The Y Combinator-backed startup calls itself "Cursor for CTOs." Instead of asking teams to log more, it watches the work itself - code, specs, designs, meetings - and reports what's shipping and what's stuck.
Every engineering leader hits the same wall at roughly the same size. The team gets big enough that no single person can see all of it anymore. Projects that looked fine on Monday quietly stall by Thursday. The founder finds out last, usually in a meeting, usually too late to do anything cheap about it. Mesmer, a company that came through Y Combinator's 2025 batch, is built around that specific moment of blindness.
The pitch it uses is deliberately compact: "Cursor for CTOs." Cursor is the AI code editor engineers have adopted in droves; the joke lands because it names the pattern rather than the feature. The best tools live where the work already happens. For an engineer, that place is the editor. For the person who manages a room full of engineers, the equivalent has always been meetings, Slack pings, and a spreadsheet of dubious accuracy. Mesmer's argument is that a manager's tool should sit in the same place the work does - inside the code, the specs, the designs, the calendar - and read it directly.
Most software that promises "engineering productivity" asks the team to feed it. Log this ticket, tag that story, update the status field before standup. The data is only as honest as everyone's discipline on a busy day, which is to say not very. Mesmer's design choice is to ask for nothing. It connects to the systems an engineering org already runs - GitHub, Slack, Jira, Linear, GitLab, Datadog, Bitbucket - and reads the artifacts that get produced anyway. Out of that, it assembles a live picture of what moved, what shipped, and what's sitting still.
The output arrives in two shapes. The first is a weekly report: a plain-language digest of what progressed, what shipped, and what stalled across the organization. The second is conversational. A leader can ask it the questions they'd otherwise interrupt three people to answer - why is this project late, who should own that piece of work, what does this part of the architecture actually look like, what changed in the roadmap. It is meant to behave less like a dashboard and more like a chief of staff who happens to have read every pull request.
There's a specific behavior this design tries to kill: the update tax. In a scaling company, a startling share of senior time goes to manufacturing status - the standup, the weekly sync, the deck that summarizes the syncs, the follow-up thread that corrects the deck. None of it writes code. Mesmer's wager is that if the picture assembles itself from the artifacts, most of that ceremony becomes optional. The manager's job narrows to the part that actually needs a human: deciding what to do about what the system surfaces.
The company describes one of its core features as friction mapping - an org-wide diagnostic that traces work through its stages and flags where it keeps getting stuck. Software doesn't usually die in the writing. It dies in the waiting: the review that sits for two days, the QA queue nobody owns, the merge that keeps getting bumped, the deploy that waits on one person's approval. Laid out as a pipeline, the bottleneck is often obvious in a way it never is from inside the day-to-day.
The company frames all of this as closing the gap to best-in-class - benchmark where you are, find where you stall, ship the fix. The word that keeps recurring in its material is compound: small productivity gains that stack, month over month, until the org moves at a noticeably different speed. It's a familiar promise in developer tooling, and the honest read is that it's easy to claim and hard to prove. What makes Mesmer's version at least plausible is that it starts from the friction it can actually see, rather than from a generic playbook bolted onto every team the same way.
Alongside the diagnosis, Mesmer offers to act. It pitches "hands-free operations" - automated management chores like per-engineer feedback and team digests - and an auto-merge capability that clears the trivial pull requests so human reviewers spend their attention on the changes that actually deserve it. Rounding it out is a benchmarking layer that measures a team's current numbers against best-in-class targets, and "Mesmer CTO," a senior-advisor persona meant to weigh in on the strategic calls.
The best workflow tool is the one nobody has to remember to use.
The design principle underneath Mesmer's "watch the work" approachMesmer's three founders were not observers of this problem. They were CTOs living inside it. João de Paula, the CEO, is a Brazilian entrepreneur with roughly fifteen years of building behind him. He co-founded Glio, which went through Y Combinator's summer 2013 batch and is frequently cited as the first Latin American company to do so, and later helped build Origin Financial, a company valued around $400 million. Lucas Bruno Silva, the CTO, spent nearly two decades as an engineering leader, including a stint as Director of Engineering at Origin. Sergio Oliveira Bergmann, the VP of Engineering, brings a long run in fintech and digital-banking leadership.
That shared résumé is the quiet advantage. The founders are selling to their own past selves - people running teams of twenty or more engineers who remember exactly what it felt like to lose the thread of their own organization. When the person building the tool has held the buyer's job, the demo tends to skip the part where you explain the pain.
It also shapes what Mesmer chooses to be sensitive about. Founders who have managed engineers know how quickly a productivity tool can curdle into a surveillance one - and how fast that kills adoption on the very teams it's meant to help. Mesmer's framing leans hard the other way: it talks about watching the work rather than watching the workers, about digests and team-level signal rather than individual scorecards for their own sake. That's partly positioning, but it also reads like a lesson learned from having been on the receiving end of a badly aimed metric.
They built the tool for the job they used to have.
On Mesmer's founder-market fitThe early figures are the kind a young company likes to lead with. Mesmer says it reached six-figure annual recurring revenue within a week of launching, that it has more than ten live customers whose valuations add up past $3 billion, and that it has reviewed over 70 million lines of code. Named customers span a range of organizations, and one case study - Rohlik Group, a European online grocer - is cited with a 156% engineering-productivity improvement across five months. Figures like these are self-reported and worth treating as directional rather than audited, but the direction is clear enough.
Which points at the real obstacle. A product that reads a company's source code is asking for a level of access most vendors never get near. Mesmer's answer is to lead with the security posture rather than bury it: SOC 2 Type II certification, a stated policy of no data retention and no training of language models on customer code, private-network deployment for teams that need it, and the usual enterprise controls - SSO, SCIM, role-based access, audit logging. For a tool at this layer, that isn't a footnote. It's the thing that has to be true before a CTO will even take the first call.
Mesmer is not the first company to point analytics at an engineering org. Tools like LinearB, Jellyfish, Swarmia, DX and Faros AI have spent years turning delivery data into charts for leadership. The difference Mesmer is betting on is one of posture. Those products largely surface metrics and leave the interpretation to you. Mesmer wants to read the work, reach a conclusion, and where it can, take the action - closer to an operator than a report. Its truest competitor may not be another vendor at all, but the status quo it's arguing against: the standing meeting and the dashboard nobody trusts.
The business runs the way you'd expect for a tool at this altitude: direct outreach to founders and CTOs, a demo, and priority onboarding for teams past a certain size, sold as an ongoing subscription that layers on top of the tools a company already pays for. That's a sales-led motion, not a self-serve one, and it fits the product - you don't casually swipe a credit card to hand a vendor read access to your entire codebase. The early customer list, weighted toward organizations with real valuations behind them, suggests the pitch is landing with buyers who have the most to lose from flying blind.
Whether an AI chief of staff becomes a category or a feature is the open question. What's less arguable is the shape of the shift underneath it - engineering leadership moving from gathering the signal by hand to reading a signal that assembles itself. Mesmer is one early, specific bet on how that plays out, made by people who spent a decade collecting the scar tissue it's built from. The company is young enough that most of its story is still unwritten; what it has so far is a clear thesis, a founding team that has lived the problem, and a set of early customers willing to let it read the one thing companies guard most closely.