Mateusz Mucha is CEO of Omni Calculator, a platform helping 15M+ monthly users make better decisions through expert-reviewed calculators.getty​The automation question is the intuitive one: find a manual task and see whether AI can do it faster or cheaper. It's not a bad place to start, but it stops at optimizing what you already do, and I think there's a more valuable question to ask.Why AI Changes Which Projects Are Worth StartingAI reduces cost and time, which can produce two outcomes: an existing activity becomes cheaper, or an activity that wasn't worth starting before becomes viable. The mechanism is the same, but when the cost reduction is large enough, it changes which projects make sense, not just how efficiently you run the ones you already have.Here's an example: Earlier this year, we launched an omni calculator app. I built the first version myself using Claude Code in under a month, despite not having touched code in over a decade.It wasn't that Omni had suddenly acquired a technical capability we didn't have before. We had developers, and a mobile app had been on the list for years. The problem was that building one would've taken more time and resources than we were willing to commit, so we didn't.What changed was the economics. With AI-assisted coding tools, I could move quickly enough that the project finally made sense because the cost of execution dropped far enough to turn something we'd shelved into something we could ship.This second category, capability expansion, is harder to pitch internally, as automation projects have a cost-saving baseline to point to and capability expansion doesn't.How To Look For These OpportunitiesThe best signals come from your team's shelved projects and your customers' unanswered requests. Below are four questions to guide you:1. What were we not doing that we can now start?A good starting point is to go back through projects your team rejected in the past few years because the cost or time couldn't be justified. Not the ones cut for strategic reasons, but the ones the team agreed were worth doing, with cost the only thing in the way.Those are the projects worth revisiting. The question to ask yourself isn't whether AI makes them possible now, but whether AI has lowered the cost enough to cross the threshold you set when you turned them down.2. What are customers asking for that we've said no to?Customer requests are a different kind of signal, as the demand has already been validated. The question is whether you declined because it wasn't the right product decision, or because the cost of delivering it couldn't be justified.Go back through feature requests, service requests or product asks that were turned down due to cost or time constraints. Those are worth revisiting for the same reason as shelved internal projects: the economics may have changed even if the ask hasn't.3. Was the blocker economics?Not every shelved project is a candidate for capability expansion. Some might have been turned down because the strategic fit wasn't right, the market wasn't ready or the team had more pressing priorities. Cheaper execution doesn't change any of that.Here's the test I'd apply: If the cost and time had been half what you estimated when you rejected the project, would you have done it? If yes, it's worth revisiting. If the answer is still no, the blocker was something other than economics, and that constraint probably hasn't changed.4. Do we have what it takes to do it well?While AI changes the cost of execution, it doesn't provide judgment, domain knowledge or the ability to tell good from mediocre.When Omni launched a mobile app, I vibe-coded the first version myself, but the honest version of that story is more complicated. After the initial build, our development team found security vulnerabilities AI had introduced, and we fixed them before the launch.Furthermore, the app isn't just code I wrote with AI assistance. It's built on 12 years of accumulated understanding about what people need help calculating, and the judgment of dozens of developers, domain experts and researchers.That institutional knowledge determined which calculators to include, how to structure them and what a useful result looks like for someone with a specific problem.ConclusionFinding manual tasks and making them cheaper with AI is a legitimate place to start. But the more valuable version of that question is about which projects you can now start, not just which processes you can speed up.The biggest AI opportunity for your organization might not look like an AI project at all, but rather something your team decided not to build three years ago.​Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?