Skip to content

Instantly share code, notes, and snippets.

Show Gist options
  • Select an option

  • Save manduks/a050bba9f1673a7217116c280c3a887e to your computer and use it in GitHub Desktop.

Select an option

Save manduks/a050bba9f1673a7217116c280c3a887e to your computer and use it in GitHub Desktop.
AI Audit Checklist for Solo Coaches & Consultants — 14 Questions to Run Before You Spend Another Dollar on ChatGPT (April 2026)

Most solo coaches and consultants we talk to have already paid for ChatGPT Plus. Half of them have paid for Claude Pro, Notion AI, and at least one "AI for coaches" course on top.

None of them can answer this question: what changed in your week?

That's the gap an AI audit closes. Not "what AI tools exist" — that's a Google search. The audit answers: given your specific workflow, your specific clients, and your specific Tuesday afternoon, where does AI save you 5+ hours per week and where does it just burn $20/month?

This is the checklist we use on $997 client engagements, simplified to 14 questions you can run on yourself today. There's a free fillable template at the bottom.


TL;DR — 14 questions worth a real answer

# Question What "no" or "I don't know" usually means
1 Where does my week actually go, by hour? You're optimizing the wrong workflow
2 What 3 tasks repeat every week? Your highest-ROI automation candidates
3 Which of those tasks needs my judgment vs my typing? Typing tasks = AI. Judgment tasks = stay you.
4 What does each repeating task cost per hour (revenue × hours billed)? You can't price an automation without this
5 What tools have I paid for and never used past month 1? $40-$200/mo of dead subscriptions, usually
6 Which client deliverable do I dread the most? This is your #1 candidate, not your "easiest" task
7 When I describe my work to AI, do I get generic output? You're missing context, not "the right tool"
8 What's my privacy floor — what data must NEVER leave my device? Determines free vs paid vs local-first models
9 What's my technical floor — can I actually use APIs, or only chat UIs? Eliminates 60% of "AI for coaches" advice
10 What 5 hours/week saved would I actually do something with? If the answer is "scroll Twitter" — don't bother
11 What does my best client get from me that AI cannot replicate? This is your moat — automate everything else
12 If I spent $1,000 on AI tooling tomorrow, would I know which 3 to buy? If no → the audit is the cheapest version of that decision
13 What does failure look like in 90 days? Without this, you can't tell ROI from busywork
14 Who would I hire to do this audit if money were no object? If your answer is "a $250/hr consultant" — that's the math

Free fillable version of this checklist — answers tracked in a Notion-style template, plus a 1-page summary at the end you can hand to a contractor: midastools.co/audit-template

Done-for-you version — we run all 14 questions, plus 6 more, plus deliver a 15-25 page custom report with tool recommendations, monthly cost, hours-saved math, and a 30-min review call: $997 AI Clarity Assessment. Full refund if we don't save you 5+ hrs/wk.


The universal AI-audit question formula

Every question above fills five slots. When you write your own audit questions for your business, follow the same shape:

[ROLE]:        who you are when you do the task (writer? researcher? CFO? therapist?)
[TASK]:        the smallest unit that recurs — "draft Tuesday's nurture email", not "marketing"
[FREQUENCY]:   per day, per week, per client, per launch
[TIME COST]:   minutes of focused work, not calendar hours
[JUDGMENT %]:  what fraction is "decisions only you can make" vs "typing it out"

A task with 80%+ judgment is not an AI candidate, even if AI can technically do it. Your discovery calls aren't a bottleneck — your follow-up notes are.


1. Where does my week actually go, by hour?

Track every 30-minute block for 5 working days. Don't estimate — log.

For each block, write:
  TASK:       what you actually did
  CLIENT:     who paid for it (if anyone)
  REVENUE:    $ value of that block to you, honestly
  ENERGY:     +1 / 0 / -1 (did this leave you sharper or fried?)
  REPEATS:    weekly / monthly / one-off

After 5 days, sort by REPEATS=weekly and ENERGY=-1. Those are your audit
candidates. Everything else is noise.

Most coaches think their week is 60% client delivery. Real tracking shows it's 25% delivery + 30% prospect/admin + 25% content + 20% fragmented context-switching. You can't fix what you don't measure.

2. What 3 tasks repeat every week?

From the log above, name your top 3 weekly-recurring tasks. Write them as:

  Task 1: [name] — frequency: [N times/week] — total time: [hours]
  Task 2: [name] — frequency: [N times/week] — total time: [hours]
  Task 3: [name] — frequency: [N times/week] — total time: [hours]

If two of the three sound similar, merge them. You should have at most
3 distinct workflows, not 7.

Stop here if you can't fill it in. The audit can't help if you don't know what your week is.

3. Which tasks need my judgment vs my typing?

For each of the 3 tasks above, label each step as one of:

  J = judgment (only I can decide)
  T = typing (a competent assistant could type this)
  R = research (gathering known information)

Most "I'll do it myself" workflows are 20% J + 60% T + 20% R.
The 80% (T + R) is your AI surface area. Don't try to automate the J.

This is the question that kills the most "AI replaces coaches" panic. Coaches who bill for judgment ($300/hr) and try to "automate coaching" lose. Coaches who automate the typing around their judgment win.

4. What does each repeating task cost per hour?

For each of your top 3 tasks:

  Hours/week:        [X]
  Your billable rate: [$Y/hour]
  Annual cost:        [X × Y × 50 weeks] = [$Z]

If you can't bill the hours back (e.g., admin, content), use your
opportunity cost — what would you do with that hour? If the answer is
"another client call at $300/hr", the math is the same.

This number is what makes the audit obvious. A 3-hour/week task at a $250 effective rate costs you $37,500/year. A $50/month tool that cuts it in half pays back in 11 days.

5. What tools have I paid for and never used?

Open your last 90 days of credit-card statements. List every recurring
SaaS charge under "AI" / "productivity" / "writing" / "design". For each:

  Tool:            [name]
  $ /mo:           [X]
  Last opened:     [date]
  Hours used last 7 days: [N]

Anything where "last opened" > 30 days ago: cancel today. You don't need
an audit to do this — you just have to look.

Average solo coach we audit has $147/month in AI tooling, of which $89 is unused. Cancel-the-unused alone pays for the assessment in 11 months.

6. Which client deliverable do I dread the most?

For each recurring deliverable (proposals, reports, follow-ups, content):

  Deliverable:    [name]
  Dread level:    1 (love) → 5 (procrastinate for 3 days)
  Hours/instance: [X]
  Per month:      [N instances]

The dread-5, hours-2+, multiple-per-month deliverable is your
highest-leverage audit target. Not the easiest one. The dreadful one.

Easy tasks aren't bottlenecks. Dreaded tasks are. Solving the dread is what unlocks the other 4 hours.

7. When I describe my work to AI, do I get generic output?

Open ChatGPT or Claude. Paste this exact prompt:

  "I'm a [your title]. I [one-sentence what you do] for
  [your ICP]. I'm trying to [specific thing] for a client who
  [specific thing about them]. Write me a [deliverable] that
  [specific outcome]."

If the output sounds like a Medium blog post — generic, "in today's
fast-paced world" — you don't have a tool problem. You have a
context problem. Every audit fixes this in question #11.

Free fix: try the prompt enhancer — paste your messy draft, get back something with the right context shape. Won't fix bad context but will surface what's missing.

8. What's my privacy floor — what data MUST NOT leave my device?

Categorize the data you touch weekly:

  Tier A (never leaves device): client SSNs, medical, attorney-client privileged,
                                custody disputes, signed NDAs, board-confidential
  Tier B (only encrypted SaaS):  client revenue figures, child names, addresses,
                                 unsigned contracts
  Tier C (any tool is fine):     your own marketing copy, public testimonials,
                                 industry research, anonymized examples

Tier A means local-only models (Ollama, LM Studio, Apple Intelligence on-device).
Tier B means the paid tier of a major provider with DPA + zero-retention.
Tier C is open season.

If you don't know your floor, you'll either over-restrict (and use no AI)
or under-restrict (and breach your client's trust). Neither helps.

This is the question that disqualifies 60% of "AI for therapists" courses — they recommend tools that violate the therapist's licensing-board data-handling rules.

9. Can I actually use APIs, or only chat UIs?

Honestly score yourself 1-5 on each:

  ChatGPT/Claude chat UI:     [1-5]
  Custom GPTs / Claude projects: [1-5]
  Zapier / Make.com:           [1-5]
  Browser automation (Manus, Browser Use): [1-5]
  Local API key + Python:      [1-5]
  Self-hosted models:          [1-5]

Where you're below 3, the gap is training, not tooling. Where you're
already 3+, the gap is workflow design. The audit recommendation
depends entirely on this score — recommending Zapier to a 1 is cruel.

10. What 5 hours/week saved would I actually do something with?

Write the answer in one sentence. Examples:

  Yes:  "I'd onboard 2 more clients at $1,500/mo each = $3,000 MRR"
  Yes:  "I'd finish the cohort course I've been drafting for 14 months"
  Yes:  "I'd recover from burnout and not blow up my marriage"
  No:   "I'd... I dunno, watch more shows?"
  No:   "I'd start scrolling X again"

If the answer is no, save the $997 and don't run the audit.
The constraint isn't time — it's something else.

This is the most honest question on the list. Half the people we talk to discover at this question that they don't have a time problem.

11. What does my best client get from me that AI cannot replicate?

List the top 3 things your favorite client says about working with you.
Now mark each:

  M = Moat (AI cannot replicate this in 5 years)
  C = Capacity (AI could do this, you just have to)
  A = Adminish (AI absolutely could already)

Most coaches discover their M is presence + accountability + reading the
person in front of them. Their C is "weekly follow-up emails, summaries,
session notes". Their A is everything else.

Automate everything that is C or A. Protect everything that is M.

If you don't know your M, the audit will surface it for you — that's a deliverable, not just a recommendation.

12. If I spent $1,000 on AI tooling tomorrow, would I know which 3 to buy?

Write down the 3 tools you'd buy, with:

  Tool:        [name]
  $/mo:        [cost]
  Use case:    [one specific recurring task it fixes]
  Why this:    [one sentence on why this over the other 12 contenders]

If you can't fill in all 4 columns for 3 tools, you're going to make a
wrong $1,000 bet. The audit IS that decision, $3 per tool researched,
plus the work of actually testing them in your workflow.

13. What does failure look like in 90 days?

Write 3 sentences:

  If the audit recommendations work, in 90 days I will:
    [X concrete observable thing]

  If the audit recommendations partially work, in 90 days I will:
    [Y honest middle-state]

  If the audit recommendations don't help at all, in 90 days I will:
    [Z honest worst case — "no change in hours, $50/mo wasted on tools"]

Without this, you can't tell ROI from busywork. With this, you can.

This is what we put on page 2 of every $997 deliverable. If you're DIYing the audit, write it for yourself anyway.

14. Who would I hire to do this audit if money were no object?

Be honest. Not "myself with more time". Not "ChatGPT for free".

  Person/firm:  [name]
  Their rate:   [$/hr]
  Why them:     [one specific reason]

If your answer is "a $250-500/hr consultant who actually knows my industry",
that's the alternative. The math:

  10-hour assessment × $300/hr = $3,000
  Our productized version:        $997, 7 days, written deliverable + call

That's it. That's the pitch. We don't try to convince you we're better
than the $300/hr consultant — we tell you when to hire one and when to
buy the productized version.

DIY vs $997 audit vs $10K consultant

Slot DIY checklist (this gist) $997 productized audit $10K bespoke consultant
Time to deliverable 6-10 hours of your week 7 days 4-8 weeks
Discovery call None 60 min 3-4 sessions
Custom analysis You do it We do it They do it deeper
Tool recommendations Self-research 3-5, tested for you 5-15, with vendor relationships
Implementation order You guess Ranked by ROI Ranked + project-managed
Hours-saved math Your gut Calculated per workflow Calculated + measured
Refund if wrong N/A Yes — full refund unless 5+ hrs/wk saved Usually not
Best for Anyone with 6 free hours Solo coaches/consultants $5K-$25K MRR Teams of 3+, $50K+ MRR

You are not the audience for the $10K consultant if you're a solo expert. You probably are the audience for one of the first two columns.


Common mistakes (from 50+ informal audits)

  1. Treating "AI for coaches" courses as substitutes for an audit. Courses teach you tools. An audit tells you which tools, in what order, for your week. They're complementary, not competitive.

  2. Auditing the easiest workflow first. Easy workflows aren't bottlenecks — they're just easy. Audit the dreaded one (question #6). The unlock is in the dread.

  3. Skipping question #10. Half of all audit-curious people don't actually have a time problem. They have an energy problem, a focus problem, or a confidence problem. None of those are fixed by Notion AI.

  4. Confusing "AI saved me time on this email" with "AI saved me 5 hrs/week". One email is a vibe; 5 hrs/week is a measurable change in your calendar. Track question #1 again 30 days after implementing.

  5. Hiring an "AI consultant" who doesn't know your industry. A generic AI consultant will recommend the same 5 tools to a real estate agent, a pastor, an IP attorney, and a cohort-course coach. The recommendation depends on the industry's privacy floor, content cadence, and client deliverable shape. Ask for industry references.


Resources

  • Free fillable template — the 14 questions in a Notion-style table with a 1-page summary at the end: midastools.co/audit-template
  • $997 AI Clarity Assessment — done-for-you version of this checklist plus 6 more questions, custom 15-25 page report, 30-min review call, full refund if we don't save you 5+ hrs/wk: midastools.co/ai-audit
  • For Coaches & Consultants — the broader product line we built for this audience: midastools.co/for-coaches
  • Free prompt enhancer — fixes the "generic AI output" problem from question #7 by upgrading your context: midastools.co/prompt-enhancer
  • Done-for-you implementation — for the recommendations from the audit you don't want to install yourself: midastools.co/services

Built by Midas Tools. We sell clarity, not automation. — April 2026

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment