Inside the FDE Team: How an Engagement Runs
The engagement process for OpenAI FDEs involves navigating complex interactions with various teams while managing customer expectations. Effective communication is crucial; messages must clearly outline problems, consequences, evidence, requests, and follow-through to avoid ambiguity. FDEs must identify decision-makers and ensure that each phase of engagement is backed by specific evidence. Observational visits to customers can yield valuable insights that enhance collaboration and decision-making. Ultimately, successful FDEs demonstrate ownership of outcomes by articulating clear responsibilities and tracking progress meticulously.
We'll cover the following...
Cedar’s two-week trial concluded on Friday. The following Monday morning, the FDE shared this update in the engagement channel:
“Product, Research, Security, GRC, GTM: Cedar is blocked on several issues from the trial. Please fix before the HR desk rollout next week. I’ve told the sponsor we’re on track.”
By Thursday, nothing had moved. Everyone had read the message. Nobody had agreed to do anything, because the message never said what anyone was being asked to do.
An OpenAI FDE owns the outcome of an engagement, but almost none of the people who can unblock it report to them. Customer engineers control the customer’s systems. Domain specialists decide what a correct answer looks like. Several OpenAI teams own pieces the FDE needs. Interviewers probe this directly, often with some version of: how do you move a blocked deployment through people you don’t manage?
Who does an OpenAI FDE work with?
OpenAI’s posting says FDEs work closely with Product, Research, Partnerships, GRC, Security, and GTM teams. It does not publish reporting lines or say that every team joins every engagement, so the useful skill is knowing what to bring to each one.
Team | What an FDE Typically Brings Them |
Research | A model behavior that reproduces outside the customer’s integration |
Product | A recurring workflow need that might point to an OpenAI product surface |
Security | A security concern that needs assessment before a connection is used |
GRC (governance, risk, and compliance) | Assurance questions from the customer, and who can accept the answer |
Partnerships | Coordination on the customer relationship beyond the engineering team |
GTM (go-to-market) | How to describe scope and readiness so commercial expectations match what was proven |
Customer teams retain responsibility for their systems, data, and business decisions throughout the engagement. At Cedar, the IT team manages the ticketing system, the knowledge team maintains the FAQ, and service desk leads define the criteria for acceptable responses. Without clear ownership boundaries, an FDE may escalate issues to OpenAI that require decisions or changes from the customer's teams.
The teams listed in this table are examples of possible owners, not confirmed assignments. At the start of an engagement, confirm who has decision-making authority, who implements changes, and who handles operational issues.
Why did the broadcast message fail?
Before reading the diagnosis, review the Monday message and identify every phrase that fails to specify a clear action ...
The Monday ...