Start here: Build a role, not a prompt
The biggest mistake will be giving a Dot vague instructions like:
“Help me with my business.”
Instead, define five things:
Role — What is this Dot responsible for?
Goal — What outcome should it continuously improve?
Context — What information and apps should it watch?
Boundaries — What can it do alone vs. ask you about?
Output — How should it bring work back to you?
Then the workflow sections should look like:
01. The Morning Operator
Start your day with decisions, not inboxes
Most people don’t begin their day doing meaningful work. They begin by checking five different places just to understand what happened while they were away.
Email. Slack. Calendar. Tasks. Notifications. Then, 30 minutes later, you finally know what you’re supposed to care about.
This is a perfect first responsibility for a Dot.
What to give it access to
Your calendar, inbox, team communications, project tools, and any source where important work tends to appear.
Mission
Every morning, review what has changed since my last check-in. Identify anything urgent, blocked, overdue, waiting on me, or likely to become a problem. Prioritize what deserves my attention and prepare anything you can before involving me.
Ask it to return
Today’s priorities
The three things most worth your attention.
Already handled
Research, summaries, drafts, or preparation it completed for you.
Needs your decision
Only the items where your judgment is actually required.
Worth watching
Things that matter, but not yet.
The goal here isn’t to create a prettier to-do list.
It’s to stop spending the first part of your day figuring out what your day is.
02. The Revenue Operator
Catch the money that usually disappears between the cracks
You probably don’t need an AI to tell you to “make more money.”
What is actually useful is having something notice the small situations where money is already available — but nobody followed through.
An unpaid invoice. A proposal that went quiet. A client who asked about additional work. A contract approaching renewal. An expense you forgot to submit.
These things are easy to miss because none of them feel urgent individually.
Together, they add up.
Mission
Watch for financial opportunities or obligations that may be getting overlooked. Look for unpaid invoices, forgotten follow-ups, upcoming renewals, proposals without a response, reimbursable expenses, or conversations that could become additional business. Prepare the next step whenever possible and ask me before taking any consequential external action.
Ask it to surface
Money owed
Invoices, payments, reimbursements.
Money at risk
Renewals or deals likely to lapse.
Money available
Upsells, follow-ups, or opportunities already present in your conversations.
Next action
The exact email, invoice, proposal, or reminder needed.
You shouldn’t have to remember every place where money might be hiding.
03. The AI Release Radar
Stop trying to manually keep up with everything
If you work anywhere near AI, you already know the problem.
There is always another launch.
Another model. Another feature. Another “game-changing” demo. Another thread telling you that everything has changed again.
The advantage isn’t seeing more news.
It’s knowing which updates deserve your attention.
Mission
Track meaningful developments in the areas I care about. Ignore minor announcements and surface only changes that introduce an important capability, remove a limitation, create a new opportunity, or materially change how people work.
Then ask your Dot to go one step further.
For every important development, explain what changed, why it matters, what most people will probably focus on, and what implication they may be missing.
Your ideal output
What happened
The release in plain English.
Why it matters
The actual consequence.
What others will say
The obvious narrative.
What they might miss
The angle worth thinking about.
What to do now
Test it, ignore it, build with it, or watch it.
That turns information overload into leverage.
04. The Content Operator
Give your ideas somewhere to go before they disappear
Most content systems don’t suffer from a lack of ideas.
They suffer from fragmentation.
A good thought is sitting in a voice note. A customer says something interesting in Slack. A founder mentions a useful insight on a call. A new report drops. Something performs unexpectedly well.
Then nobody connects any of it.
Your Dot can.
Mission
Learn what I publish, who I create for, what I consider useful, and what good content looks like to me. Watch the source material I give you and identify opportunities to turn it into finished content. Prioritize useful or surprising ideas over generic commentary.
You can also give it a permanent editorial rule:
Don’t create content simply because a topic is trending. Every piece should teach something, reveal something, challenge an assumption, provide a useful asset, or give someone a reason to share it.
Have it prepare
Ideas worth pursuing
Finished first drafts
Alternative hooks
Repurposing opportunities
Follow-ups based on performance
The objective is not to generate more content.
It’s to capture more of the good material you’re already producing every day.
05. The Product Feedback Pipeline
Turn repeated complaints into work that actually gets done
Every product team has some version of the same problem.
Users mention the same annoyance repeatedly.
Everyone agrees it matters.
It gets added somewhere.
Then three months later, someone asks why it still hasn’t been fixed.
A Dot is particularly useful here because it can help connect the entire chain.
Mission
Continuously review customer feedback, support conversations, bug reports, and product discussions. Identify repeated patterns, group related issues, estimate their impact, and surface improvements that appear both valuable and realistically actionable.
For appropriate issues:
Investigate the problem, determine the likely cause, propose the smallest effective fix, and prepare the work for review.
Have it classify issues by
Frequency
How often does this appear?
Severity
How painful is it?
Impact
Who does it affect?
Effort
How difficult is the fix likely to be?
Confidence
Do we actually understand the problem?
Instead of:
Feedback → backlog
you’re trying to move toward:
Feedback → evidence → solution → review
06. The Sales Deal Desk
Never reopen a deal wondering, “Where were we?”
Complex sales conversations generate an absurd amount of context.
Requirements. Objections. Pricing discussions. Technical questions. Promises. Follow-ups. Internal notes.
And the larger the opportunity becomes, the easier it is for one unresolved detail to quietly slow everything down.
Mission
Maintain an up-to-date understanding of every important opportunity. Track what the customer wants, what we have promised, unresolved questions, technical requirements, objections, risks, decision-makers, and the most useful next action.
Whenever something changes, ask it to update:
Current deal state
What the customer cares about
Outstanding blockers
What we still need to prove
Recommended next step
Draft response or material needed
The Dot handles the memory.
Your team gets to focus on judgment, negotiation, and relationships.
07. The Launch Operator
Keep launches aligned when the plan inevitably changes
Every launch begins with a clean plan.
Then the product changes.
The positioning changes.
A feature gets delayed.
Pricing moves.
Somebody updates the landing page but forgets the email sequence.
Now five different versions of the truth exist.
That is exactly the kind of coordination work an always-on agent can help with.
Mission
Understand the current launch strategy, audience, positioning, timeline, product scope, and messaging. When something changes, identify every downstream asset affected and surface inconsistencies before they reach customers.
Give it access to the relevant:
Product documentation
Landing pages
Email campaigns
Sales collateral
Social posts
FAQs
Creative briefs
Launch schedule
Then ask:
Whenever new information conflicts with existing material, tell me what changed, what is now inaccurate, and what needs to be updated.
You are essentially giving the launch a memory of its own.
08. The Opportunity Scout
Look for evidence before you look for ideas
Asking AI to generate startup ideas is easy.
Getting useful ones is harder.
A better approach is to have your Dot look for repeated pain.
What are people constantly complaining about?
What requires too many manual steps?
Where are companies paying humans to move information between systems?
What do people keep trying to solve with spreadsheets, workarounds, or five different tools?
Mission
Look for recurring problems, inefficient workflows, repeated complaints, expensive manual processes, unmet requests, and situations where new AI capabilities make a previously difficult solution possible.
For every interesting opportunity, require:
The problem
What exactly is frustrating?
The buyer
Who cares enough to pay?
The current workaround
How do they solve it today?
Why now
What recently became possible?
The smallest useful product
Not the dream company — the first useful version.
The validation test
How could we test demand before building it?
That changes the question from:
“What should I build?”
to:
“Where is the evidence that something needs to exist?”
09. The Personal Chief of Staff
Reduce the amount of your life you have to keep in your head
Some of the most exhausting work isn’t difficult.
It’s remembering.
Remembering the document you need for next week. The person you promised to reply to. The trip you still haven’t prepared for. The appointment you need to schedule. The thing you’ve moved to tomorrow four days in a row.
Your Dot can become a second layer of awareness around your life.
Mission
Understand my current priorities, commitments, deadlines, travel, appointments, and unfinished obligations. Look for conflicts, things I am likely to forget, preparations I should make in advance, and administrative work that can be handled or prepared without consuming my attention.
Have it organize things into:
Coming up
Prepare now
Waiting on someone
Needs my decision
Can be delegated
Probably not worth doing
That last category matters too.
A good chief of staff shouldn’t just help you do more.
It should occasionally tell you what you can stop doing.
10. The Proactive Operator
The most interesting use of Dots may be the work you never assign
This is where the concept becomes more interesting.
Most AI still begins with you.
You recognize a problem.
You decide AI could help.
You formulate the request.
You wait for the result.
But once an agent can maintain context over time, you can begin giving it something broader:
permission to notice.
Mission
When you don’t have an explicit task from me, look for useful work connected to my active goals. Focus on overlooked opportunities, unfinished commitments, inconsistencies, approaching deadlines, recurring problems, and information that materially changes something I care about.
Then add an important filter:
Don’t interrupt me with observations that are merely interesting. Investigate first. Bring me something only when you have a useful recommendation, prepared work, a meaningful warning, or a clear decision that requires me.
That distinction is everything.
You don’t want an AI employee constantly saying:
“I noticed something.”
You want:
“I noticed something, investigated it, and here’s what I think we should do.”
Set boundaries before you expand autonomy
The goal isn’t to make your Dot autonomous as quickly as possible.
The goal is to make it reliably useful.
Start by giving it freedom where mistakes are inexpensive:
Usually safe to delegate
Research
Reading
Analysis
Organization
Summarization
Drafting
Preparing documents
Comparing options
Then keep consequential actions behind approval:
Ask first
Sending messages
Publishing publicly
Spending money
Deleting information
Changing important account settings
Making commitments
Accepting terms or contracts
As you use your Dot, watch where you repeatedly approve the same type of action.
Those patterns tell you where a new rule may eventually make sense.
Autonomy should be earned through repetition, not granted because the technology feels exciting.
The 7-Day Dot Challenge
If you’re getting access to Dots, resist the temptation to create ten of them immediately.
Start with one.
Give it a real responsibility and spend a week teaching it what good looks like.
Day 1 — Choose the job
Pick something you already deal with repeatedly.
Day 2 — Give it context
Connect the information it genuinely needs.
Day 3 — Show it your standards
Give examples of good work, bad work, and previous decisions.
Day 4 — Give it real work
Not a test. Something you would otherwise have done yourself.
Day 5 — Correct it
Tell it specifically where its judgment was wrong.
Day 6 — Define recurring rules
Turn repeated corrections into clearer instructions.
Day 7 — Ask it what’s next
And finish the week with this:
Based on everything you’ve learned about how I work, what am I still doing manually that you think you should own?
That question gets to the real opportunity.
The biggest benefit of Dots probably won’t come from discovering one magical prompt.
It will come from gradually identifying entire categories of work you no longer need to personally carry.
So as you experiment, keep returning to one question:
What should I never have to do manually again?
THE AI LEVERAGE
Stay ahead of what matters in AI.
By @bestapps.ai on Instagram
