Cloudflare did not start as a distributed company. Before the pandemic, they were primarily office-based and San Francisco-centric. What followed was not a gradual evolution — it was a series of deliberate, sometimes uncomfortable decisions that have compounded into one of the more unusual talent strategies in the technology sector.
At Running Remote 2026, Kelly Russell, Chief People Officer at Cloudflare, gave a candid account of what it has actually taken to build an adaptive talent organisation in a company that spans hub offices across six primary cities, 65% hub-based employees, and a genuinely global workforce.
The hub strategy in practice
Cloudflare’s primary hubs are San Francisco, Austin, London, Lisbon, Bengaluru, and Singapore. Secondary hubs in Seattle, Boston, and Atlanta give employees WeWork-style access without a full office commitment. The shift from 30% hub-based when Kelly joined to 65% today has been intentional and data-driven, tracked through a hub dashboard that monitors workforce distribution, attendance patterns, and the most common days for teams to coordinate presence.
The model gives individual executives discretion on office requirements rather than mandating specific days company-wide. Interns are required three days per week and typically come in five. Other teams self-organise, with Tuesday through Thursday emerging naturally in locations like Austin.
Engagement survey results showed distributed employees slightly more engaged than hub-based employees — a finding that cuts against the prevailing pressure in the technology sector to mandate in-office time. Cloudflare takes that data seriously.
The 1,111 intern programme
The most striking initiative Kelly shared was Cloudflare’s plan to hire 1,111 interns — a number named after their 1.1.1.1 DNS product — representing an 1800% increase from their typical 60-70 Austin-only summer programme. The internship has been expanded internationally, runs year-round across all hubs, and is based on a clear strategic rationale.
The CEO’s reasoning was twofold. First, bring in AI-native early career talent to help the organisation transform from inside — because leadership, however capable, did not grow up in the AI era. Second, get real project work done by digitally native people who see no distinction between AI-assisted and traditional work. One intern presentation prompted the CTO to declare an immediate hire in a way that was accidentally broadcast over an unmuted mic. The programme is producing that quality of work.
AI transformation from the inside out
When Kelly joined, the CEO was sceptical about AI. Within six weeks, he had shifted entirely — and from that point, the expectation has been clear and consistent. Employees who are unwilling to embrace AI, after receiving all available tools and enablement, will likely not be a good fit for the company moving forward. That is not a vague cultural preference. It is an explicit operating expectation.
The People team received dedicated engineer support to identify the manual processes most worth automating and implement solutions. Onboarding — previously 40 steps — is now 75% automated. A week-long hands-on AI workshop in Austin let non-technical employees build their own websites and discover firsthand that the technology was accessible to them. The lesson: experiential learning outperforms any presentation.
Cloudflare scrapped its entire talent development strategy for 2026 and rebuilt it from scratch, focused exclusively on AI enablement for leaders. The reasoning: AI creates bigger leadership challenges than hybrid work, and if leaders do not model AI adoption visibly, their teams will not follow. The rebuild was not just logistical — it was a signal about what the organisation believes matters most right now.
An unexpected dynamic has emerged alongside the transformation: high-performing AI adopters are frustrated with peers who resist adoption. In some cases, lower performers have become more productive with AI while previous high performers have struggled to adapt. That inversion of the performance curve is one of the more interesting challenges the People team is navigating.
Office design as an answer to a real question
One thread in the session that resonated practically: if you are going to ask people to come to an office, the office has to answer the question “why would I come in if I’m on Zoom all day?”
Cloudflare invests in diverse workspace types — conference rooms with couches, soundproof phone booths, sheltered brainstorming areas — as well as social infrastructure: weekly provided lunches, game rooms, location-specific Gchat channels that have become organic community hubs. Off-sites intentionally block lunch time for connection rather than working sessions, sometimes locating lunch on a different floor to create the serendipitous encounters that are genuinely difficult to engineer.
All-hands and kickoff sessions are conducted virtually so everyone sees them simultaneously regardless of location. That commitment to the level playing field has not wavered even as hub presence has grown. It is one of the clearest signals of what the company actually values.
Building for the uniquely human
The session ended on a theme that runs through much of what Cloudflare is doing: as AI handles more of the operational load, the question becomes what humans are uniquely good at. The intern programme is designed to inject that perspective. The AI workshop was designed to remove the fear that prevents it. The talent development strategy rebuild is designed to make leaders capable of asking the question well.
