Guide
Your AI agent's first week: a day-by-day setup
Most new AI agents stall in week one because they get a chat, not a job. This is the day-by-day plan we'd copy: pick one repeating job, run it read-only, move to drafts you approve, then review on day 7 before adding a second job. It works with Grok Bot, ChatGPT, Claude or an n8n flow. Which job to pick first is covered in What should my AI agent do first?; how to grow it into a team is in How to set up an AI team. Time figures here are estimates, not measurements.
What to hand to an agent first
Day 1 — Pick one job and write it down
One sentence: what it does, when, and where the result goes. Example: 'Every weekday at 8:00, send me today's calendar, emails that need a reply, and anything overdue.' A morning brief is the safest first job because it only reads.
Day 2 — Connect read-only
Connect only what the job needs (often calendar and email), with read access where the tool allows it. No passwords or card numbers in instructions.
Day 3 — Put it on a schedule
Scheduled runs make it a habit, not a chat. In ChatGPT, Free accounts can run a task at most once a day in a flexible window; paid plans allow exact times and up to hourly (OpenAI Help, Scheduled tasks). Other tools have their own schedulers.
Day 4 — Fix the output format
Tell it exactly what a good result looks like: short, sorted, quiet when nothing changed. Most first-week fixes are format, not intelligence.
Day 5 — Add drafts, not sends
Let it draft replies or follow-ups for you to approve. Anything that sends a message or changes data should wait for your yes; some tools pause the task for approval automatically.
Day 6 — Write the handoff rules
List what always comes to you: money, complaints, legal or health questions, anything public.
Day 7 — Review, then add job two
Ask: did I use it every day? What did I fix by hand? Keep, change or drop it. Only then add the next job. Estimate: expect roughly 15–30 minutes a day of your attention in week one while you tune it.
Not sure which jobs fit your business?
Build my free AI plan →Ready-made use cases
- 01CEO morning brief in 5 minutes
One message each morning: calendar, urgent emails, money in/out, and the 3 things that move the business today.
~3h/week · estimate - 02Inbox triage with drafted replies
The agent sorts your inbox into reply / FYI / ignore and drafts replies in your voice for you to approve.
~5h/week · estimate - 03Never miss a lead follow-up
The agent checks your CRM or inbox for leads that went quiet and drafts the next follow-up.
~4h/week · estimate - 04Weekly CEO review & plan
Every Friday the agent compiles wins, numbers, stuck items and proposes next week's priorities.
~2h/week · estimate - 05Meeting prep dossier before every call
30 minutes before a call you get who they are, what they want, past emails and suggested questions.
~2h/week · estimate
Example: Grok Bot setup
Grok Bot routine: CEO morning brief in 5 minutes Schedule: Daily 7:30 Every morning, read my calendar for today, my unread emails from the last 24h and yesterday's revenue. Send me a brief: (1) meetings with one prep note each, (2) emails that need my reply with a suggested one-line answer, (3) money in/out, (4) the 3 highest-leverage tasks for today. Keep it under 200 words. Rules: draft first, never send or post anything external without my approval. Keep summaries short.
Grok Bot, ChatGPT, Claude and n8n versions: all templates
Setup steps
- Write the job in one sentence: what, when, where the result goes.
- Connect only the apps it needs, read-only where possible.
- Put it on a schedule (daily is enough to start).
- Fix the output format until you'd read it every day.
- Turn on drafts you approve; no auto-sending in week one.
- Write handoff rules for money, complaints, legal and anything public.
- On day 7, review and only then add a second job.
Keep these human
- One job in week one. Adding three at once makes it impossible to tell what's working.
- Read-only first, then drafts, then sending — never the other way round.
- Keep passwords, card numbers and health or financial details out of instructions and task titles.
- Anything public, financial or legal stays with you.
- Start in draft mode: the agent prepares, you approve, then you loosen the leash routine by routine.
FAQ
What should an AI agent do first for Your AI agent's first week?
Day 1 — Pick one job and write it down: One sentence: what it does, when, and where the result goes. Example: 'Every weekday at 8:00, send me today's calendar, emails that need a reply, and anything overdue.' A morning brief is the safest first job because it only reads. Day 2 — Connect read-only: Connect only what the job needs (often calendar and email), with read access where the tool allows it. No passwords or card numbers in instructions. Day 3 — Put it on a schedule: Scheduled runs make it a habit, not a chat. In ChatGPT, Free accounts can run a task at most once a day in a flexible window; paid plans allow exact times and up to hourly (OpenAI Help, Scheduled tasks). Other tools have their own schedulers.
What should stay human?
One job in week one. Adding three at once makes it impossible to tell what's working. Read-only first, then drafts, then sending — never the other way round. Keep passwords, card numbers and health or financial details out of instructions and task titles. Anything public, financial or legal stays with you. Start in draft mode: the agent prepares, you approve, then you loosen the leash routine by routine.
Is there a ready-made template?
Yes. Start with "CEO morning brief in 5 minutes": copy-ready Grok Bot, ChatGPT, Claude and n8n templates are at https://whatshouldmybotdo.com/use-cases/ceo-morning-brief#templates.