Why AI-native startups need half the headcount

PLUS: a Google DeepMind exec on when to bet on AI tools and when to wait.

From the aibl team

How far are we from having personal AIs that do our shopping for us, or our financial management? Probably further away than a startup founder would say in their pitch deck, but sooner than our businesses will be prepared for. This will be one of those changes that perpetually lies on the horizon, loses steam in the press, then suddenly crashes over our marketing business models.

It’s one of the intriguing concepts that came up in our interview with Rupert Kemp, who has one of the coolest titles you’ll come across: Director of Strategy and Special Projects, Office of the CEO at Google DeepMind.

In our talk we cover the challenges specific to the mid-market and how to balance waiting with moving too quickly, among other questions.

Then we encourage you to keep reading and learn about the fascinating working paper from HBS scholars on their research into the differences between true AI-native startups and their more traditional peers.

Here’s a taste. The organisational differences they identified trace almost entirely to whether AI is integrated into products, not from simply giving the workforce access to AI tools. About two-thirds of AI-native startups embed AI directly into their products, and they benefit from smaller, flatter organisations that generate more revenue per head. When the authors measured internal tool use through job postings, it did not predict these same benefits.

Q&A with Rupert Kemp, Director of Strategy and Special Projects, Office of the CEO, Google DeepMind

Rupert Kemp joined Google DeepMind in 2017 after completing his MBA at London Business School. Before that, he spent eight years at a brand and intangible asset valuation consultancy, leading projects for major global brands and helping launch its Africa office. At DeepMind, he works on strategy and special projects within the CEO’s office, with a front-row seat to where the technology is heading before most organisations have clocked it.

What capabilities should mid-market business leaders be watching right now?

“Agent systems are going to be incredibly powerful, both personally and professionally. Companies that can harness the power of AI systems enabling their employees to be more effective and efficient are likely going to be able to capture more value or be more efficient with how they go about it.”

The technology moves fast. How should a mid-market business think about when to commit to a tool or approach, and when to hold off?

“The technology is moving incredibly fast, making long-term bets on specific applications risky. Instead, I would invest in foundational solutions, such as robust data infrastructure, flexible API layers, and unified cloud environments, that provide a stable base for multiple tools. This modular approach ensures you remain nimble and agile, allowing you to swap out individual AI models or providers as superior versions emerge. Agility is the primary competitive advantage in this landscape; by avoiding vendor lock-in and focusing on interoperable systems, businesses can integrate cutting-edge breakthroughs in months rather than years.”

You built yourself a personalised training app using Gemini in a couple of hours. What’s the equivalent starting point for a non-technical business leader?

“Something super powerful and incredibly easy is creating infographics from text. Go to Gemini, drop in a bunch of text or docs, ask it to simply represent the content as an infographic. It’s a pretty powerful way to convey a lot of information graphically, rather than hitting people with a wall of text.”

“I also find Gemini in Google Slides is incredibly useful as a co-creator of slides. You give it a range of your content and it will draft something up for you to review and edit from there.”

aibl Research: What Can We Learn From AI-Native Firms?

We spent some time with the new HBS working paper by Hyunjin Kim (INSEAD) and Rembrand Koning, because it offers the clearest picture so far of how firms built around AI are organised. The essence is that the AI natives do look structurally different, and the differences point to where AI value comes from.

The authors studied Y Combinator startups from 2020 to 2024, plus the wider universe of US venture-backed firms, linking each to workforce data on team size, seniority, and hierarchy. Then they compared those with the normies and identified four patterns.

They are smaller. AI-native firms run roughly 25% leaner than comparable non-AI startups. Three years after founding, they employ about half as many people. Yet they raise similar funding and hit comparable valuations, meaning they generate far more value per head.

Product spotlight of the week

This week the aibl team has been tracking Sage People, the Salesforce-built HR platform Sage picked up from UK firm Fairsail back in 2017. It’s built for global or multi-site mid-market businesses, 200 to 5,000 employees, running HR across several countries at once.

What it solves is fragmentation. Core HR, leave, performance, and talent all sit in one Salesforce-native system, with statutory support built in across dozens of countries, rather than patched together from local tools.

Last year Sage added Workforce Intelligence, an analytics layer that pushes HR teams past basic headcount reporting into proper workforce planning.

Pricing is bespoke, quoted per organisation based on headcount and configuration, so it’s one to scope with Sage directly rather than self-serve sign-up.

Copyright © 2026 aibl. All rights reserved.
Office 318, X Arding & Hobbs Building, 7 St. John’s Road, London SW11 1QN, United Kingdom