Isenberg posts the 5 steps forward deployed engineers run
A 53 minute masterclass on work that Isenberg says is basically nowhere online.
Greg Isenberg published a 53 minute masterclass on how forward deployed engineers deploy agents inside Fortune 500 companies, and put the five steps in the post itself. He says some forward deployed engineers are paid over $1M a year, and that almost nobody talks about how they do it in any detail. His framing is blunt about why the thread exists at all. It is, in his words, basically nowhere online.
They start by listening, not building.
They use the cheapest model that gets the job done. Most of this work doesn't need the most expensive AI.
The guest is someone Isenberg says does the work all day long for Fortune 500 companies. The episode is on @startupideaspod, with a credit to @vasuman, and Isenberg describes it as a full course on the topic available free.
The five steps
One, they start by listening instead of building. They interview the people doing the work and let agents quietly read the company's data for a few weeks. Isenberg says the real process is almost always 3x longer than the one written on paper, which is the sort of thing you only find by watching rather than asking.
Two, every step gets sorted into one of four buckets. Delete it, automate it with a simple rule, give it to an AI agent, or keep a human on it. Isenberg says a surprising amount ends up in the first bucket.
Three, they build inside the tools the company already uses, so nobody has to learn a new app. When a human needs to sign off, it is a Slack message.
Four, they use the cheapest model that gets the job done. Isenberg says most of this work does not need the most expensive AI.
Five, they measure before and after, then come back six months later and prove it worked.
The business read
Nothing in that list is about model choice, and the two steps that carry the most weight are the cheapest ones. Deleting a step costs nothing to run and never breaks. Measuring before and after is what turns a deployment into a renewal, which is presumably part of how the fee gets to $1M.
For a solo founder selling agent work into bigger companies, the shape of the offer is the useful part here. The deliverable is not an app. It is a few weeks of listening, a sorted list of steps, some automation living inside Slack and whatever the client already runs, and a number at the end. Isenberg says the episode includes concrete client examples, and pitches it at anyone who wants to be a forward deployed engineer or make their own company AI-native.
This is one practitioner's account relayed by Isenberg, not a survey, so treat the $1M figure and the four-bucket split as what that practitioner reports rather than a measured norm.