‎I’d probably build around a boring problem that businesses already pay people to solve. AI is great, but the real value is when it takes a repetitive task off someone’s plate completely.

I’d focus on something where AI can actually take action, not just generate text. Reading an issue is useful, but reading it, deciding what needs to happen, and actually doing it is where I think the bigger opportunity is.
‎

Honestly, I’d look at customer support. There’s still a lot of repetitive work around tickets, refunds, order updates, and follow ups that AI could handle while still handing complicated cases to a human.

‎I think the biggest shift is from AI as a feature to AI as the worker. Instead of building another tool that helps someone do a task, I’d look for a task AI can reliably take over from start to finish.
‎

I’d look at the gaps between tools rather than another standalone task. A lot of work gets tedious because someone has to take information from one system, interpret it, update another, then notify someone else. If AI could reliably handle those handoffs while leaving a clear trail of what it did, that seems like a problem worth solving from scratch.

I’d build for a very specific user rather than trying to make something everyone can use. If you can solve one person’s problem really well, you have a much clearer path to expanding later.

‎

Something around sales operations would interest me. There’s so much information spread across emails, calls, CRMs and spreadsheets. An AI that turns all that into actual next steps could be genuinely useful.
‎

I’d start with a problem first and only use AI where it actually makes the solution better. It’s tempting to start with “what can AI do?” but I think asking “what annoying problem keeps happening?” is a better starting point.

‎

A place I would seriously look at and consider is AI that helps small teams remember what’s already happened. Important details get buried in emails, chats, docs, and notes, so people end up repeating work or missing follow ups. Something that quietly keeps track of decisions, promises, and unfinished tasks could be genuinely useful.

‎I’d build something that saves people time but also connects directly to the systems they already use. AI generating an answer is nice, but AI that can update the CRM, create the ticket, send the email, and move the task forward is much more useful.

For me, the interesting opportunities are in the messy workflows businesses still run manually. If a company is copying information between five different tools every day, that feels like a much better AI SaaS opportunity than another writing tool.
‎

If I were launching an AI SaaS from scratch today, I wouldn’t build another wrapper that just generates text or images. I would build what I think of as agentic middleware — a very specific, outcome-driven vertical agent.

I’d probably avoid building another general AI assistant. There are already too many of those. I’d rather pick one specific industry and solve one painful workflow extremely well.
‎

I’d focus less on building “another AI tool” and more on solving one specific problem really well. There are still so many everyday tasks that are repetitive and time-consuming. If AI can make one of those tasks significantly easier without adding more complexity, that’s where I’d start.

If I was launching an AI SaaS today, I’d avoid generic copilots or wrappers that just summarize or write emails.

It’s not SaaS with an AI feature anymore, it’s AI-first products that own end-to-end execution.

The real gap now isn’t content – it’s trust and execution. The stickiest problems in 2026 are where AI stops suggesting and starts actually doing the workflow.