Services: The New Software (6 months later)

"The next $1T company will be a software company masquerading as a services firm."

We've been building, selling, and running an AI-native services startup in the life sciences industry for the past 6 months.

When we raised our latest round, the story investors wanted to hear was how we'd rip the expert out of the loop entirely.

That assumption broadly matched the thesis behind Sequoia's article “Services: The New Software” published earlier this year.

Sequoia's @julienbek argued that, over time, today's judgement will become tomorrow's intelligence: “AI autopilots” will slowly absorb the work that still requires human judgement today.

But they've missed one fundamental nuance on the other side of judgement: accountability.

When a service is done right, it's great, but when it's done wrong, it is often disastrous, with real-world consequences: a bad regulatory filing to the FDA can have an opportunity cost of $1 billion.

No one would ever trust anything that has nothing to lose, and AI agents have nothing to lose.

Sequoia calls these agents autopilots; it's the right metaphor - autopilots have been flying planes for decades.

But there's still a pilot in every cockpit.

Not because the autopilots can't fly, but because the pilot has a license to lose, and a seat on the plane.

YC's @charliewarren defined the category's existential risk as variance in the work itself: software ships the same product every time, services don't, and they get judged on their worst delivery.

Accountability is how you cap variance. The pilot isn't there for the 10,000 normal flights; he's there for the abnormal one.

Clients choose us for the same reason: whatever the AI automates, qualified in-house experts own the outcome and intervene where needed.

Clients don't care about the tool, they just want someone to be accountable when it matters, and that has nothing to do with the technology.

Sell the capabilities of an agent behind the accountability of a credible human, and, voilà, you have yourself a scalable AI-native services company.

How we are reshaping an entire industry

Panacea helps biotechs and medtechs navigate the FDA faster, better, and at a lower cost.

Traditionally, regulatory work was outsourced to hourly consulting firms, who, by definition, earn more revenue from prolonged timelines.

Because of this, the entire regulatory consulting space in the life sciences industry has a very low NPS score.

We built an internal AI platform that collapsed the time it took to deliver this regulatory work, and then replaced traditional hourly billing with a fixed milestone-based pricing structure.

We aligned our incentives with our clients: the faster we work, the bigger our margins get, the happier our clients are.

We signed $500,000 in contracts within the first two months.

90% of our seven-figure pipeline is inbound.

The technical upside of AI-native services

On the platform

There is something beautiful about owning your users' environment end-to-end.

Our platform's users are our employees.

They don't file feature requests and wait months; they're sitting right next to us every day.

Every time they intervene with our platform's outputs, we make sure that they never have to intervene again.

We can make breaking changes, redesign workflows, replace systems, and deploy aggressive experiments without worrying about general-purpose UX, backwards compatibility, feature discoverability, etc.

This gives our engineers the technical freedom to attack the problem directly without maintaining a generalized customer-facing product.

On the AI budget

Our contracts are in the six- to seven-figure range.

Unlike SaaS platforms, our AI budget isn't governed by software unit economics.

We can throw money at the inference and infrastructure needed to solve the client's problems without sacrificing quality or speed, simply because the contract values allow us to.

Why should VCs care if humans are never going away?

A $1B SaaS startup with one engineer.

A $1B AI-native services startup with one expert.

Same underlying bet.

We brought the expert in-house, centralized their work into a single platform, automated every repeatable manual task, and turned every intervention into feedback that improved our specialized agents and reduced the expert's involvement over time.

We built an internal operating system for regulatory consulting, and sold it as a service for 10x what we could have charged for the software alone.

With delivery automated end-to-end, the expert moved upstream from executing the work to acquiring and converting customers.

The trillion-dollar opportunity doesn't require replacing experts. It only requires expert labor to stop scaling with revenue.