Obnovit
← All resources

One Engineer With AI vs a Three Week Proposal Team

24 hours.

That is roughly how much working time it took me to pull together the key elements of a proposal for two related projects to retrofit turbine inlet air chillers into two existing facilities. If that phrase means nothing to you, here is the plain version. You cool the air going into a gas turbine, and the turbine produces more power. The equipment is large and the engineering behind it is dense, which is exactly why proposals for it usually take weeks, not days.

I did it independently, working with a couple different LLMs. Less than 24 hours of actual work, spread across two weeks in the cracks of my normal schedule, at a fraction of my time.

For reference, a team at an EPC firm pulled together the same proposal. It took them well over three weeks, and on the order of 100 to 200 hours of their time.

To be clear about scope, what I put together were the key elements of a proposal, not the finished, submission-ready package. That included the process design, the sizing, and the clarification questions that needed to go back to the owner. All of it took roughly 10 to 25 percent of the normal time and effort. And on a straight schedule basis, if I had sat down and worked through it without interruption, the compression would have landed in about the same range.

One engineer, working part time, matched the core output of a full proposal team. That number is worth sitting with for a second, because it raises an obvious question about what happens to the team.

Growing capacity without growing headcount

The numbers point to why, and the clearest example isn’t even from engineering.

When total stations began replacing transit-and-tape surveying, the productivity of a field crew changed dramatically. Work that once required multiple people and extensive manual measuring could increasingly be completed by a much smaller crew, and eventually, with robotic total stations, by a single surveyor working alone.

But the surveyor did not become less important. The judgment about where to set up, what to measure, how to validate the work, and what the numbers actually meant remained human.

What changed was capacity. A surveying firm could collect more data and complete fieldwork faster, taking on more work with the same team.

The same shift is playing out in proposal work. Capacity grows while the size of the team stays the same.

Many engineering firms I talk to run into the same wall. They carry a backlog of opportunities they cannot chase, because proposal work eats too much time from the resources who are the most constrained, the senior engineers and team members. Every bid you can’t get to or have to no-bid is revenue you never had a chance to win. If one engineer plus AI can produce the key elements of a proposal in a quarter of the hours, the same-size firm gets more at-bats with the team it already has.

The judgment stayed with the engineer

This is the part that matters most, and the part most AI headlines skip entirely.

The AI handled the production work. Research. Structuring. First-pass calculations. Drafting. Formatting. Those are the exact tasks that eat up those 100-plus team hours on a normal proposal, and that part compressed.

The judgment behind it stayed human. I decided what the owner actually needed, and I knew which design conditions were missing and belonged on the clarification list. Deciding which numbers were reasonable and which needed pushback took the same kind of experience I’d expect from any senior hire. [Add a specific example here: one clarification question or number you personally caught and pushed back on, and why it mattered.] Then I checked the whole thing the way I would check a junior engineer’s work, because that’s exactly what it was, fast, capable, tireless, and in need of an experienced set of eyes on every page.

Skip that check, and the same 24 hours produces something that merely looks like a proposal. It cannot be trusted as one without the judgment that went into building it.

The number worth remembering here is 10 to 25 percent, the share of the production work AI can now carry on a proposal like this one, when an engineer knows how to direct the tool and check what comes back. The other 75 to 90 percent of that effort did not disappear. It came back to you, to spend on the work only a human can do, and on the bids you never used to have time to chase.

If you’re wrestling with what this means for your team, DM me and tell me what your proposal process looks like today. I’ll tell you where AI can take real hours out of it, and how much capacity that opens up for the bids you’re not getting to right now.

Know where you stand. Choose your next move.

Get your free AI Score