What I Learned at a Unicorn
(AKA How Lisa Was Born)

Hypergrowth is exhilarating — until 75 people suddenly report to a manager who’s only ever led one intern. Here’s what I saw, and why it led me to build Lisa.

A couple of years ago, I jumped ship from my comfy corner of a big gaming studio to join a bona fide unicorn startup. At the time, it felt like the obvious move — the energy, the mission, the headcount hockey-stick that everyone in tech dreams about. What I didn’t fully appreciate was what that hockey-stick actually looked like from inside the people team.

What I witnessed during hypergrowth — managers suddenly leading 75 people who’d only ever managed one intern, title inflation, retention headaches, and enough legal landmines to make your head spin — is exactly why Lisa exists.

The Manager Problem Nobody Talks About

Here’s the thing about hypergrowth: the company scales, but the managers don’t always scale with it. You promote your best individual contributors because they’re brilliant, and six months later they’re managing a team of fifteen people across three time zones. They’re trying to figure out how to have a performance conversation for the first time while simultaneously navigating a leave request they’ve never seen before.

In a well-resourced enterprise, that manager would have an HRBP sitting two desks away. At a startup doubling every eight months, you’re lucky if your HR team has time to answer Slack messages before 11pm.

“Your best ICs become managers overnight. They’re brilliant at their craft and completely unprepared for the people side. That gap doesn’t close on its own.”

I watched genuinely talented managers make entirely preventable mistakes — not because they were careless, but because they didn’t know what they didn’t know. A performance conversation that should have been documented wasn’t. A leave request that needed to go through a specific state process got handled informally. A team conflict that had a clear path to resolution sat unaddressed for months because nobody had the vocabulary to name it.

The Dirty Secret of HR Reach

I’ve spent 25+ years in HR across Fortune 100s and startups. One thing is consistent everywhere: there are never enough HRBPs to go around.

At a big company, the ratio might be 1 HRBP to 150 employees — and that HRBP is stretched thin managing executive relationships, strategic projects, investigations, and compliance work. The frontline manager with a tricky comp conversation isn’t getting their call returned the same day.

At a startup, it’s often worse. You might have one generalist covering everything: recruiting, onboarding, benefits, compliance, and the occasional crisis. There is no dedicated HRBP. There’s just a very stressed human trying to keep up.

The Reality

Most managers make their most consequential people decisions without ever speaking to an HR professional. They Google it, ask a friend, or just guess. The variance in outcomes is enormous — and entirely avoidable.

The managers who get good outcomes are usually the ones who either happened to get lucky with their HR partner, had a mentor who’d been through it before, or spent years accumulating the scar tissue that comes from making mistakes. None of those scale.

What Hypergrowth Actually Looks Like

Let me paint you a picture of a typical week at the unicorn. It’s not the polished version you see in the TechCrunch profiles.

On Monday, a manager comes to me because someone on their team got an outside offer and they need to counter. They have no idea what market rate is, they don’t know their company’s comp bands, and they’re terrified of making a wrong move. The conversation takes 90 minutes to get to a decision that should have taken 15.

By Wednesday, there’s a team conflict that’s been brewing for three months. Two senior engineers who can’t be in the same design review anymore. The manager tried to address it informally but made it worse, partly because they used language that (with the benefit of hindsight) wasn’t quite right for the situation.

Friday brings a California CFRA leave request from an employee who may also qualify for FMLA. The manager assumes the laws automatically create separate banks of leave, even though how the protections interact depends on eligibility and the facts and should be reviewed by HR or counsel.

None of these managers were bad at their jobs. They were operating at the edge of their knowledge, without access to real-time expert guidance, in a company moving too fast to build the HR infrastructure that would have helped them.

The Idea Behind Lisa

The frustrating thing wasn’t that these problems existed. It was that an experienced HRBP could often surface the right questions and next steps quickly, including when a situation required case-specific review by HR or counsel. The gap wasn’t always about complexity; it was about access.

What if every manager had access to structured guidance on demand? Not a chatbot that merely recites policy, but a system designed to ask useful discovery questions, use governed sources, coach toward a responsible next step, and recognize when authorized HR or Legal professionals need to take over?

That’s the question that led to Lisa. Not “can we automate HR?” but “can we scale the expertise of a great HRBP to reach every manager who needs it?”

“Lisa isn’t a replacement for HR. She’s the HRBP that HR never had enough headcount to hire.”

The distinction matters. Lisa doesn’t replace the relationship between HR and the business. She amplifies it. Your HR team can still focus on the strategic, the sensitive, and the complex — while every manager who needs guidance on a performance conversation, a comp decision, or a leave request has somewhere to turn.

What We Built (And How We Evaluate It)

Lisa is a deployed evaluation system with governed model routing, specialist consultation, source receipts, deterministic safety rules, and human-review boundaries. Evaluation is part of the product, not a one-time score attached to an earlier version.

Earlier development used structured simulated scenarios and multi-category rubrics. Those totals and scores belong to earlier model, prompt, and tool lineages, so we do not present them as current performance. Current validation is being rebuilt from fresh, blinded evidence with the exact model, prompt, tools, sources, and independent reviews recorded.

Lisa also includes deterministic hard stops for defined high-risk situations. In those cases, ordinary coaching stops and the response directs the manager toward authorized HR or Legal review; configured escalation workflows can support that handoff. Lisa does not provide legal advice, make employment decisions, or replace case-specific professional review.

The Part That Still Gets Me

I’ve been in HR long enough to be skeptical of technology promises. I’ve seen a lot of HR tech that claimed to transform the function and mostly just created more administrative work.

What’s different here isn’t the technology — it’s the problem we’re solving. The manager access gap isn’t only a data or workflow problem. It’s an expertise problem. Lisa is designed to bring more of an experienced practitioner’s questioning, context gathering, and risk awareness into that first conversation.

When I test Lisa against a California harassment scenario, I am looking for specific behavior: does it gather the right context, avoid unsupported conclusions, recognize urgency, and stop ordinary coaching when an authorized human needs to take over? I think about every manager I watched struggle through that kind of situation without support, and the outcomes that could have been different.

That’s why this matters. Not as a productivity tool or a cost-reduction play, but as a genuine attempt to close a gap that’s been causing preventable harm for a long time.


About the Author

Kristina Treytl, MS, is the co-founder of HRBS.ai and the practitioner behind Lisa. With over 25 years of HR experience across Fortune 100 companies and high-growth startups, Kris brings a practitioner’s lens to every aspect of what Lisa says and how she says it. She lives and works in the Bay Area.

Latest from LinkedIn

More from Kris, straight from the source

The Unicorn chapter was just the start. Here’s what Kris has been writing about lately — pulled live from her LinkedIn.

See Lisa in Action

Lisa is deployed for controlled evaluation and pilot configuration. The current environment is approved for synthetic and demonstration data only. Do not enter real employee data or PHI. Real-data pilots require approved provider transport, contractual controls, and customer-specific configuration.

Request a Demo