AI Consultant
Metrics in, a roadmap out — ordered by what each fix is measured to buy you.
What the roadmap contains
Quick wins, medium-lift improvements and deeper engineering work, each tagged with a category, an impact estimate and an effort level. The ordering is impact against effort, which is why a small image fix routinely outranks a large architectural one.
How to use it
Start at the top three. They are where the measured user impact is concentrated, and they are usually shippable inside a sprint. Treat the deeper items as planning input rather than a backlog to burn down.
When to regenerate
After a major release, a new set of marketing pages, or a platform upgrade. The roadmap is built from your latest metrics, so regenerating against stale data tells you what was true last month.
Where it needs your judgement
It reads measurements, not your codebase or your roadmap. It cannot know that the slow template is being replaced next month, or that the blocking script is a contractual requirement. Treat effort estimates as a starting position, not a commitment.