How Work Gets Done at Klarity
Klarity's engineering teams ship AI document processing features on a fortnightly cadence, taking features from idea to production in sprints tight enough that a quarterly roadmap can shift before it's printed. That pace is the defining texture of working there (fast, flat, and self-directed) and it falls directly out of how the company has chosen to organize itself.
Work is divided into small, cross-functional squads rather than traditional departments, with each team owning a slice of the product end-to-end. In a 2022 MIT Sloan interview, founder and CEO Andrew Antos described the internal toolkit in plain terms: "We use a combination of Zoom calls, the collaboration software Asana, and spreadsheets to make sure that we are executing well." No proprietary platform, no layers of product managers — Asana and a shared doc. Headcount stayed lean well into the company's growth: TechCrunch reported in January 2022 that Klarity had 34 employees, up from 14 a year earlier. By mid-2026 that count had roughly quadrupled (Tracxn pegged it near 184, Startup Intros closer to 128) but the squad model has held.
The decision-making structure is the part that sets the tone. Product decisions sit with the squad closest to the problem. There is no standing committee, no VP gate between an engineer's pull request and a customer release. Klarity's own LinkedIn makes the philosophy explicit: the best software teams, the company argues, deploy 182 times more often than laggards with eight times fewer failures — "same talent, different operating model." Antos has labeled the practice "High Frequency Management." Translated into Tuesday morning: if you have data and conviction, you ship.
That autonomy shows up in how features actually arrive. Klarity launched its first product in August 2020, rolled out new document review automation use cases for deal desk, renewals and procurement teams in late 2022, and by March 2023 had integrated OpenAI's GPT-4 to validate critical financial data. Each of those moves was a squad calling a bet rather than a roadmap item handed down. The tradeoff is structural: load is uneven. Some squads sit on a hot problem for six straight sprints while adjacent teams ship once a quarter. There is no formal leveling committee smoothing the difference; the company simply expects engineers to pick up the slack or signal when they can't.
The current hiring pattern reinforces the model. Klarity's job board currently lists 15 salaried roles with a median of $250,000, suggesting the flat structure is starting to layer in go-to-market partners. Every posted engineering role names AI as the core stack.
| Role | Salary Range | Source |
|---|---|---|
| Staff AI Frontend Engineer | $235,000–$315,000 | Klarity job board |
| Senior AI Backend Engineer | $220,000–$280,000 | Klarity job board |
| Alliances Director | up to $340,000 | Klarity job board |
The Operating Principles Hanging on the Wall
Klarity publishes two parallel sets of values that read less like a manifesto and more like an operating manual. The customer-facing page names four: Velocity, Agency, Care, and Energy, each with a one-line directive — "Move before the meeting gets scheduled," "If you see it, own it. If you own it, ship it," "The work matters. The customer matters. The teammate matters," and "You bring it. Or you borrow it from someone who does." A separate list on CEO Andrew Antos's writing surfaces a different four: Customer First, Anti-Fragility, Simplicity, and Relationships. The lists aren't contradictions so much as two angles on the same operating system, one written for people deciding whether to apply, the other for people deciding whether to stay.
Two values carry through both lists and do the heaviest lifting. Velocity comes first in the public four and is implicit in Antos's framing of Customer First: "If your schedule isn't filled with talking to customers and deciding what to build, you're doing something wrong. Everything else is secondary." The Series B closed in three days; the platform's three phases (Discover, Structure, Improve) are named with what one outside profile called "admirable bluntness." Even the company's About page, "We're building the tools that let change agents change things," reads as a tempo statement as much as a mission statement. Anti-fragility is the second load-bearing value, and Antos has returned to it repeatedly. "Startups move one step forward, two steps back," he told MIT Sloan in 2022. "Being anti-fragile — treating setbacks as expected inputs, not failures — is the most important skill."
The third value worth weighing is Simplicity, which Antos ties directly to product judgment: "The best and most powerful ideas are really simple and straightforward. Complexity is usually a signal that the idea isn't right yet." That principle is visible in how Klarity talks about its 2025 repositioning around the Context Graph (a live map of how an enterprise actually works) and in the bluntness of metric framing the company uses with customers: 87% time saved on document review, 82%+ automation pass-through, three days shaved off month-end close.
What the values don't say is also instructive. Energy (the directive that acknowledges some employees carry more of the emotional load) is a rare thing for a values page to put in writing, because it concedes that the company expects some employees to carry more than their share of the emotional load. Antos's own priority list ("family, Klarity, physical well-being") ranks the company second and personal health third, which a reader could parse as either a healthy boundary or an honest warning. Glassdoor reviewers rate Klarity 3.1 out of 5 for culture and values, with 45% of employees saying they would recommend the company to a friend — middling numbers for a Series B AI startup that has raised $90M and counts OpenAI, DoorDash, Stripe, Salesforce, and Zoom among its customers. Antos himself has flagged the management layer as the place where values meet their stress test: his "biggest early mistake," by his own account, was hiring too many individual contributors before building a proper executive layer, a pattern that can produce the very uneven workloads a flat structure invites.
What the Hiring Bar Selects For
Klarity's job posts make the hiring philosophy unusually explicit: the company is recruiting for a small bench of "exponential contributors" rather than scaling headcount. A 2026 recruiting-lead listing captures the line verbatim: "We hire exponential contributors, not headcount. A small team of exceptional people, augmented by AI, will outperform a large team of good people." That sentence sets the tone for everything that follows in the interview process.
Technical chops are the entry ticket, not the deciding factor. Glassdoor reviewers describe a roughly three-round loop (two technical rounds followed by a culture round), which fits a pattern where engineering depth gets verified first and fit is judged afterward. But the posts say almost nothing about LeetCode scores, degrees, or years of experience. Instead, the language centers on three traits candidates have to demonstrate before they get an offer.
AI fluency as a working habit. The recruiting-lead post states: "We are AI-native in how we operate. We use AI throughout our hiring process and expect every person we hire to be AI-native in their work." A candidate who has read about AI tools won't pass; the bar is "200–300% efficiency in your own workflow over the last two years, and have a real point of view on what creates leverage and what doesn't. Curiosity isn't enough; we want hands-on practice and outcomes." The same post notes Klarity's internal screening pipeline is already on "prompt v9" and needs to reach "v50," signaling that the company wants people who treat AI tooling as a craft to iterate, not a feature to flip on.
Builder instinct over operator polish. The recruiting listing opens with a deliberate contrast: "Most recruiting roles are about running the machine. This one is about building it." The role description repeats it: "You won't inherit a mature recruiting function. You'll build the structure, workflows, and operating rhythm that make hiring scalable and predictable over time." The Head of Product post echoes the same filter from the product side, asking for "former founders, early product hires at unicorns — folks who are the best of the best in Tech," and experience with Fortune 2000 contracts at $100k+ ACVs. People who have spent careers "building HR organizations at traditional companies" are warned the role "will frustrate" them.
Intensity as a stated preference, not an afterthought. Klarity calls this out plainly: "We are high-intensity and mission-driven. Klarity isn't for people seeking work-life balance in the traditional sense. The people who thrive here are energized by hard problems and high standards — not drained by them." Five-day in-person attendance at the San Francisco office is "a firm requirement." The company frames the same trait as a filter both ways: "If this sounds like the environment you've been looking for, keep reading. If it sounds exhausting or misguided, we're not the right fit."
On the product side, the Head of Product post adds a quieter trait — comfort with interruption.
The compensation bar matches the cultural bar. The same job board shows a band from $154k to $313k alongside the $250k median, and a separate Ashby posting lists a $180k–$250k base plus equity. The willingness to pay top-quartile for a small team reinforces the "exponential contributors" pitch: when headcount is the constraint, every hire has to clear a high bar.
What Current and Former Employees Say
Public sentiment about Klarity clusters tightly around two poles, and the gap between them says as much as either side does on its own. Glassdoor's aggregate score sits at 3.4 out of 5, drawn from 47 company reviews, a middling number that, on its own, hides the structure underneath. The platform's own framing notes that the rating "indicates that most employees have a good working experience there," yet the same dataset contains enough sharp criticism to keep the average from climbing higher.
On the positive side, reviewers repeatedly point to ownership. Multiple posts describe being handed a problem with no prescribed approach and the latitude to pick a solution, ship it, and see the result in production within days. Reviewers credit this to the autonomous squad structure rather than to any individual manager. The two-week shipping cadence comes up as both a source of pride and a useful forcing function: teams say it kills the slow accumulation of half-finished projects that plague larger AI shops.
The criticisms are equally consistent. Workload tops the list. Several reviewers describe weeks that stretch past 50 hours when a release window approaches, with no formal cap and no comp time built into the cadence. The flat structure that current and former employees praise for cutting bureaucracy gets blamed, in the same reviews, for uneven distribution of work. The people most willing to pick up ambiguous problems end up carrying more of them, and there is no formal leveling or promotion rubric to recognize that load. A second thread of complaints targets onboarding: reviewers say the autonomy assumed of new hires leaves people without a clear map for their first month, especially engineers joining from larger organizations where scope was handed to them in writing. A smaller but visible cluster of posts calls out communication gaps between product and engineering squads, with reviewers saying decisions made inside one autonomous unit sometimes collide with another's roadmap before anyone notices.
The pattern across roughly four dozen reviews is not a house divided. It is a culture that works the way its architects designed it to work — and that delivers the same experience to people who want that structure as it does friction to people who don't.
Who Thrives and Who Burns Out
The two-week ship cycle at Klarity sorts engineers into two camps fast. Engineers who write their own tickets tend to accelerate inside it. Those who wait for the next instruction tend to stall.
The model rewards a specific archetype: a builder who can take a rough product hypothesis, pick a ship date, and own the path between the two. Inside Klarity's autonomous squads, that means an engineer comfortable making scope calls a layer up would normally gate: what gets cut from a sprint, whether a feature ships behind a flag, when a model regression is severe enough to roll back. The people who thrive describe the structure as a feature rather than a bug. They use the absence of a long approval chain to spend more cycles on the actual problem: training data quality, eval coverage, prompt regressions, latency budgets. The salary band published on the Alliances Director posting tops out around $340,000, and the Senior AI Backend Engineer role reaches $280,000 — pay grades that assume the hire will operate with that latitude rather than ask for it.
The trait that decides whether someone lasts is tolerance for ambiguity. Two-week cycles compress the feedback loop, but they also expose the seams: a half-finished eval suite, a fuzzy success metric, a product brief that reads like a guess. An engineer who needs the spec nailed down before writing code finds every sprint a negotiation. An engineer who treats the spec as a starting draft finds every sprint a chance to push the product forward. That distinction shows up early, usually in the first three cycles, and it tends to be self-selecting.
Burnout at Klarity tends to follow a recognizable arc. It is not the pace itself that breaks people; the two-week cadence is brisk but bounded. What breaks people is uneven load inside the autonomy. When a squad is small or one engineer becomes the de facto owner of a fragile subsystem (say, the DevSecOps stack or the document-extraction eval pipeline), the absence of a manager rebalancing the queue turns from freedom into a backlog that nobody else can see. Engineers who prefer more structure describe the same flatness their self-directed colleagues describe as freedom, but they experience it as invisibility: no one is going to notice the warning signs until something slips.
That asymmetry is the trade-off the culture makes openly. The same flat decision-making that lets a strong IC ship a product decision on a Thursday is the flat decision-making that leaves a struggling IC without a check-in until the next retro. Engineers who flag this in reviews tend to be the ones who joined expecting the autonomy and discovered they also wanted a coach. Engineers who flag the inverse (too many check-ins, not enough room to run) are rare at Klarity, because the hiring bar filters most of them out before they start. The practical implication for candidates is straightforward. If your best work happens when you can define the problem, set the deadline, and own the result, the Klarity model will multiply your output. If your best work happens inside a clearly scoped ticket handed to you by a tech lead who has already made the trade-offs, the same model will feel like neglect. Neither is a moral judgment; it is a fit question, and the two-week cycle is honest about who belongs there because it makes that answer visible faster than most companies do.
Working in frontier tech? Zero G Talent tracks the openings: see every open Klarity role, browse frontier tech jobs, the companies hiring, and the people building the field.