← Braintrust
Braintrust role-play cheat sheet
For the mock cold call at the end of most SDR interviews.
What they sell, in one line
Software that shows AI teams how their AI agents behave in real use and helps them test and improve them.
Who you'd be calling
AI engineers, heads of AI or ML, and engineering leaders at companies building AI products, from startups to enterprises.
Your opener
“Hi [Name], this is [Your Name] from Braintrust. I'm calling because we help teams building AI agents see what's going wrong in production and fix it faster. Do you have a minute to tell me how your team checks quality today?”
Problems you solve for them
- Hard to see why an AI agent gave a bad answer
- Quality testing of AI features is manual and inconsistent
- Problems in production found only after customers complain
Discovery questions to ask
- Which AI features or agents are you running in production?
- How do you find out when an agent gives a bad answer?
- How do you test changes before they go live?
Objections they'll throw at you
“I'm busy right now.”
→ Understood. Could I send a calendar invite for a 10-minute chat later this week?
“We're not interested.”
→ Fair. Are your agents performing well enough that you don't need extra visibility, or is it just not the right time?
“We already use something for that.”
→ Good, so you take quality seriously. What does it show you today, and what questions about your agents can't you answer yet?
“Just send me an email.”
→ Happy to. What's the main thing you'd want to know about your agents so I can point you to the right info?
Competitors worth knowing
LangSmith (LangChain), Arize, Langfuse
Interview tips for this company
- Learn what an AI agent, a trace, and an evaluation (eval) mean in simple terms before the interview.
- Your buyers are engineers, so be honest, brief, and curious. Don't pretend to know their stack, ask about it.
Written from public information about Braintrust to help you practice. Ask an insider what their role play looked like; it's the best prep there is.