The First Context Graph for Your People

Every company makes a handful of decisions each month that go on to shape the next five years: who gets promoted, who leads the next project, whose role gets expanded or narrowed. And if you ask any leader, they’ll say that most of these decisions come down to gut feel and memory.

AI is finally getting smart enough to help. Until recently, a model could not reliably summarize a year of someone’s work, let alone weigh it against expectations. Reasoning was the bottleneck. But the frontier models of 2025 crossed that threshold. They can now reason about ambiguous, high-stakes situations at a level that was unthinkable two years ago.

The bottleneck has shifted from the capabilities of the model to what the model knows about your organization. Things like who actually works together, what great looks like on your team, what managers have observed over time, and what each person should be allowed to see.

That context exists. It is just fragmented across HR systems, calendars, Slack, project tools, Google Docs, private notes, past performance reviews, and human memory. Collecting it is tedious. Organizing it is hard. Keeping it current is even harder.

And companies have never moved as quickly as they do today; roles change, teams reorganize, people work across more tools. The amount of context required to understand someone’s contribution has exploded.

At the same time, the decisions themselves have never been more critical. AI is raising the ceiling on what every individual can contribute, which means the stakes on every people decision are rising with it. As teams flatten and each person carries more leverage, putting people in the right environment, on the right problems, with the right support matters more than ever.

This is the problem Windmill is solving.

Windmill is building the first context graph for your people: the structure connecting people to the work they do, the standards they’re measured against, and the observations of their peers, reports, and managers. It’s a living, cited understanding of your workforce, enabling HR to lead strategically, managers to see their teams clearly, and every person’s work to speak for itself.

What is the context graph?

A context graph brings together the layers of information required to understand people at work.

A context graph for your people needs four layers: people, evidence, standards, and perspectives.

Together, they answer four key questions:

• People: Who is this person, where do they sit in the organization, and who do they work with?
• Evidence: What work happened, who drove it, and when?
• Standards: What was expected?
• Perspectives: What have the people closest to the work observed?

People lay the foundation. Evidence gives the facts. Standards give the rubric. Perspectives give you the nuance. Without all four, you are left with an incomplete picture.

Layer 1: People

The foundation of the graph is the People layer.

It starts with the basics: who someone is, what role they are in, who they report to, when they joined, and how their role has changed over time. Most of this information lives somewhere in an HR system.

Who someone is also determines what they can see. Take Slack: some channels are open to the whole company, others are limited to a specific team, and DMs stay between the people in them. The same structure shows up in every tool a company uses. For AI to be useful in people workflows, the graph has to mirror these permissions and keep them current, filtering every query so nothing is visible to anyone who shouldn’t see it.

Then there is the organizational network: how each person at the company works together. This isn’t something you can pull from an org chart; it takes analysis of real work patterns over time to understand who collaborates frequently, which teams are tightly linked, and which peers have enough context to give meaningful feedback.

Layer 2: Evidence

On top of People sits Evidence.

This is the ground truth of what actually happened over time. Each employee’s individual activity, from the work they produced to the impact it had.

Every company already has this evidence, it’s just spread across different places. For an engineer, Git commits, Linear tickets, and Slack threads may already tell the story. For a salesperson, CRM activity, deal progression, and call recordings. The challenge is collecting it consistently, connecting it to the right people, preserving its history, and keeping it up to date.

This is where most tools stop short. Surface-level integrations do not create a useful graph. You have to go deep enough to understand workflows, relationships, ownership, and how work changes over time.

Windmill’s Evidence layer is built through 30+ native integrations with the tools where work already happens — project management systems, code repositories, communication platforms, calendars, docs, and more.

Layer 3: Standards

Evidence alone is not enough. You need Standards to give facts meaning.

A person can be highly active and still be working on the wrong things. A team can be busy and still be misaligned. A manager can see lots of output and still not know whether it meets the bar.

The Standards layer captures what good looks like in this company, on this team, and in this role. That includes role expectations, goals, team priorities, company values, and operating principles. It also includes more local standards: how a function evaluates tradeoffs, what behaviors are rewarded, what success looks like in practice.

Without Standards, evidence is just activity. You can see what happened, but you cannot evaluate whether it was the right thing, done in the right way, against the right expectations.

Layer 4: Perspectives

The final layer is Perspectives.

People decisions rarely reduce to a single right answer. The output matters, but so do trust, communication, growth, leadership, and consistency over time. Managers and peers constantly form observations that never make it cleanly into a system of record: a manager notices someone stepping up in a difficult moment, a teammate sees someone becoming a bottleneck, someone’s metrics look fine but the people around them are struggling.

These observations are often the most important input into a decision. In most companies, they live in memory, in hallway conversations, or nowhere at all.

This is the layer where Windmill captures feedback, private notes, one-on-ones, and manager observations, combining the visible work with the informed perspective of the people closest to it. People own these observations and the decisions that come from them. The graph just makes sure their observations are preserved, permissioned, and surfaced when needed, making decisions more informed and easier to make, but never taking them out of human hands.

A good system should let that context be captured in low-friction ways. Rather than an additional system managers have to maintain, it should meet them where they already are.

Why existing tools do not solve this

Every company has fragments of this graph but almost none have it assembled in one coherent system.

Your HRIS knows who someone is, but not what they worked on. Your productivity tools know what moved, but not what was expected. Existing performance tools are workflow layers sitting on top of forms. They collect inputs at review time; they do not build and maintain a living model of your people that can be accessed at any point during the year.

This matters more than it used to. Every person in your company can point a personal AI at whatever fragments they have and get back a confident, well-written answer that’s wrong, and you usually won’t find out until after the decision is made. This is dangerous, already happening, and will continue to happen if you don’t proactively solve the problem.

This is what Windmill does

Windmill does the tedious work of assembling this graph: collecting context from the systems where work already happens, capturing the human observations that usually go unrecorded, and keeping all of it continuously updated. And it does all of this with permissions built in, so people only see what they should be able to see.

Performance reviews are the first proof point. The reason Windmill’s AI performance reviews feel dramatically better is because there is a context graph behind the review: who the person is, what they worked on, what was expected of them, and what managers and peers have observed over time. Both managers and ICs enter the process with a cohesive record of performance before writing a single word.

Once that graph exists, many other workflows become possible: staffing decisions, promotion cases, coaching recommendations, succession planning, and organizational diagnostics. The review is one expression of the graph. It will not be the last.

Why this matters

AI is making it easier for individuals to build impressive personal workflows. That works well for low-stakes tasks. One person may have a great setup for writing, research, or presentations. Another may not. The cost of inconsistency is low.

But people decisions are not low-stakes.

Promotions, reviews, compensation, coaching, and staffing decisions shape careers and set standards across a company. A bad decision does not just produce a bad output in the moment. It can change the trajectory of a person and unintentionally reset the standards of the organization.

When the stakes are this high, companies need a real system for people context: shared across managers, continuously updated, and grounded in what actually happened.

The best leaders have always known something the tools have never quite captured: people grow when they’re seen clearly, and they leave when they’re not.

The context graph is what it takes for clarity to scale. And that is what Windmill is building.