Find out where AI can actually save you time

You are doing a time audit of my work week to find where AI could genuinely give me hours back. Be honest in both directions: I need to know what AI can really take over today, and what it cannot, so I do not waste a weekend automating the wrong thing. Work in four phases.

PHASE 1: MAP MY WEEK.
Interview me one short question at a time:
1. What my business is and my role in it day to day.
2. Walk me through yesterday, or my last normal workday: what I actually did, in order, roughly how long each thing took.
3. What eats time every single week that has nothing to do with why I started this business.
4. What I do after hours or on weekends because there is no time during the day.
5. The task I most dread, and the one I keep putting off.
If my answers are broad ("paperwork"), dig once: which paperwork, for whom, how often.

PHASE 2: BUILD THE TASK LIST.
From my answers, list every distinct recurring task you heard, with my rough hours per week on each. Show me the list and ask what I missed. Do not score anything yet.

PHASE 3: SCORE HONESTLY.
For each task, assess what today's AI can really do, using three honest buckets:
- Hand it over: AI does 80 percent or more with light review. Typical: drafting routine messages, summarizing documents, turning notes into structured output, first-pass research.
- Split it: AI does the heavy lifting, I stay in the loop for judgment, relationships, or accuracy stakes.
- Keep it: AI should not own this today, and one honest line on why: it needs my hands, my license, my relationships, or the error cost is too high.
Weigh hours saved against my real week, not a hypothetical one. A task I do for 20 minutes a month does not make the top five however automatable it is.

PHASE 4: THE TOP FIVE.
Deliver:
1. My top five, ranked by realistic hours back per week: the task, the bucket, the estimated time back, and exactly what the AI-assisted version of the task looks like, step by step, so I can picture Tuesday.
2. The honest "keep doing these by hand" list, so I trust the rest.
3. The one to start with this week: the best mix of hours saved and easy setup, and the very first step, small enough to do today.

Rules:
- Never inflate. If the realistic saving is one hour a week, say one hour.
- Every estimate ties back to the hours I gave you in the interview, and the math shows.
- If my week has a bigger problem than AI can fix (underpricing, a role I should hire for), say that plainly in one line. I asked for honesty, not upsell.