Cloudflare Builds Coaching Foundation and Cuts Call Admin by 40%

Gustav Bergman joined as Cloudflare AI's first GTM hire, responsible for building a commercial motion from scratch. With a technical buyer base that expects precision and a small team expected to grow to 20 to 30 people, getting the foundation right mattered, a discipline that carried through to the team Cloudflare ultimately acquired.

‍

The Challenge

When Gustav joined as Cloudflare AI's first sales hire, there was no infrastructure for recording, reviewing, or learning from customer conversations. Meetings were happening on Zoom with developer prospects and technical evaluators, but nothing was being captured in a way that fed back into deal management or coaching.

‍

Call notes went into Attio manually after each conversation. That process took 15 to 20 minutes per call, created inconsistencies in how deals were logged, and depended entirely on what Gustav could remember by the time he sat down to write it up. There was no structured view of which objections were coming up repeatedly, which talk tracks were landing, or which deals were at risk because of a poorly handled technical question.

‍

The absence of structure compounded over time. Without recorded conversations, there was no feedback loop for improving how calls were run. Without reliable CRM data, there was no clean pipeline signal. And with a team expansion on the horizon, there was no coaching framework to onboard new reps against. Every new hire would have to start from scratch if nothing was built now.

‍

The Solution

‍

Gustav evaluated Outdoo to solve three problems at once: reduce the manual overhead of post-call work, build a review system that would improve how he ran technical sales conversations, and create a coaching infrastructure that would still be useful when the team grew.

‍

Practicing technical discovery against realistic developer buyer personas

Cloudflare AI's prospects are developers and engineering-led product teams. They come into sales conversations with detailed knowledge of the infrastructure space, precise questions about model performance and pricing, and a low tolerance for vague answers. A poorly prepared discovery call with this buyer closes doors fast.

‍

Gustav used Outdoo's AI roleplay to practice discovery conversations against developer personas that reflected Cloudflare AI's real buyer profiles. Objections around model latency, API reliability, cost at scale, and comparisons to hosting models directly came up repeatedly in live calls. Working through those objections in practice sessions before a live conversation meant he could engage more confidently and spend less of the actual call recovering from an unexpected challenge.

‍

“The technical objections in developer sales are specific and they come early. Being able to practice them against something that pushes back the way a real engineer does changes how you go into the actual conversation. ”

Gustav Bergman GTM, Cloudflare AI Infra (earlier Replicate)

‍

Building structured preparation for expansion and renewal conversations

As pipeline grew, the conversations shifted. Customers who had already adopted the platform came back with architecture questions, usage-based pricing discussions, and requests about enterprise features. These conversations required a different kind of preparation than initial discovery.

‍

Gustav used post-sales roleplay simulations to prepare for these interactions: expansion calls with technical decision-makers who already understood the product, renewal discussions where pricing and contract terms were in play, and escalations where a deployment issue had created friction. Practicing those scenarios ahead of time meant the call had a plan going in rather than an improvised response to wherever the buyer took it.

‍

“Developer buyers who already use the product are harder to impress the second time around. They have specific expectations and specific feedback. Preparing for those conversations with structured practice makes the difference between a confident expansion call and a defensive one.”

Gustav Bergman GTM, Cloudflare AI Infra (earlier Replicate)

‍

AI coaching that scores every call and builds a repeatable performance baseline

Outdoo's AI coaching layer scored each recorded conversation against defined criteria: discovery depth, objection handling, next-steps clarity, messaging consistency, and technical fluency. Rather than relying on memory or occasional self-review, Gustav had a systematic record of where calls were strong and where they were falling short.

‍

The scoring data served a second purpose. By the time Cloudflare AI was ready to bring more people into the sales team, there was already a library of scored calls, a defined set of coaching parameters, and a set of roleplay scenarios grounded in real conversations. New reps would not be starting from a blank slate. They would have a structured onboarding path built on actual performance data from calls that had already happened.

‍

“Building the coaching data now means the first hire I bring in has something to learn from. I am not inventing a process when the team grows. The foundation is already there.”

Gustav Bergman GTM, Cloudflare AI Infra (earlier Replicate)

‍

The Impact

Outdoo gave Cloudflare AI's first sales hire a way to run a more rigorous sales process without adding operational overhead. The combination of automated call summaries, CRM sync with Attio, and AI scoring replaced hours of manual work each week and created a structured record of performance that would carry forward as the team expanded.

‍

Call admin time reduced by 40% across all customer meetings

Automated summaries and CRM sync with Attio eliminated the manual note-logging that previously followed every customer call. What took 15 to 20 minutes per conversation was handled automatically, and the output was more consistent than manual notes. Time recovered from admin went directly into pipeline development and deal preparation.

‍

Objection handling consistency improved through structured practice

Technical objections that previously surfaced unpredictably in live calls became predictable through roleplay. Gustav could rehearse the specific challenges that developer buyers raised around performance, pricing, and architecture before those conversations happened. That preparation showed up in live calls as more direct, confident responses that kept conversations on track instead of shifting into damage control.

‍

A coaching baseline built before the first team expansion

Scored call data, roleplay libraries, and defined coaching criteria were in place before Cloudflare AI added its first sales hire. That infrastructure meant new reps would complete structured practice against realistic buyer personas before speaking to actual prospects. The gap that typically exists between a new hire's first week and their first well-run discovery call would be narrower because the coaching system had been built and validated in advance.

‍

‍