Thomas Baucom’s calendar perpetually looks like “a bad game of Tetris.” Working as a forward deployed engineer (FDE) at productivity platform Superhuman, he’s often on back-to-back calls talking to customers across industries, hearing out their problems and requests. Sometimes he’s jumping on a plane to be by their side.
Previously a customer success manager, he stepped into his new role two months ago. Baucom said the job is similar in that it’s also focused on customer insights. As a customer success manager, he was also talking all day, every day with customers about how they wanted to use the products. The key difference is now he can spring into action on what he hears.
“In that role before, I had to say, ‘well, we're working on that,’ or ‘we can put that in the backlog, but it doesn't work like that today,’” he said. “The big difference in the FDE role is I can now say, ‘I can fix that.’ I have the ability to work in our codebase, go on-site with you and build those things out. I can go much much deeper.”
That more direct line from insight to action is why companies of all stages and sizes are embracing the FDE model, which revolves around embedding directly into customer organizations to help them adopt the product, solve specific problems and scale usage.
Job listings for FDE roles increased more than 729% between April 2025 and April 2026, according to data from Indeed shared with Business Insider. Amazon in June announced a $1 billion investment toward a new FDE unit, and just two days later, Microsoft announced its own $2.5 billion FDE initiative. Anthropic, OpenAI and Salesforce have also all announced similar initiatives earlier in 2026.
Now that some adopters are well into their FDE journey, they’re starting to see the outcomes — in particular, rapid, iterative speed to deployment. They’re also learning some lessons around how to leverage the FDEs effectively.
Need for speed
Superhuman started embracing the FDE model around four months ago, and Baucom said he’s already shipped around 20 customer projects. He’s seen first-hand that speed is the most undeniable benefit, and others agree.
“What was taking months is taking weeks,” said Mike Park, CIO at Infor, which has done a big push into FDE engineering starting about a year ago.
Kevin Wu, co-founder and CEO at Leaping AI, a voice AI startup that hired its first FDE last year, also cited speed as the biggest benefit and said this is particularly true when they go on-site.
Kevin Wu with two recent FDE hires. Credits: Leaping AI
“Flying to the customer actually helps because you get everybody in one building, and then you just finalize the last few steps on day one,” he said. On day two, they train the customer on how to use it and monitor the performance to see what can be immediately improved. After that, they gather feedback, integrate it, and have weekly virtual check-ins from there on out.
Beyond speed, FDE teams have noticed that customers appreciate the hands-on approach.
For one example, an Infor FDE worked with a customer in industrial manufacturing to build a system that flags itself when it needs attention, so the customer no longer has to watch the dashboards constantly. The challenge with a use case like this, Park explained, is that companies have different levels of comfort trusting AI. The FDE worked with them directly to understand their tolerance and tailor the solution to their exact stage of AI adoption, and help them feel more comfortable along the way, he said.
Mike Park from Infor. Credits: Infor
For a recent project, Baucom transformed a multinational publisher’s 400-page style guide into a series of custom agents, with one for each of its many sub brands. As a writer works, the agent reads the text and flags any deviations from the relevant style guide. It underlines errors for the writer to review, shows the correction and points to the corresponding entry in the style guide. The customer got a fast turnaround and a bespoke solution built around keeping its brand consistent.
“We're going to cater this solution to you. We're going to make it easy. We're still going to have the human element there, and that seems to really resonate at least with folks I get on the phone with.”
The part that isn't automated
Overall, the main lesson Baucom has learned so far about being an FDE is that AI is still intimidating for a lot of folks, which makes a strong human touch more valuable, not less.
“People still want to have a Thomas they can Slack,” he said.
While FDE is typically thought of as a customer-facing role, companies have also learned the benefits of the model can be leveraged internally. Baucom said he’s built prototypes for tools for his teammates to be able to run better demos and agents to help prep for calls, for example.
For Park, another lesson from Infor’s FDE work so far is how crucial it is to stay up to speed. As models progress, teams need to keep evolving and iterating quickly, he said. For this reason, he believes the fast pace of AI itself is a driver for the FDE model.
“With this ability to iterate faster, I think [the FDE model] allows you to take advantage of the technology even faster than you would have before because of this concept,” he said. “And I think that's where it’s very powerful.”
Top photo: Thomas Baucom, Credits: Superhuman
