Your First 90 Days in AI Automation, Week by Week

This assumes evenings and weekends rather than full time, because that is the realistic situation for most people. If you have more hours, the same order applies, compressed.

Weeks 1 to 3: build one thing that works

One platform. Not three. Learn it by building rather than by watching, and pick a fictional business so nothing is blocked on someone else.

The target is one complete system: an enquiry arrives, gets tagged, receives a reply within a minute, is qualified with two or three questions, is offered times, books itself, gets a reminder, and is followed up if it does not show. All of it connected.

Complete matters more than impressive. A plain system that works end to end teaches you more than five half-built clever ones, because the lessons live in the joins.

Weeks 4 to 5: choose who, and go and ask

Pick a type of business using the four traits: steady enquiries, high customer value, appointment-based, busy owner. Weight it heavily toward whoever you can actually reach.

Then contact ten of them and do not pitch. Ask how enquiries reach them and what happens when nobody can pick up. You are gathering the words they use, which becomes your outreach copy later.

This week feels unproductive because nothing is being built. It is the highest-value fortnight in the whole plan. Everything after is aimed at people you now understand.

Weeks 6 to 9: outreach and adaptation, at the same time

This is where most people stop, so name it in advance: these four weeks are mostly silence, and silence is the normal texture of this stage rather than a verdict on you.

Two things run in parallel:

1. 01Adapt your system to the niche. Their language, their stages, their common questions, their objections. This becomes the thing you sell repeatedly.
2. 02Contact businesses consistently, every single week, whether or not anything came back last week. Consistency beats intensity here, and a burst followed by two quiet weeks produces nothing.

Expect most messages to be ignored. Expect most replies to be no. Expect early conversations to go badly while you learn what to say. All three are normal and none of them means the plan is failing.

Weeks 10 to 13: close one, then use it

Take the first client at a price that feels slightly low and over-deliver past the point of comfort. You are buying a case study, a testimonial, referrals and your own belief. Do not work free, because free clients do not implement.

Deliver fast. Onboard them properly so they actually use it in week one, because a client who sees value early stays and a client who does not leaves in month two.

Then, at the moment they say something good, ask for two things: a sentence you can quote, and whether they know anyone else with the same problem. That single conversation is usually worth more than another month of cold outreach.

What to ignore for all 90 days

• A second platform. One, properly, until it is boring.
• A perfect website. Nobody is searching for you yet.
• Logos, business cards, a company name you agonise over.
• Paid ads. You do not yet know what converts, so you would be paying to learn what outreach teaches free.
• Any new tool that launches during these 90 days. It will still be there in six months, better documented.
• Comparing your week 6 to someone else's year 3.

If you reach day 90 with no client

This is common and it is not a reason to quit. Diagnose honestly, because the fix differs completely:

Four of those five are fixed by doing something different, not by learning more. That is the pattern worth internalising: at day 90 the answer is almost never another course.