Most enterprise AI agents are stuck in a pilot’s purgatory. They can execute a single, isolated task, but they lack the institutional memory to navigate a multi-step business process. The problem isn’t the model’s intelligence, but its context. A new London-based startup, Trace, is betting the solution is a shared knowledge graph that sits between your team and the agents, turning vague prompts into executable workflows.
Trace, part of Y Combinator’s Summer 2025 cohort, announced a $3 million seed round in February 2026 [TechCrunch, February 2026]. The financing was led by Y Combinator, Zeno Ventures, Transpose Platform Management, Goodwater Capital, Formosa Capital, and WeFunder, with angels Benjamin Bryant and Kevin Moore also participating [TechCrunch, February 2026]. The capital is earmarked for deepening the platform’s context-graph engineering and expanding the product and engineering teams to drive enterprise adoption [The SaaS News, March 2026].
The Orchestration Wedge
Trace’s product is a visual workflow builder that connects to a company’s existing systems,Slack, Jira, Notion, email, and Airtable are named integrations [Yes Press, July 2026]. It builds a persistent, organizational knowledge graph from these sources. Teams can then design processes that assign specific steps to AI agents or human employees, with approval gates and handoffs managed by the platform. The core thesis is that AI agents need a persistent, structured understanding of company data, roles, and processes to move from demo to daily driver.
The Team Behind the Graph
The founders bring complementary backgrounds in product and data engineering at scale.
| Role | Name | Prior Experience |
|---|---|---|
| Co-Founder & CEO | Tim Cherkasov | Product Manager at digital-asset custodian Copper, where he delivered an institutional offering for State Street [Y Combinator, 2025-2026]. |
| Co-Founder & CTO | Artur Romanov | Senior Engineer at Deliveroo, where he built a grocery data-enrichment ML pipeline covering millions of items [Y Combinator, 2025-2026]. |
This blend of fintech product rigor and large-scale data systems engineering informs Trace’s approach. It’s not just another chatbot wrapper; it’s an infrastructure play to become the routing layer for operational work.
The Scale and Skepticism Test
For a workflow orchestrator, the technical breakdown is straightforward but critical. The platform must perform two core functions reliably: continuously syncing and indexing context from disparate SaaS tools, and then reliably executing defined workflows with low latency. The first is a data engineering challenge, familiar to Romanov’s background. The second is a systems reliability problem, where any failure in a multi-step agentic process breaks user trust completely.
The sober assessment lies in what could go wrong at scale. Trace’s value is contingent on deep, ongoing integration with a company’s core systems. This creates a significant implementation burden and potential performance bottlenecks as the context graph grows. Furthermore, the product competes with the internal tooling capabilities of large platforms like Salesforce or Microsoft, which are increasingly baking similar orchestration features directly into their ecosystems. Trace’s wedge is its agnosticism and focus, but that also means it must constantly prove it’s better than the native path.
Sources
- [TechCrunch, February 2026] Trace raises $3M to solve the AI agent adoption problem in enterprise | https://techcrunch.com/2026/02/26/trace-raises-3-million-to-solve-the-agent-adoption-problem/
- [The SaaS News, March 2026] Trace Raises $3 Million in Seed Round | https://www.thesaasnews.com/news/trace-raises-3-million-in-seed-round/
- [Y Combinator, 2025-2026] Trace: Context-aware orchestration between humans and AI | https://www.ycombinator.com/companies/trace-so
- [Yes Press, July 2026] Trace (YC S25): The Manager Built for the Human-AI Workforce | https://yespress.io/trace-yc-s25