Stacy L. Chin warns that writing SBIR proposals entirely with AI like ChatGPT gets you rejected. Reviewers want human logic, a tight narrative linking problem to innovation to execution, and proof you can do the work. Use AI only for outlining and early drafts, then rewrite with your expertise to tighten the story and connect every piece back to your goals.
Do you think AI tools can simply just write your SBIR proposal for you? I've seen so many companies just solely rely on ChatGPT or Claude but only have reviewers completely dismiss their application. So I love my AI tools, don't get me wrong, and they can help. If you only rely on them, you're just not going to win. And that's because reviewers are not funding generic writing. They want logic. They want the human component. They are looking for clarity and persuasion from you as a PI. Because sometimes the AI can sound way too polished. It just lacks compelling narrative that connects the problem to the innovation and to the R&D approach. So if you want to use AI tools in your SBIR applications, here's how you do it properly. Only use AI for structure and early drafts, and then take it to refine it by adding your expertise. That means tighten the story, make the logic airtight, and tie back every component to the proposal goals. Because if your proposal looks generic, reviewers can call it out immediately, and then you just become another generic proposal that was rejected in the pile. Because in the SBIR world, strong proposals isn't just about good writing, it's about making reviewers believe that you can do it and you should get awarded. And that's the advice I tell my founders when I ask them,
