Start with the workflow, not the AI
Most AI projects stall because they start from the technology and look for a problem. The ones that pay off start from a task that is repetitive, text-heavy and done many times a week: sorting leads, answering the same questions, copying data between tools or writing first drafts.
That is why this tool asks for one workflow. The more concrete you are about who does what and where time is lost, the more specific the ideas you get back.
How the ideas are ranked
Impact is how much time, money or revenue an idea could affect. Effort runs from S, a no-code automation you could set up yourself, to L, a custom integration with your own systems.
Start with high impact and low effort. A quick win builds trust in AI inside the team and makes the bigger projects easier to justify.
Keep a human in the loop
AI is good at drafts, sorting and summaries. It is not good at judgement calls with real consequences, such as pricing, refunds or anything legal. Every result marks what should stay with a person, at least until the automation has proven itself.