Enterprise leaders have spent the last two years experimenting with AI agents — and quietly discovering a frustrating truth: brilliant models don’t automatically translate into operational impact.
London-based startup Trace believes it has identified the missing piece.
Fresh out of the Y Combinator 2025 summer cohort, Trace has raised $3 million in seed funding to tackle what it calls the “AI agent adoption problem.” Investors include YC, Zeno Ventures, Goodwater Capital and several angel backers.
But the story here isn’t the capital. It’s the thesis.
The Problem: Agents Without Context
OpenAI, Anthropic and other labs have built increasingly capable AI systems. Tools like OpenAI’s enterprise products and Anthropic’s Claude models can draft reports, analyze data and automate tasks.
Yet inside most enterprises, these agents remain underutilized.
According to Trace CEO Tim Cherkasov, the issue isn’t intelligence — it’s placement.
“AI labs are building brilliant interns,” he has said publicly. “We’re building the manager that knows where to put them.”
In other words, companies have access to powerful agents, but lack the orchestration layer that connects them meaningfully to business workflows.
From Prompt Engineering to Context Engineering
Trace’s system begins by building a knowledge graph from a company’s existing tools, such as Slack, email, Airtable, and project management systems.
By mapping relationships, processes, and communication flows, Trace creates an operational blueprint of the organization.
When a user enters a high-level objective — for example, “Build a microsite” or “Draft a 2027 sales strategy” — the system generates a step-by-step workflow. It assigns certain tasks to AI agents and others to human contributors.
Crucially, when an agent is invoked, it receives precisely scoped data from the knowledge graph.
This shift reflects a broader industry evolution.
In 2024, enterprise AI was largely about prompt engineering — refining instructions to coax better outputs. By 2025, the conversation has moved toward context engineering: embedding structural understanding into AI deployments.
As CTO Artur Romanov has framed it, whoever delivers the best context at the right time becomes foundational infrastructure for AI-first companies.
Competitive Landscape: A Crowded Field
Trace enters an increasingly competitive market.
Anthropic recently introduced enterprise-focused agent integrations, while workplace platforms like Atlassian are embedding native AI features into tools such as Jira.
The challenge for startups like Trace is differentiation.
Instead of competing directly with pre-built departmental agents, Trace positions itself as the orchestration layer — a system that coordinates both external AI models and internal human workflows.
This approach mirrors the rise of workflow automation leaders in previous SaaS waves, where the companies controlling integration layers often captured durable value.
Leadership in the AI Infrastructure Era
Cherkasov and Romanov represent a new generation of AI founders focused less on model-building and more on deployment infrastructure.
This is a critical distinction.
According to analysis from McKinsey & Company, enterprises struggle not with AI experimentation, but with scaling adoption across departments.
The bottleneck is rarely capability. It is integration.
Trace’s knowledge graph model aims to address that bottleneck directly, reducing friction for onboarding agents and minimizing manual configuration.
For enterprises wary of AI complexity, the simplicity of orchestration may prove decisive.
Why This Founder Matters for 2026
By 2026, the enterprise AI conversation will likely shift from “Can agents perform tasks?” to “Can agents operate cohesively across the organization?”
Companies that master internal AI orchestration will outperform those deploying siloed tools.
If Trace successfully positions itself as the connective infrastructure layer — not just another agent provider — it could become an essential enabler of AI-native operations.
The $3 million seed round signals early confidence. But the larger opportunity lies in shaping how companies architect AI systems from the inside out.
In the AI arms race, model builders capture headlines.
Infrastructure builders quietly define outcomes.
Trace is betting that context — not code — will determine which enterprises truly become AI-first.
And if that bet proves right, Cherkasov’s leadership could position him as one of the more consequential operators in enterprise AI’s second wave.
UK-based journalist covering UAE entrepreneurship, executive branding, and leadership growth across global business ecosystems.
