Trace raised $3M to build the context layer for AI at work

Today we're announcing that Trace has raised $3 million to build the context and orchestration layer for AI agents inside real companies.

Frontier models can already reason across domains, write working code, and handle multi-step tasks that seemed impossible two years ago. In software engineering, this shift already happened - tools like Claude Code and Cursor changed what a single person can build in a day. We're building toward that same moment for the rest of the business.

Our goal is to be the go-to platform for companies deploying AI agents across operations, client management, HR, finance, and sales. As AI gets more capable, context has become the bottleneck. Our mission is to remove it.

Most companies are stuck where individual employees use AI for one-off tasks, but nobody can scale beyond that. The agent doesn't know your org structure, your processes, or the decisions made in a messaging thread six weeks ago. Without that, the output is generic enough that someone has to rewrite it anyway.

Our vision is simple: AI agents need to understand how a company works before they can do useful work inside it. Trace connects to a company's existing tools, builds a knowledge graph of the organization, and lets teams orchestrate workflows that combine AI and humans.

To get there, we've raised $3M from investors and operators who understand the space, including:

• Y Combinator

• Xeno Ventures

• Goodwater Capital

• Transpose Platform Management

• Formosa Capital

• Benjamin Bryant, Kevin Moore, and other operators via WeFunder

How it works

When a company connects their tools, Trace builds a knowledge graph of entities and relationships — people, teams, projects, clients, tickets. When a workflow runs, each agent queries the graph for exactly the context it needs rather than starting from zero.

Teams describe processes in plain English. Trace turns them into visual workflows — a graph of tasks with dependencies, each with a title, status, assignee, input, and output. AI tasks execute automatically. Human tasks pause and notify the right person. Approval gates ensure nothing irreversible happens without sign-off.

The more workflows run, the more the system learns about how the company operates. After a few weeks, it starts suggesting automations the team didn't think to ask for.

Company Facts

• YC Batch. Y Combinator Summer 2025 (S25).

• Team. Trace is a two-person founding team based in London.

• Product Hunt. Won Product of the Day, Product of the Week, and Product of the Month.

• YC Directory. Most upvoted startup in the YC Summer 2025 batch.

• Traction. Over 550+ active workflows are now live across the customer base. Approximately 10-14% of tasks are currently assigned to our agents with customers seeing clear ROI from reduced headcount needs.

• Use cases. Customers primarily use Trace for back-office automation - account management, client onboarding, document processing, HR screening, and sales operations.

What the funding enables

We're expanding the agent suite, deepening tool integrations, and building a third-party SDK so any agent can plug into Trace's workflow engine.

No single company is going to build every agent for every industry. We think the right model is an ecosystem — Trace as the orchestration and context layer, with specialized agents from third parties handling vertical-specific work. We want to be the infrastructure that any AI agent can operate within, with full organizational context and human oversight built in.

We're also moving upmarket as companies that started with a single workflow expand across their organizations. The context layer gets more valuable with every team that plugs in.