. When a workflow qualifies, the product outputs a prioritized use-case roadmap along with the JSON configuration needed to deploy an agent directly inside OpenAI's agent builder. The timing is deliberate: the company framed the release of OpenAI's o3 model (GPT-5.2) on the same day as a tailwind rather than a threat.

Positioning against model risk is central to the pitch. Rather than building on static model capabilities, Worktrace bets that the share of enterprise workflows amenable to agentic automation will expand as models improve, making its discovery layer more valuable over time. Angela explicitly positions Worktrace as complementary to OpenAI rather than competitive, arguing that distributing AI into enterprise contexts requires ground-level workflow knowledge that OpenAI lacks the headcount to acquire at scale.

The product integrates with existing enterprise tools including Salesforce rather than replacing them, and the company acknowledges a boundary between its agent-discovery focus and broader process-mining tools that surface SaaS adoption gaps or operational bottlenecks. Worktrace is specifically scoped to workflows that can be automated with agents today.

The forward-deployed AI consultant model, where vendors embed human experts inside client organizations, is a direct competitive comparison. Angela argues that scarce AI expertise and continuous model change make that model economically unsustainable for most buyers, particularly mid-market companies that cannot absorb multi-month on-site engagements on an ongoing basis.