Key Takeaways

  • Cloudflare scaled its summer internship cohort from 200 to 1,000 people across a 5,000-person company to drive bottom-up AI adoption.
  • Cloudflare CTO Dane Knecht found that managers paired with AI-native interns adopted AI tools significantly faster than peers working without them.
  • When employees thought Cloudflare had rolled out automated internal systems, the automation was actually interns writing scale files by hand with AI.
  • Cloudflare uses personal AI habits as an interview filter, treating candidates who do not experiment with LLMs outside of work as uncurious.

Reverse Mentorship Beats Executive Mandates

Most enterprise software companies run AI rollouts from the top down. An executive team signs an enterprise agreement with an LLM provider, schedules mandatory company-wide training, and wonders why senior engineers still write code the exact same way they did three years ago.

Cloudflare took the opposite path. Two years ago, the company brought in 200 college interns and noticed something distinct about how they worked. As Knecht observed, “those interns I think were really the first fully kind of AI native, where that if you took AI away from them I don't think they would be able to work.”

The interns did not need an executive memo or a best-practices workshop to use LLMs. They simply did not know how to build software without them. More importantly, their habits spread upward. Knecht watched the engineering managers who supervised those interns change their own workflows: “we actually saw the managers that had interns from that summer actually adopt AI faster so then this past summer, this past year, we decided we're going to hire a thousand interns.”

Injecting 1,000 AI-reliant college students into a 5,000-person organization functioned like an operational contagion. Rather than telling senior staff to change, Cloudflare surrounded them with junior team members who solved problems in half the time by default.

The Intern Engine as an Audit Tool

Instead of bringing in management consultants to audit internal workflows and build AI transition roadmaps, Cloudflare used its intern class as human exploratory agents.

The interns found the bottlenecks in company operations by tackling manual tasks directly with LLMs. At one point, internal teams assumed the company had deployed an autonomous software pipeline. Knecht noted: “for a while people thought it was actually all AI and automated. It was actually a bunch of interns that were helping people write scale files. But that let us kind of get to see in a way that's not going out and hiring a consulting firm to come out and catalog every single thing to be done.”

Before you can write code that automates an internal process, you have to map the edge cases. Interns armed with LLMs uncovered those edges in real time. They became the bridge between brittle manual tasks and permanent automated systems.

Screen for Personal Curiosity Over Tool Proficiency

Cloudflare's strategy reveals a direct hiring rule: do not test whether an engineer knows how to use a specific AI tool at an enterprise level. Test whether they are curious enough to play with models in their personal life.

Knecht views an engineer's lack of outside experimentation as an automatic warning sign. “One of the biggest things we've always looked for is curiosity,” Knecht said. “And so if people aren't using AI at home in their personal life and all that, I mean that's a pretty big red flag. How can you not be embracing AI if you have a level of curiosity out there?”

Cloudflare operates at a traffic scale where off-the-shelf software fails, meaning engineers must invent bespoke network architectures. If an applicant lacks the baseline curiosity to mess around with free model endpoints on a weekend, they will not invent novel solutions under pressure.

What to Do With This

Add one screening question to every candidate interview this week: "Show me something unusual you built, tested, or broke with an LLM at home this month." If their answer relies entirely on tasks assigned to them at work, pass on the candidate.