Tried AI to write all our client proposals. Won 0 out of 7 pitches. Went back to writing them myself. Won 3 of the next 5.
The experiment seemed obvious. I was spending 3-4 hours per proposal. An AI tool could generate a first draft in 15 minutes. Customise for 30 minutes. Total time: under an hour. A 75% time savings on a task I do 8-10 times per month.
Switched to AI-generated proposals in January. Ran it for 7 pitches.
Lost all 7.
The proposals were good. Clean formatting. Clear structure. Comprehensive scope. Professional language. By any objective quality measure, they were better than my handwritten ones.
They were also generic. The AI produced proposals that could have come from any consulting firm in my sector. Correct but undifferentiated. The structure was professional and the voice was nobody's.
My handwritten proposals are messier. They have my specific observations about the client's business woven into the scope. They reference things from the discovery call by name. They contain my opinions, stated directly, about what the client should prioritize and why. The AI couldn't replicate the judgment because the judgment is mine.
Went back to writing proposals myself in March. Won 3 of the next 5.
The 3 clients who signed all mentioned the same thing in different words: the proposal felt like I had already started thinking about their problem before they hired me. That's because I had. The writing process is the thinking process. When I write the proposal, I'm forced to form opinions about the client's situation. When AI writes it, I'm reviewing someone else's thinking that happens to be nobody's.
Now I use AI for the mechanical parts. The formatting. The pricing table. The terms section nobody reads. The narrative section, the part that demonstrates judgment and earns trust, I write every word myself.
The proposal that wins is not the most professional. It's the most personal.