How To Pick The Right AI Company To Bet On & Operationalize Champion Building w/ Bryan Cox

Bryan Cox is the VP of Worldwide Sales at Braintrust, the eval and observability platform that builders and frontier model providers alike rely on to ship AI agents.

Before Braintrust, Bryan spent nearly five years at Grafana, joining at Series A/B and helping build one of the most respected go-to-market motions in developer tooling.

Earlier in his career he cut his teeth selling complex cloud and data-center infrastructure, including a stint at Flexera, learning what large strategic "whale-hunting" deals look like in action.

A former college tennis player and self-described history major who fell into sales, Bryan treats the work like the craft it is, and is moving his family from Santa Barbara to the Bay Area to get closer to the action.

Bryan was in accelerators at Grafana. Seven-figure deal, hundreds of thousands in commission on the line. All they needed was one click from someone outside the deal, and she said no.

Then his champion called: "I honestly could not have predicted this. But you know what? I just don't need this. I'm about to retire."

Multiple POCs, a leadership trip to a Vancouver steakhouse, months of work. Gone. The lesson: "Your time as a seller is valuable, and the people you attach to are even more valuable." Most reps are so desperate to have a champion that they glom onto whoever's talking to them, even when that person can't get the deal done.

Bryan's take is that most orgs treat champion building as magic. "There's like five reps at a given company that are just artistic about it, and they have no idea why."

Braintrust is making it boring on purpose: champion-building spreadsheets in their QBRs, mapping and testing people across personas, and literal plays for things like getting a buyer onto a cell phone. The formula on every call: have a reason for reaching out, teach them something, give them something, and communicate about risk. All of it leads back to trust.

And with 2-3 month AI sales cycles, there's barely time to build one champion, let alone several. Which is exactly why it has to be operationalized.

Before saying yes to Braintrust, Bryan did 40 calls: investors who knew the company, investors who didn't, past champions and execs, current customers, and people in the space he respected. His trick: ask the same question over and over so you can actually form an opinion, the same way you'd interview an org to find pain in a deal.

He also had, in his words, "long romantic conversations with Claude and ChatGPT" about the space. Dumb questions, why a CTO should care, who else could build this.

His signal checklist for evaluating an AI company: durable TAM you can explain in one breath, who's buying it (all the builders buying Braintrust was the tell), whether the NDR proves the sales cycle is working, and whether it can jump from digital native to enterprise. "If you can't easily jump from digital native to enterprise, your company can only go so far."

A spicy one from the guy selling AI infrastructure: "I see a little bit too much DIY AI, both in the industry and on my team."

His view is that it's rev ops' job to strip out the busywork, not each rep's job to duct-tape their own agent stack. "The reps that are interested in building their own agents are maybe in the wrong role." His best reps lean on simple tools like Notion AI, don't overcomplicate it, and some workflows are still just a spreadsheet that's worked for years.

Everything is good in its right place and in moderation. The job is selling.

Bryan's version of enablement for 20 years: interview the top six reps, put it in a deck, assign a video and a quiz, clap. "Reading books about how to play soccer, then getting thrown on the field."

His replacement: voice agents built on their own ICP, evaluated in Braintrust itself, so reps rehearse new business meetings and feature pitches before the stakes are real. He practices with them himself.

The inspiration came from one of their investors, a famous basketball player, telling the Kobe story at an all-hands: an hour of drills with no ball, 30 minutes dribbling against a wall with each hand. The greats practice. Most reps, new and experienced, never do. And the experienced ones are the worst offenders, because they think they're already great at everything.

(00:00) Cold open: the one-click "no" that exploded a seven-figure deal
(01:22) Welcome to the Crew podcast + sponsors
(02:52) From early cloud computing to Flexera's whale-hunting deals
(06:02) Why Grafana produced so much elite sales talent
(07:23) Choosing Braintrust: going where the highest-value problem is
(09:34) Job as craft: loving the technical, hard-to-understand sale
(13:31) What separates great reps from good ones on enablement
(14:35) Sales as an ultramarathon: balance, burnout, and your "why"
(22:05) The motion offense: circling a deal instead of running waterfall
(25:08) The seven-figure deal that died, and operationalizing champions
(29:41) How to interview a company: TAM, durability, and who's buying
(33:10) The 40-call diligence process before joining Braintrust
(36:50) Why everyone's moving to voice agents for ramp and enablement
(42:25) The AI tool stack and a hot take on too much DIY
(44:21) Rapid fire: GTM teams he admires and what he tests for in hiring
(46:47) "Why are you in sales?" and the Shrek/onion test
(50:44) The most creative way to break into Braintrust (hint: it's not through Bryan)