GPT EPD · one AI experiment a day
Inspired on Thursday.Unchanged by Monday.
That is the honest problem with AI talks, mine included. You leave with a full notebook and a feeling, and by Monday the feeling has gone and the inbox is exactly where you left it.
So: thirty days instead of one evening. Fifteen minutes a day. And at the end of it you are holding thirty things you made — not thirty prompts you saved.
Starts 1 October 2026
#GPTEPDWhatever happens, post it. It goes up with the rest of them.
An homage to VPD — a video every day. Same idea, different reps: instead of filming something, you run one experiment.
Marketing & Visibilityyou keep
- 01Teach GPT about your firmFirm brief
- 02Find your favourite clientsIdeal client profile
- 03Map their problemsClient problem map
- 04Borrow their languageClient language bank
- 05Challenge your assumptionsValidated client brief
… and twenty-five more days after these five, each one starting from the last.
Day one is fifteen minutes and a conversation.
01
Marketing & Visibility ·Context setting
Teach GPT about your firm
Describe services, clients, location and approach; ask what context is missing.
That is the whole of day one. You end it holding a written description of your own practice that is good enough the model stops guessing — which is the thing that makes every day after it work.
#GPTEPDDay one costs fifteen minutes. Telling someone how it went costs two, and it is the part that turns this from a reading list into a room.
None of these days stand on their own. That is the entire trick.
Day 01
Teach GPT about your firm
Describe services, clients, location and approach; ask what context is missing.
Leaves you: Firm brief
Day 02
Find your favourite clients
Describe five anonymized good-fit clients and ask for common traits.
Leaves you: Ideal client profile
Day 03
Map their problems
Use the profile to identify goals, worries and recurring questions.
Leaves you: Client problem map
Day three does not say describe your clients. It says use the profile — because you built one on day two. Day nine writes a positioning statement against the brief you wrote on day five. Day twenty-two rebuilds your homepage using all prior assets. And on day twenty-nine you take the firm brief, the audience, the voice and the positioning and fold them into one set of instructions you can hand the model every morning.
A prompt library cannot do that. It just gets longer.
Day 22 · Fix the homepage
“Rework the opening, proof and next step using all prior assets”
Day 29 · Build your copilot
“Combine firm brief, audience, voice, positioning and review rules”
Nine chains. Every line of every one of them.
Nothing is held back and there is nothing to hand over to see it. Pick the chain where a small change would travel furthest in your firm, then read down it — the column alongside keeps count of what you would be holding by the time you got there.
40 days each
360 in all
Chain 01 of 09
Marketing & Visibility
Become known for what you know.
Nothing yet — read down and watch it fill.
What you're holding
of 30 assets
30 connected · 10 stretch
Day zero
You own nothing yet. Read down the list and watch that change.
Context & audience
Days 01–05
- 01
Teach GPT about your firm
Describe services, clients, location and approach; ask what context is missing
Context setting
KeepAsset you keep: Firm brief
- 02
Find your favourite clients
Describe five anonymized good-fit clients and ask for common traits
Example selection
KeepAsset you keep: Ideal client profile
- 03
Map their problems
Use the profile to identify goals, worries and recurring questions
Audience analysis
KeepAsset you keep: Client problem map
- 04
Borrow their language
Role-play a client describing those problems without accounting jargon
Role-play
KeepAsset you keep: Client language bank
- 05
Challenge your assumptions
Ask GPT which claims in your client profile need real-world validation
Assumption testing
KeepAsset you keep: Validated client brief
Positioning
Days 06–10
- 06
Audit your introduction
Dictate how you explain your firm; ask what sounds vague
Message critique
KeepAsset you keep: Messaging audit
- 07
Find your difference
Compare your approach and experience with common alternatives
Comparison
KeepAsset you keep: Differentiator list
- 08
Define poor fit
Identify people your firm would struggle to serve well
Constraint setting
KeepAsset you keep: Not-for-us profile
- 09
Write your positioning
Iterate a one-sentence promise against your client brief
Iteration
KeepAsset you keep: Positioning statement
- 10
Pass the plain-language test
Explain your promise to a ten-year-old without losing meaning
Simplification
KeepAsset you keep: Plain-English description
Expertise & voice
Days 11–15
- 11
Let GPT interview you
Answer ten probing questions by voice on one area of expertise
Conversational AI
KeepAsset you keep: Expert interview
- 12
Mine the interview
Extract genuine lessons, stories and opinions from the transcript
Extraction
KeepAsset you keep: Expertise library
- 13
Find your point of view
Compare your own posts or notes and name recurring beliefs
Pattern recognition
KeepAsset you keep: Point-of-view list
- 14
Argue against yourself
Have GPT challenge three beliefs; revise what you still stand behind
Counterargument
KeepAsset you keep: Stronger points of view
- 15
Choose content pillars
Group client problems and expertise into three to five themes
Classification
KeepAsset you keep: Content pillars
Content
Days 16–20
- 16
Build an idea bank
Generate ideas from the pillars and discard anything generic
Idea generation
KeepAsset you keep: Curated idea bank
- 17
Write the ugly first draft
Draft one post yourself; use GPT only to improve clarity
Editing
KeepAsset you keep: Signature post
- 18
Find a real story
Have GPT interview you for an experience behind a lesson
Story discovery
KeepAsset you keep: Story post
- 19
Turn one into four
Adapt a strong post to email, video outline, carousel and FAQ
Transformation
KeepAsset you keep: Content package
- 20
Teach your voice
Compare genuine writing samples and define style rules
Voice analysis
KeepAsset you keep: Voice guide
Discoverability
Days 21–25
- 21
Audit your website
Role-play the ideal visitor and mark doubts and dead ends
Perspective taking
KeepAsset you keep: Website audit
- 22
Fix the homepage
Rework the opening, proof and next step using all prior assets
Synthesis
KeepAsset you keep: Homepage outline
- 23
Collect real questions
Map the questions clients ask before choosing a firm
Question mapping
KeepAsset you keep: Question library
- 24
Build topic clusters
Group questions into themes that show actual expertise
Topic clustering
KeepAsset you keep: Authority map
- 25
Check AI discoverability
Compare public evidence of expertise with the authority map
Evidence audit
KeepAsset you keep: Discovery improvement plan
A working system
Days 26–30
- 26
Build a useful download
Turn one recurring question into a checklist or scorecard
Product thinking
KeepAsset you keep: Lead magnet draft
- 27
Map the journey
Connect discovery, site, download, consultation and follow-up
Journey mapping
KeepAsset you keep: Client journey
- 28
Plan a small campaign
Set one goal and sequence content and calls to action
Campaign planning
KeepAsset you keep: 30-day campaign
- 29
Build your copilot
Combine firm brief, audience, voice, positioning and review rules
Reusable instructions
KeepAsset you keep: Marketing copilot instructions
- 30
Audit the whole build
Ask a skeptical GPT for the weakest link and one measurable test
Strategic critique
KeepAsset you keep: Next-month experiment plan
Keep experimenting
Experiments 31–35
Optional
- 31
Interview a client
Use consented feedback to check your client-language assumptions
Interview analysis
KeepAsset you keep: Validated language notes
- 32
Test three hooks
Publish three ethical variations and compare meaningful response
A/B testing
KeepAsset you keep: Hook test log
- 33
Audit search results
Review actual results manually; ask GPT to group gaps
Search analysis
KeepAsset you keep: Search evidence map
- 34
Build an FAQ page
Turn vetted client questions into concise reviewed answers
Answer drafting
KeepAsset you keep: FAQ draft
- 35
Make a video series
Sequence five expert stories into short talking outlines
Series planning
KeepAsset you keep: Video series plan
Keep experimenting
Experiments 36–40
Optional
- 36
Design a webinar
Turn a high-interest question into a useful 30-minute session
Workshop design
KeepAsset you keep: Webinar outline
- 37
Build a referral page
Explain who you help and how introductions work
Referral framing
KeepAsset you keep: Referral page draft
- 38
Compare channels
Use real aggregate data to compare effort and outcomes
Channel analysis
KeepAsset you keep: Channel scorecard
- 39
Create a review loop
Define who checks claims, confidentiality, tone and accuracy
Quality control
KeepAsset you keep: Editorial checklist
- 40
Decide what to stop
Rank activities by evidence, effort and strategic fit
Prioritization
KeepAsset you keep: Stop-start-continue list
Chain 02 of 09
Sales & Business Development
Make the next conversation count.
Nothing yet — read down and watch it fill.
What you're holding
of 30 assets
30 connected · 10 stretch
Day zero
You own nothing yet. Read down the list and watch that change.
Fit & qualification
Days 01–05
- 01
Teach GPT your sales process
Describe services, price approach and inquiry-to-client steps
Context setting
KeepAsset you keep: Sales brief
- 02
Find your best prospects
Describe anonymized clients you would love to clone
Pattern recognition
KeepAsset you keep: Ideal prospect profile
- 03
Find poor-fit patterns
Review anonymized difficult engagements for early signals
Contrast analysis
KeepAsset you keep: Poor-fit profile
- 04
Build a qualification scorecard
Turn fit and poor-fit traits into objective criteria
Criteria design
KeepAsset you keep: Lead scorecard
- 05
Test the scorecard
Score five fictional prospects; revise ambiguous criteria
Simulation
KeepAsset you keep: Validated scorecard
Better conversations
Days 06–10
- 06
Find what you need to know
Ask which facts matter before recommending a service
Question generation
KeepAsset you keep: Discovery question bank
- 07
Build a conversation flow
Arrange questions so a consultation feels natural
Conversation design
KeepAsset you keep: Discovery flow
- 08
Role-play a good-fit prospect
Ask GPT to reveal information only when you ask well
Role-play
KeepAsset you keep: Practice transcript
- 09
Role-play a skeptic
Practise with a rushed or price-sensitive fictional prospect
Adaptive role-play
KeepAsset you keep: Difficult-prospect transcript
- 10
Coach your performance
Review both role-plays for missed questions and talking time
Coaching analysis
KeepAsset you keep: Coaching report
Diagnose first
Days 11–15
- 11
Look beneath the question
Explore possible motives behind ten common opening questions
Needs inference
KeepAsset you keep: Underlying-needs map
- 12
Ask a better second question
Write follow-ups that expose stakes and timing
Follow-up questioning
KeepAsset you keep: Follow-up question bank
- 13
Separate symptom from problem
Analyze fictional scenarios and flag missing facts
Diagnostic reasoning
KeepAsset you keep: Diagnostic framework
- 14
Practise delaying advice
Role-play technical questions while asking for context first
Conversational discipline
KeepAsset you keep: Diagnostic practice log
- 15
Summarize discovery
Convert a fictional call into goals, problems and unknowns
Structured synthesis
KeepAsset you keep: Discovery summary template
Communicate value
Days 16–20
- 16
Map client outcomes
Describe what clients gain beyond each deliverable
Outcome mapping
KeepAsset you keep: Outcome map
- 17
Translate features into value
Rewrite service descriptions in practical client language
Value translation
KeepAsset you keep: Value language bank
- 18
Study objections
List hesitations and plausible concerns beneath each
Objection analysis
KeepAsset you keep: Objection map
- 19
Practise live objections
Answer as yourself; ask GPT to coach, not write a script
Live practice
KeepAsset you keep: Objection practice log
- 20
Explain your price
Clarify scope, value and fit without apologetic language
Message refinement
KeepAsset you keep: Pricing guide
Decision & follow-up
Days 21–25
- 21
Structure a proposal
Decide what an informed buyer needs to see
Proposal structure
KeepAsset you keep: Proposal framework
- 22
Personalize the opening
Anchor a draft proposal in fictional discovery findings
Personalization
KeepAsset you keep: Client-centred opening
- 23
Write a useful recap
Summarize what was heard and the agreed next action
Call synthesis
KeepAsset you keep: Follow-up template
- 24
Investigate ghosting
Map several reasons for silence to respectful next steps
Scenario planning
KeepAsset you keep: Follow-up playbook
- 25
Ask for introductions
Identify natural moments for a specific referral request
Referral mapping
KeepAsset you keep: Referral system
A repeatable system
Days 26–30
- 26
Map the whole funnel
Diagram inquiry through onboarding and flag drop-offs
Process mapping
KeepAsset you keep: Sales funnel map
- 27
Study lost deals
Classify anonymized declines without assuming price was the cause
Case classification
KeepAsset you keep: Lost-opportunity analysis
- 28
Redesign the workflow
Assign humans, systems and AI distinct reviewable roles
Workflow redesign
KeepAsset you keep: Future sales workflow
- 29
Build your sales copilot
Combine qualification, discovery, value and proposal rules
Reusable instructions
KeepAsset you keep: Sales copilot instructions
- 30
Audit the sales engine
Challenge assumptions and choose one next-month test
System audit
KeepAsset you keep: Sales experiment plan
Keep experimenting
Experiments 31–35
Optional
- 31
Test an intake form
Compare questions against what discovery truly needs
Form critique
KeepAsset you keep: Intake form revision
- 32
Score ten leads
Apply criteria to fictional records and inspect inconsistency
Score calibration
KeepAsset you keep: Scorecard calibration
- 33
Role-play switching
Practise a prospect reluctant to leave an incumbent
Switching role-play
KeepAsset you keep: Switching conversation notes
- 34
Handle no decision
Explore what information a stalled buyer is missing
Decision analysis
KeepAsset you keep: Decision-support checklist
- 35
Audit proposal length
Remove sections that do not help a buyer decide
Document editing
KeepAsset you keep: Lean proposal template
Keep experimenting
Experiments 36–40
Optional
- 36
Map onboarding handoff
Specify what the team must know after a yes
Handoff design
KeepAsset you keep: Handoff checklist
- 37
Build a sales FAQ
Answer repeated questions using approved firm facts
FAQ drafting
KeepAsset you keep: Reviewed FAQ draft
- 38
Review lead sources
Compare actual aggregate fit and conversion by source
Source analysis
KeepAsset you keep: Lead-source scorecard
- 39
Train a teammate
Create role-play scenarios and objective coaching prompts
Coaching design
KeepAsset you keep: Practice kit
- 40
Retire weak tactics
Choose what to change based on evidence and feedback
Evidence-based prioritization
KeepAsset you keep: Sales improvement backlog
Chain 03 of 09
Client Experience
Make care feel effortless.
Nothing yet — read down and watch it fill.
What you're holding
of 30 assets
30 connected · 10 stretch
Day zero
You own nothing yet. Read down the list and watch that change.
See the journey
Days 01–05
- 01
Describe the current journey
List client steps from first inquiry through renewal
Journey mapping
KeepAsset you keep: Journey map
- 02
Identify anxious moments
Role-play a new client and mark moments of uncertainty
Perspective taking
KeepAsset you keep: Friction list
- 03
Gather real questions
Collect anonymized questions the team answers repeatedly
Question extraction
KeepAsset you keep: Question bank
- 04
Set a service promise
Write what clients can reasonably expect and when
Promise drafting
KeepAsset you keep: Service promise
- 05
Test the promise
Compare it with actual capacity and revise unrealistic claims
Feasibility testing
KeepAsset you keep: Validated promise
Onboarding
Days 06–10
- 06
Map onboarding
List tasks, owners, timing and client dependencies
Workflow mapping
KeepAsset you keep: Onboarding workflow
- 07
Simplify the welcome
Rewrite a welcome email around the next three actions
Plain-language writing
KeepAsset you keep: Welcome draft
- 08
Build a document request
Group required information with clear reasons and examples
Request design
KeepAsset you keep: Request checklist
- 09
Explain the first month
Create a day-by-day view of what happens after signing
Sequence planning
KeepAsset you keep: First-month guide
- 10
Test as a new client
Role-play misunderstanding the instructions and refine them
Usability testing
KeepAsset you keep: Onboarding test notes
Communication
Days 11–15
- 11
Define communication triggers
Identify events that merit a proactive update
Trigger design
KeepAsset you keep: Trigger map
- 12
Write progress updates
Draft short messages for three common milestones
Template drafting
KeepAsset you keep: Update templates
- 13
Make delays clear
Write an honest note explaining a delay and revised next step
Transparent writing
KeepAsset you keep: Delay template
- 14
Create a response standard
Set realistic channel, owner and response expectations
Standard setting
KeepAsset you keep: Communication standard
- 15
Audit tone
Test templates against anxious, busy and technical clients
Tone critique
KeepAsset you keep: Tone revision notes
Recurring care
Days 16–20
- 16
Map recurring work
Show when annual or monthly decisions and requests arise
Calendar mapping
KeepAsset you keep: Client calendar
- 17
Make tax season easier
Sequence requests, deadlines and simple reminders
Campaign sequencing
KeepAsset you keep: Seasonal communication plan
- 18
Explain the deliverable
Draft a plain-English summary of what the client receives
Explanation
KeepAsset you keep: Deliverable explainer
- 19
Prepare difficult news
Role-play a costly surprise with empathy and clarity
Empathy practice
KeepAsset you keep: Difficult-news guide
- 20
Close the loop
Define how a request becomes a confirmed resolution
Closure design
KeepAsset you keep: Resolution checklist
Handoffs & quality
Days 21–25
- 21
Build an escalation path
Specify when an issue reaches a manager or partner
Escalation logic
KeepAsset you keep: Escalation map
- 22
Design a handoff
Transfer context between team members without making clients repeat it
Context transfer
KeepAsset you keep: Handoff brief
- 23
Create meeting prep
Turn permitted history into goals and open questions
Briefing synthesis
KeepAsset you keep: Meeting prep template
- 24
Create meeting follow-through
Summarize decisions, owners and dates for review
Action extraction
KeepAsset you keep: Meeting recap template
- 25
Check accessibility
Test reading level, format and language of key messages
Accessibility review
KeepAsset you keep: Accessibility improvements
Learn and improve
Days 26–30
- 26
Ask for feedback
Write two short questions that invite specific, honest answers
Feedback design
KeepAsset you keep: Feedback prompt
- 27
Read feedback patterns
Cluster anonymized responses without overgeneralizing
Theme clustering
KeepAsset you keep: Feedback themes
- 28
Choose one fix
Rank client pain by frequency, severity and effort
Priority ranking
KeepAsset you keep: Improvement priority
- 29
Build a client-care copilot
Package approved process and tone with escalation rules
Reusable instructions
KeepAsset you keep: Copilot instructions
- 30
Measure next month
Define a simple baseline and one service improvement test
Pilot design
KeepAsset you keep: Client experience pilot
Keep experimenting
Experiments 31–35
Optional
- 31
Test a portal invite
Ask a fictional client to follow every step
Usability role-play
KeepAsset you keep: Portal usability notes
- 32
Reduce duplicate asks
Compare request points and consolidate them
Duplicate detection
KeepAsset you keep: Request consolidation
- 33
Build a status glossary
Explain common workflow states in everyday language
Terminology design
KeepAsset you keep: Status glossary
- 34
Design a check-in
Create an agenda for a meaningful annual review
Meeting design
KeepAsset you keep: Annual review guide
- 35
Recover a mistake
Draft apology, correction and follow-up steps
Service recovery
KeepAsset you keep: Service recovery playbook
Keep experimenting
Experiments 36–40
Optional
- 36
Examine complaints
Classify anonymized complaints by root cause
Root-cause analysis
KeepAsset you keep: Complaint pattern map
- 37
Map moments of delight
Find helpful surprises that are actually deliverable
Opportunity spotting
KeepAsset you keep: Care opportunity list
- 38
Test a client survey
Pilot two questions and see if responses are actionable
Survey testing
KeepAsset you keep: Survey revision
- 39
Review retention
Compare aggregate exit reasons with the journey map
Retention analysis
KeepAsset you keep: Retention hypotheses
- 40
Remove one annoyance
Pilot an improvement and record client feedback
Before-after evaluation
KeepAsset you keep: Before-after case
Chain 04 of 09
Operations & Workflow
Turn the messy middle into a system.
Nothing yet — read down and watch it fill.
What you're holding
of 30 assets
30 connected · 10 stretch
Day zero
You own nothing yet. Read down the list and watch that change.
Capture the work
Days 01–05
- 01
Choose one recurring process
Name its trigger, outcome and current owner
Scope definition
KeepAsset you keep: Process brief
- 02
Tell the story by voice
Explain each step as if training a new colleague
Voice capture
KeepAsset you keep: Voice transcript
- 03
Turn talk into steps
Convert the transcript into a numbered SOP draft
SOP drafting
KeepAsset you keep: SOP draft
- 04
Find missing decisions
Ask where instructions say “use judgment” without guidance
Ambiguity detection
KeepAsset you keep: Decision gaps
- 05
Walk a real example
Test the SOP with fictional inputs and note where it breaks
Walkthrough testing
KeepAsset you keep: Walkthrough notes
Map the handoffs
Days 06–10
- 06
Map systems and handoffs
Show where work enters, moves and exits
Process mapping
KeepAsset you keep: Process map
- 07
Find duplicate entry
Mark information copied between systems
Duplicate detection
KeepAsset you keep: Duplicate-work list
- 08
Find wait states
Identify approval, client and system delays
Delay analysis
KeepAsset you keep: Wait-state map
- 09
Clarify ownership
Assign one owner and reviewer to each step
Ownership mapping
KeepAsset you keep: Responsibility map
- 10
Set completion criteria
Define what “done” means and what evidence remains
Acceptance criteria
KeepAsset you keep: Done checklist
Control exceptions
Days 11–15
- 11
List exceptions
Name cases that cannot safely follow the standard path
Exception discovery
KeepAsset you keep: Exception register
- 12
Sort by risk
Rank exceptions by frequency, impact and reversibility
Risk ranking
KeepAsset you keep: Risk matrix
- 13
Build a review point
Specify when a qualified person must check output
Control design
KeepAsset you keep: Human-review rule
- 14
Create a QA checklist
Turn frequent errors into a short independent check
Checklist building
KeepAsset you keep: QA checklist
- 15
Test edge cases
Run three fictional unusual cases against the SOP
Edge-case simulation
KeepAsset you keep: Edge-case notes
Simplify and design
Days 16–20
- 16
Simplify before automating
Remove unnecessary steps and clarify dependencies
Process simplification
KeepAsset you keep: Simplified map
- 17
Separate deterministic rules
Identify steps a system can handle with explicit logic
Rules analysis
KeepAsset you keep: Rules inventory
- 18
Locate AI assistance
Mark where AI could draft, extract or flag for review
AI task matching
KeepAsset you keep: AI opportunity map
- 19
Define safe inputs
Set data classification, approvals and access limits
Data classification
KeepAsset you keep: Input policy draft
- 20
Design fallback
Decide what happens when AI or integration fails
Fallback planning
KeepAsset you keep: Fallback plan
Run a safe prototype
Days 21–25
- 21
Prototype one extraction
Try a fake sample document and compare fields manually
Document extraction
KeepAsset you keep: Extraction test
- 22
Prototype one summary
Summarize a fictional request into tasks and owners
Task extraction
KeepAsset you keep: Task-summary test
- 23
Prototype one checklist
Have GPT apply your SOP to fictional case details
Checklist application
KeepAsset you keep: Checklist test
- 24
Measure the baseline
Time the current process and count errors or rework
Baseline measurement
KeepAsset you keep: Baseline scorecard
- 25
Compare the prototype
Run the same sample and inspect time and accuracy
Comparative testing
KeepAsset you keep: Pilot evidence
Pilot and improve
Days 26–30
- 26
Write a control plan
Define logs, approvals, error handling and retention
Control documentation
KeepAsset you keep: Control plan
- 27
Train one teammate
Ask them to follow the SOP without verbal rescue
Training test
KeepAsset you keep: Training feedback
- 28
Revise the workflow
Fix ambiguity revealed by the teammate and tests
Revision
KeepAsset you keep: SOP revision
- 29
Build a workflow assistant
Package approved instructions with source and escalation rules
Reusable instructions
KeepAsset you keep: Assistant instructions
- 30
Choose the next test
Set owner, metric and stop condition for a month-long pilot
Pilot planning
KeepAsset you keep: 30-day pilot plan
Keep experimenting
Experiments 31–35
Optional
- 31
Map month-end
Apply the method to the close without skipping review
Close mapping
KeepAsset you keep: Month-end map
- 32
Check reconciliations
List evidence and exception rules for a fictional reconciliation
Evidence checklist
KeepAsset you keep: Reconciliation checklist
- 33
Test invoice vision
Extract fields from fake invoices and inspect misses
Vision test
KeepAsset you keep: Accuracy log
- 34
Audit spreadsheet handoffs
Trace data lineage in a safe sample workbook
Data lineage
KeepAsset you keep: Lineage map
- 35
Reduce email chasing
Prototype a single missing-information request
Request design
KeepAsset you keep: Request workflow
Keep experimenting
Experiments 36–40
Optional
- 36
Build a knowledge bot
Test approved SOP answers against source material
Retrieval testing
KeepAsset you keep: Bot test log
- 37
Run failure drills
Simulate missing files, wrong fields and a system outage
Failure simulation
KeepAsset you keep: Recovery checklist
- 38
Compare automations
Rank effort, savings and control requirements
Opportunity ranking
KeepAsset you keep: Automation backlog
- 39
Review access
Check who can view, edit and approve each step
Permission analysis
KeepAsset you keep: Access matrix
- 40
Calculate impact
Compare measured pilot outcomes with baseline
Impact evaluation
KeepAsset you keep: Pilot decision memo
Chain 05 of 09
People & Culture
Give your team room to grow.
Nothing yet — read down and watch it fill.
What you're holding
of 30 assets
30 connected · 10 stretch
Day zero
You own nothing yet. Read down the list and watch that change.
Clarify the role
Days 01–05
- 01
Describe one role
Explain what the person owns and why it matters
Role context
KeepAsset you keep: Role brief
- 02
Define outcomes
Set observable 30-, 60- and 90-day results
Outcome definition
KeepAsset you keep: Outcome scorecard
- 03
Separate skills from habits
List technical skills and ways of working
Competency mapping
KeepAsset you keep: Competency map
- 04
Find role ambiguity
Ask what a new hire might misunderstand
Ambiguity review
KeepAsset you keep: Ambiguity list
- 05
Validate with a manager
Review expectations against actual team capacity
Reality check
KeepAsset you keep: Validated role
Hire with care
Days 06–10
- 06
Draft a job ad
Write a specific ad grounded in real outcomes
Job-ad drafting
KeepAsset you keep: Job ad
- 07
Build interview questions
Ask about evidence of relevant past behaviour
Interview design
KeepAsset you keep: Interview guide
- 08
Make a scorecard
Rate candidates against job criteria, not impressions
Rubric design
KeepAsset you keep: Interview scorecard
- 09
Simulate an interview
Practise follow-up questions with a fictional candidate
Role-play
KeepAsset you keep: Practice transcript
- 10
Audit fairness
Review questions and criteria for unrelated or biased signals
Fairness review
KeepAsset you keep: Fairness checklist
Onboard and train
Days 11–15
- 11
Map first-week access
List people, tools and information a newcomer needs
Access mapping
KeepAsset you keep: First-week checklist
- 12
Build a 30-day plan
Sequence practice, feedback and increasing ownership
Learning design
KeepAsset you keep: Learning plan
- 13
Turn an SOP into training
Make a realistic safe exercise with expected evidence
Exercise design
KeepAsset you keep: Training exercise
- 14
Test understanding
Ask a learner to explain when to escalate
Knowledge testing
KeepAsset you keep: Knowledge check
- 15
Gather onboarding feedback
Ask what remained unclear after a mock run
Feedback synthesis
KeepAsset you keep: Onboarding revisions
Coach and develop
Days 16–20
- 16
Prepare a one-to-one
Create questions about blockers, learning and support
Coaching questions
KeepAsset you keep: 1:1 guide
- 17
Practise feedback
Role-play a specific issue without names or client details
Conversation practice
KeepAsset you keep: Feedback practice notes
- 18
Clarify a difficult message
Test how a teammate might interpret your draft
Interpretation test
KeepAsset you keep: Clearer message
- 19
Set development goals
Turn an aspiration into measurable practice
Goal setting
KeepAsset you keep: Development goals
- 20
Create a coaching loop
Set observation, feedback and follow-up dates
Coaching cadence
KeepAsset you keep: Coaching cadence
Delegate well
Days 21–25
- 21
Audit owner tasks
List work that still depends on the firm owner
Workload inventory
KeepAsset you keep: Owner-work inventory
- 22
Group delegation candidates
Sort by risk, repeatability and employee readiness
Delegation analysis
KeepAsset you keep: Delegation map
- 23
Write a delegation brief
Define result, authority, constraints and check-ins
Instruction writing
KeepAsset you keep: Delegation brief
- 24
Test the brief
Ask GPT what questions a teammate would still have
Clarity testing
KeepAsset you keep: Brief revision
- 25
Practise a handoff
Run a fictional task through the new delegation guide
Handoff simulation
KeepAsset you keep: Handoff rehearsal
Build the system
Days 26–30
- 26
Create team meeting rules
Keep meetings centred on decisions and blockers
Meeting design
KeepAsset you keep: Meeting charter
- 27
Build a recognition habit
Notice concrete work and its client or team effect
Specific recognition
KeepAsset you keep: Recognition prompts
- 28
Review anonymous themes
Cluster feedback while preserving uncertainty
Feedback clustering
KeepAsset you keep: Culture themes
- 29
Build a people assistant
Package approved role, training and coaching guidance
Reusable instructions
KeepAsset you keep: Assistant instructions
- 30
Choose a people pilot
Test one onboarding or delegation change over a month
Pilot design
KeepAsset you keep: 30-day people pilot
Keep experimenting
Experiments 31–35
Optional
- 31
Plan a career path
Describe observable growth between role levels
Career mapping
KeepAsset you keep: Career ladder draft
- 32
Cross-train a task
Identify one single-person dependency and teach a backup
Resilience planning
KeepAsset you keep: Cross-training plan
- 33
Rehearse conflict
Role-play both perspectives of a fictional disagreement
Conflict role-play
KeepAsset you keep: Conversation prep
- 34
Improve a meeting
Measure decisions made versus time spent
Meeting analysis
KeepAsset you keep: Meeting experiment
- 35
Review work capacity
Compare aggregate demand with staffing assumptions
Capacity analysis
KeepAsset you keep: Capacity questions
Keep experimenting
Experiments 36–40
Optional
- 36
Build a buddy system
Set goals and boundaries for peer support
Peer learning
KeepAsset you keep: Buddy guide
- 37
Test knowledge access
Ask staff to find an approved SOP without help
Information retrieval
KeepAsset you keep: Findability notes
- 38
Collect small wins
Capture improvements with real examples
Evidence capture
KeepAsset you keep: Win log
- 39
Revise the scorecard
Compare scorecard with observed role outcomes
Calibration
KeepAsset you keep: Scorecard update
- 40
Name what to stop
Remove one practice that adds friction without benefit
Stop-doing analysis
KeepAsset you keep: Stop-doing decision
Chain 06 of 09
Leadership & Strategy
Spend more time on the decisions that matter.
Nothing yet — read down and watch it fill.
What you're holding
of 30 assets
30 connected · 10 stretch
Day zero
You own nothing yet. Read down the list and watch that change.
Focus attention
Days 01–05
- 01
Tell GPT your priorities
Describe the firm, goals and constraints in your words
Context setting
KeepAsset you keep: Leadership brief
- 02
Empty your head
List open ideas, decisions and worries without sorting them
Idea capture
KeepAsset you keep: Decision inventory
- 03
Rank by consequence
Separate urgent, important and reversible decisions
Priority ranking
KeepAsset you keep: Priority map
- 04
Find missing facts
Name what evidence each major choice requires
Evidence mapping
KeepAsset you keep: Evidence list
- 05
Challenge the ranking
Ask a skeptical adviser what you may be ignoring
Strategic critique
KeepAsset you keep: Revised priorities
Unblock the owner
Days 06–10
- 06
Audit your calendar
Compare time spent with stated priorities
Time analysis
KeepAsset you keep: Time allocation map
- 07
Find the owner bottleneck
Identify decisions that unnecessarily wait on you
Bottleneck analysis
KeepAsset you keep: Bottleneck list
- 08
Choose what to delegate
Classify work as keep, delegate, automate or stop
Delegation mapping
KeepAsset you keep: Delegation map
- 09
Test a difficult decision
Make the strongest case against your preferred option
Counterargument
KeepAsset you keep: Decision challenge
- 10
Write a decision memo
Record options, assumptions, trade-offs and a deadline
Decision writing
KeepAsset you keep: Decision memo
Set direction
Days 11–15
- 11
Define the firm you want
Describe desired clients, work, team and owner role
Vision drafting
KeepAsset you keep: Future-firm brief
- 12
Identify constraints
Map what limits capacity, quality or growth
Constraint mapping
KeepAsset you keep: Constraint map
- 13
Build quarterly priorities
Turn the future brief into three concrete outcomes
Quarterly planning
KeepAsset you keep: 90-day priorities
- 14
Set leading indicators
Choose numbers that change before revenue changes
Indicator design
KeepAsset you keep: KPI draft
- 15
Build a one-page scorecard
Put owner, cadence and threshold beside each metric
Scorecard design
KeepAsset you keep: Management scorecard
Understand economics
Days 16–20
- 16
Inspect service economics
Use safe aggregate data to form profitability questions
Economics analysis
KeepAsset you keep: Economics questions
- 17
Challenge pricing logic
Compare scope, effort, value and delivery risk
Pricing critique
KeepAsset you keep: Pricing assumptions
- 18
Audit recurring meetings
Ask which need a decision, a shorter format or no meeting
Meeting audit
KeepAsset you keep: Meeting decisions
- 19
Examine client mix
Segment anonymized clients by need, fit and delivery effort
Client segmentation
KeepAsset you keep: Client segmentation
- 20
Choose a strategic bet
Select one opportunity to test, not five to announce
Bet selection
KeepAsset you keep: Bet brief
Stress-test choices
Days 21–25
- 21
Run a premortem
Assume the bet failed and list plausible causes
Premortem
KeepAsset you keep: Failure scenarios
- 22
Build scenarios
Model demand or capacity up and down with assumptions visible
Scenario modelling
KeepAsset you keep: Scenario sheet
- 23
Define triggers
Name evidence that would make you change course
Trigger setting
KeepAsset you keep: Decision triggers
- 24
Find second-order effects
Ask what a successful change might disrupt elsewhere
Systems thinking
KeepAsset you keep: Dependency map
- 25
Prepare a board review
Summarize decisions, risks and unresolved questions
Review synthesis
KeepAsset you keep: Review agenda
Run the experiment
Days 26–30
- 26
Audit your tech choices
Identify overlapping tools and weak integration logic
Stack audit
KeepAsset you keep: Stack questions
- 27
Check data readiness
Ask what must be documented before AI can safely help
Readiness analysis
KeepAsset you keep: Readiness gaps
- 28
Design from scratch
Ask what a new firm would refuse to recreate today
Future-state design
KeepAsset you keep: Future-state sketch
- 29
Build a strategy copilot
Package goals, constraints, metrics and challenge rules
Reusable instructions
KeepAsset you keep: Copilot instructions
- 30
Choose a 30-day test
Define hypothesis, metric, owner and stop condition
Experiment design
KeepAsset you keep: Strategic experiment
Keep experimenting
Experiments 31–35
Optional
- 31
Simulate a rival
Ask how a new competitor could win your clients
Competitive simulation
KeepAsset you keep: Competitive response ideas
- 32
Test capacity
Model one major new client with realistic staff constraints
Capacity modelling
KeepAsset you keep: Capacity decision
- 33
Map key-person risk
Locate knowledge or approvals held by one person
Key-person analysis
KeepAsset you keep: Resilience plan
- 34
Compare growth paths
Contrast organic, niche and acquisition hypotheses
Option comparison
KeepAsset you keep: Growth option memo
- 35
Review one service
Decide whether to improve, package or retire it
Service audit
KeepAsset you keep: Service decision
Keep experimenting
Experiments 36–40
Optional
- 36
Revisit mission
Ask teammates how the stated promise appears in daily work
Culture inquiry
KeepAsset you keep: Mission evidence
- 37
Model cash
Use safe aggregate assumptions for downside planning
Cash scenarios
KeepAsset you keep: Cash questions
- 38
Examine succession
Map roles and decisions that need future owners
Succession mapping
KeepAsset you keep: Succession questions
- 39
Review the scorecard
Check which metrics led to action and which were noise
Metric evaluation
KeepAsset you keep: Scorecard revision
- 40
Close the loop
Record what the test changed in your real strategy
Learning synthesis
KeepAsset you keep: Strategy update
Chain 07 of 09
Technology & Data
Make your tools work together.
Nothing yet — read down and watch it fill.
What you're holding
of 30 assets
30 connected · 10 stretch
Day zero
You own nothing yet. Read down the list and watch that change.
Know your stack
Days 01–05
- 01
List your tools
Name what each system does and who uses it
System inventory
KeepAsset you keep: Stack inventory
- 02
Map one client record
Trace information from inquiry through service delivery
Data tracing
KeepAsset you keep: Record journey
- 03
Find the source of truth
Mark which system owns each important field
Source-of-truth analysis
KeepAsset you keep: Data ownership map
- 04
Identify duplicate entry
Find where people retype or reconcile the same value
Duplicate detection
KeepAsset you keep: Duplicate-entry list
- 05
Check assumptions
Validate the map with the people doing the work
Stakeholder validation
KeepAsset you keep: Verified system map
Own your data
Days 06–10
- 06
Define key entities
Describe clients, contacts, jobs, documents and tasks
Entity modelling
KeepAsset you keep: Entity glossary
- 07
List required fields
Decide which fields are needed for one use case
Field definition
KeepAsset you keep: Field dictionary
- 08
Trace identifiers
Map how one client is matched across systems
Identity matching
KeepAsset you keep: Matching rules draft
- 09
Map permissions
Record who can read, change or approve each field
Permission mapping
KeepAsset you keep: Access matrix
- 10
Name sensitive data
Classify information before sharing it with any AI tool
Data classification
KeepAsset you keep: Data-classification draft
Design the handoff
Days 11–15
- 11
Diagram the handoffs
Show when a process moves between tools and people
Flow diagramming
KeepAsset you keep: Handoff diagram
- 12
Identify failure modes
Imagine missing records, duplicates and stale updates
Failure analysis
KeepAsset you keep: Failure list
- 13
Choose a small integration
Pick one repetitive, reversible handoff to improve
Use-case selection
KeepAsset you keep: Integration brief
- 14
Specify the trigger
Define precisely when the handoff should run
Event specification
KeepAsset you keep: Trigger rule
- 15
Specify the output
Describe fields, destination and expected acknowledgement
Output specification
KeepAsset you keep: Output contract
Control the integration
Days 16–20
- 16
Choose review controls
Decide when a person checks a result or exception
Control design
KeepAsset you keep: Control design
- 17
Write a test case
Make fake records for happy path and edge cases
Test-data creation
KeepAsset you keep: Test dataset
- 18
Test field mapping
Compare fake input with desired destination values
Mapping validation
KeepAsset you keep: Mapping test
- 19
Test duplicates
Try repeat events and conflicting identifiers
Idempotency testing
KeepAsset you keep: Duplicate-handling notes
- 20
Test recovery
Simulate timeout and error; define safe retry behaviour
Recovery design
KeepAsset you keep: Recovery rules
Pilot with evidence
Days 21–25
- 21
Document one flow
Write plain-language steps, owners and dependencies
SOP documentation
KeepAsset you keep: Flow SOP
- 22
Create an exception queue
Define what gets flagged for human action
Exception routing
KeepAsset you keep: Exception specification
- 23
Add an audit trail
Record what happened, when and under whose authority
Audit-log design
KeepAsset you keep: Log requirements
- 24
Measure baseline
Count current time, errors and rework for the handoff
Baseline measurement
KeepAsset you keep: Baseline metrics
- 25
Run a sample pilot
Compare the improved flow on safe test cases
Pilot comparison
KeepAsset you keep: Pilot results
Roll out safely
Days 26–30
- 26
Evaluate AI access
Decide whether approved AI should see any part of the data
Access evaluation
KeepAsset you keep: AI access decision
- 27
Design a retrieval answer
Require approved sources and a visible citation or source
Grounded retrieval
KeepAsset you keep: Retrieval design
- 28
Test hallucinations
Ask unsupported questions and observe whether the assistant abstains
Abstention testing
KeepAsset you keep: Abstention test
- 29
Build a tech copilot brief
Combine system map, glossary and escalation rules
Reusable instructions
KeepAsset you keep: Tech assistant instructions
- 30
Choose a 30-day rollout
Specify owner, rollback, metric and go/no-go rule
Rollout planning
KeepAsset you keep: Rollout plan
Keep experimenting
Experiments 31–35
Optional
- 31
Audit a webhook
Check event sequence, duplicate delivery and authentication
Webhook analysis
KeepAsset you keep: Webhook checklist
- 32
Test a dashboard
Choose three decisions its measures must support
Dashboard design
KeepAsset you keep: Dashboard brief
- 33
Compare vendor claims
Translate feature claims into controlled test cases
Vendor testing
KeepAsset you keep: Vendor test plan
- 34
Review retention
Map how long records need to live and who decides
Retention analysis
KeepAsset you keep: Retention questions
- 35
Find orphaned data
Spot fields with no owner or clear use
Data cleanup
KeepAsset you keep: Cleanup backlog
Keep experimenting
Experiments 36–40
Optional
- 36
Build a permission review
Schedule periodic access checks
Access review
KeepAsset you keep: Review cadence
- 37
Model a system outage
Plan critical work while one tool is unavailable
Continuity planning
KeepAsset you keep: Continuity notes
- 38
Track data quality
Define completeness, freshness and accuracy checks
Data-quality design
KeepAsset you keep: Quality scorecard
- 39
Measure adoption
Ask users where the new flow still causes friction
Adoption analysis
KeepAsset you keep: Adoption feedback
- 40
Retire a duplicate tool
Compare actual use, cost and migration risk
Retirement analysis
KeepAsset you keep: Retirement memo
Chain 08 of 09
Advisory & Insights
Turn numbers into better questions.
Nothing yet — read down and watch it fill.
What you're holding
of 30 assets
30 connected · 10 stretch
Day zero
You own nothing yet. Read down the list and watch that change.
Define the question
Days 01–05
- 01
Describe your advisory promise
Define which decisions you help clients make
Service framing
KeepAsset you keep: Advisory brief
- 02
Choose an ideal scenario
Pick one fictional client decision worth discussing
Scenario choice
KeepAsset you keep: Scenario brief
- 03
Map the facts needed
Ask which financial and non-financial inputs matter
Information mapping
KeepAsset you keep: Information checklist
- 04
Separate fact from guess
Label supplied facts, estimates and unknowns
Assumption labelling
KeepAsset you keep: Assumption ledger
- 05
Ask what could change
List questions whose answers would alter advice
Clarifying questions
KeepAsset you keep: Clarification questions
Find useful signals
Days 06–10
- 06
Identify useful signals
Choose numbers that might signal opportunity or risk
Signal selection
KeepAsset you keep: Signal list
- 07
Build a review checklist
Turn the signals into a repeatable human review
Review design
KeepAsset you keep: Insight checklist
- 08
Practice variance questions
Use fictional comparisons to find what needs inquiry
Variance inquiry
KeepAsset you keep: Variance question bank
- 09
Explore cash timing
Use fake cash flows to ask about upcoming pressure
Cash-flow inquiry
KeepAsset you keep: Cash conversation guide
- 10
Check for misleading averages
Challenge a plausible but shallow interpretation
Statistical critique
KeepAsset you keep: Interpretation cautions
Start the conversation
Days 11–15
- 11
Role-play a client
Ask how the fictional client describes their goals
Client role-play
KeepAsset you keep: Client goals brief
- 12
Translate a metric
Explain one ratio in everyday operational terms
Plain-language explanation
KeepAsset you keep: Plain-language explainer
- 13
Connect number to choice
Identify decisions the metric might inform
Decision linkage
KeepAsset you keep: Decision map
- 14
Ask a better question
Practise asking what changed before proposing a fix
Discovery questioning
KeepAsset you keep: Discovery prompts
- 15
Create a meeting opener
Start with the client's priorities, not your dashboard
Conversation design
KeepAsset you keep: Conversation opener
Explore scenarios
Days 16–20
- 16
Model a base case
List assumptions and an illustrative fictional outcome
Base-case modelling
KeepAsset you keep: Base-case model brief
- 17
Add upside and downside
Vary one driver at a time and note dependencies
Scenario comparison
KeepAsset you keep: Scenario comparison
- 18
Test sensitivity
Find assumptions that most affect the conclusion
Sensitivity testing
KeepAsset you keep: Sensitivity questions
- 19
Mark professional boundaries
Identify tax, legal or specialist issues requiring review
Boundary recognition
KeepAsset you keep: Referral and review list
- 20
Challenge the recommendation
Ask a skeptic what evidence is missing
Recommendation critique
KeepAsset you keep: Recommendation critique
Deliver the insight
Days 21–25
- 21
Design a one-page insight
Show situation, evidence, options and unknowns
Insight structure
KeepAsset you keep: Insight page template
- 22
Draft a meeting agenda
Move from findings to implications and decisions
Agenda design
KeepAsset you keep: Advisory agenda
- 23
Role-play objections
Practise responding to a client's challenge to assumptions
Objection practice
KeepAsset you keep: Practice notes
- 24
Capture decisions
Separate client choice, adviser action and follow-up date
Decision recording
KeepAsset you keep: Decision record
- 25
Create a follow-up note
Summarize what was agreed without overstating certainty
Follow-up synthesis
KeepAsset you keep: Follow-up template
Pilot the offer
Days 26–30
- 26
Define an advisory offer
Set scope, cadence, inputs and excluded work
Offer design
KeepAsset you keep: Offer outline
- 27
Build intake rules
Identify which clients and questions suit the pilot
Pilot qualification
KeepAsset you keep: Pilot eligibility
- 28
Make a quality review
Require source checks, math checks and human sign-off
Quality review
KeepAsset you keep: Review checklist
- 29
Build an insights copilot
Package approved templates and refusal/escalation rules
Reusable instructions
KeepAsset you keep: Copilot instructions
- 30
Pilot one conversation
Use a fictional or consented case and record outcomes
Pilot evaluation
KeepAsset you keep: 30-day advisory pilot
Keep experimenting
Experiments 31–35
Optional
- 31
Compare client segments
Ask which advisory needs differ by business stage
Segmentation
KeepAsset you keep: Segment opportunity map
- 32
Model hiring
Use fictional cost and capacity assumptions
Hiring modelling
KeepAsset you keep: Hiring decision prompts
- 33
Model pricing
Explore margin and volume scenarios with visible assumptions
Pricing modelling
KeepAsset you keep: Pricing scenarios
- 34
Model investment
Ask what evidence an owner needs before spending
Investment analysis
KeepAsset you keep: Investment question list
- 35
Build a risk register
Name risks, probability assumptions and owners
Risk mapping
KeepAsset you keep: Risk register draft
Keep experimenting
Experiments 36–40
Optional
- 36
Test an executive summary
Ask a lay reader what they understand or doubt
Comprehension test
KeepAsset you keep: Summary revisions
- 37
Review a failed forecast
Compare assumptions with fictional actuals
Forecast review
KeepAsset you keep: Forecast lesson log
- 38
Build an annual review
Sequence history, future choices and action owners
Annual review design
KeepAsset you keep: Review agenda
- 39
Refine your offer
Use actual pilot feedback to adjust the service
Offer iteration
KeepAsset you keep: Offer revision
- 40
Measure usefulness
Ask whether a conversation changed a client decision
Outcome measurement
KeepAsset you keep: Outcome evidence
Chain 09 of 09
Innovation & Experiments
Put curiosity to work.
Nothing yet — read down and watch it fill.
What you're holding
of 30 assets
30 connected · 10 stretch
Day zero
You own nothing yet. Read down the list and watch that change.
Find the problem
Days 01–05
- 01
Name one annoyance
Describe a recurring problem without naming a solution
Problem framing
KeepAsset you keep: Problem statement
- 02
Find the real user
Identify who suffers and how often
User discovery
KeepAsset you keep: User profile
- 03
Describe current workarounds
Ask what people do today instead of solving it
Workaround mapping
KeepAsset you keep: Workaround map
- 04
Measure the baseline
Estimate time, error or delay with actual observations
Baseline measurement
KeepAsset you keep: Baseline log
- 05
Challenge the problem
Ask whether the annoyance is a symptom of something else
Root-cause challenge
KeepAsset you keep: Revised problem
Choose a bet
Days 06–10
- 06
Generate three approaches
Include a process fix, tool fix and stop-doing option
Option generation
KeepAsset you keep: Option set
- 07
Choose a small bet
Rank approaches by reversibility, effort and potential value
Reversible choice
KeepAsset you keep: Experiment choice
- 08
State a hypothesis
Write what will change for whom and by how much
Hypothesis writing
KeepAsset you keep: Testable hypothesis
- 09
Set a success metric
Decide what result would justify continuing
Metric design
KeepAsset you keep: Success threshold
- 10
Set a stop condition
Name evidence or risk that should end the test
Stop-rule setting
KeepAsset you keep: Stop rule
Prototype
Days 11–15
- 11
Sketch a prototype
Define the smallest version that tests the hypothesis
Prototype scoping
KeepAsset you keep: Prototype brief
- 12
Build fake inputs
Create realistic sample cases without client secrets
Synthetic test data
KeepAsset you keep: Sample dataset
- 13
Run a first attempt
Use GPT on one sample and save the unedited output
First-run capture
KeepAsset you keep: First-run evidence
- 14
Record failures
Classify what was wrong, missing or misleading
Failure classification
KeepAsset you keep: Failure log
- 15
Revise the prompt or workflow
Change one variable and rerun the same sample
Controlled iteration
KeepAsset you keep: Iteration comparison
Decide with evidence
Days 16–20
- 16
Ask a colleague to test
Let them try without step-by-step coaching
Usability testing
KeepAsset you keep: Usability feedback
- 17
Compare with current work
Measure whether the new method really improved anything
Before-after analysis
KeepAsset you keep: Before-after notes
- 18
Identify new risks
Ask what could fail if the experiment were used routinely
Risk discovery
KeepAsset you keep: Risk list
- 19
Add human checkpoints
Specify where professional judgment remains essential
Human review design
KeepAsset you keep: Review design
- 20
Decide whether to continue
Compare evidence with the original threshold
Decision analysis
KeepAsset you keep: Go/no-go memo
Prepare a pilot
Days 21–25
- 21
Turn success into steps
Write a small SOP a colleague could follow
SOP drafting
KeepAsset you keep: Pilot SOP
- 22
Estimate adoption effort
List training, permissions and support needed
Adoption mapping
KeepAsset you keep: Adoption checklist
- 23
Choose a pilot group
Select a limited setting and an accountable owner
Pilot scoping
KeepAsset you keep: Pilot scope
- 24
Plan feedback capture
Ask what users should report after each use
Feedback design
KeepAsset you keep: Feedback form
- 25
Run the small pilot
Collect real performance measures, not just enthusiasm
Pilot measurement
KeepAsset you keep: Pilot log
Learn and improve
Days 26–30
- 26
Analyse patterns
Compare wins and misses across pilot cases
Pattern analysis
KeepAsset you keep: Pattern review
- 27
Change one thing
Revise the highest-impact weakness
Single-variable change
KeepAsset you keep: Prototype version two
- 28
Retest edge cases
Check the revision against cases that broke earlier
Regression testing
KeepAsset you keep: Regression notes
- 29
Package the learning
Show evidence, unresolved issues and next recommendation
Evidence synthesis
KeepAsset you keep: Experiment brief
- 30
Choose the next 30 days
Scale, extend, redesign or stop based on evidence
Next-bet selection
KeepAsset you keep: Next-test plan
Keep experimenting
Experiments 31–35
Optional
- 31
Test another user
See whether the idea transfers across roles
Transfer test
KeepAsset you keep: Transferability notes
- 32
Compare a rival tool
Run identical samples and scoring rules
Benchmarking
KeepAsset you keep: Tool comparison
- 33
Check economics
Compare saved effort with running and review costs
Cost-benefit analysis
KeepAsset you keep: Cost-benefit draft
- 34
Plan rollback
Define how to return to the old process safely
Rollback design
KeepAsset you keep: Rollback plan
- 35
Test worst cases
Simulate errors with high consequence and low visibility
Stress testing
KeepAsset you keep: Stress-test log
Keep experimenting
Experiments 36–40
Optional
- 36
Invite criticism
Ask a skeptical team member to find blind spots
Adversarial critique
KeepAsset you keep: Critique notes
- 37
Document a failure
Share what you learned without polishing away the miss
Failure storytelling
KeepAsset you keep: Failure story
- 38
Reuse a success
Apply the pattern to a second small process
Pattern reuse
KeepAsset you keep: Transfer pilot
- 39
Build an idea queue
Score new ideas against value and testability
Backlog scoring
KeepAsset you keep: Experiment backlog
- 40
Share a VPD reel
Tell the honest arc from attempt to evidence to next step
Video storytelling
KeepAsset you keep: VPD story outline
Then let one of the nine come to you.
This deals a single day out of the 360. Read it, and if it sounds like a Tuesday evening in your firm, that is your chain — open it and start at day one instead. Press it again as many times as you like.
#GPTEPDDo it tonight, then say what happened.
One day, out of 360
Teach GPT about your firm
Describe services, clients, location and approach; ask what context is missing
- The AI skill
- Context setting
- What you keep
- Firm brief
Post it, especially when it didn't work.
Almost everything published about AI is the highlight reel, which is why so many people quietly assume they are the only one getting mediocre results. You are not. Post the real attempt — the bit that worked, the bit that didn't, and what you would try instead — and the next person gets to skip the dead end you just walked into.
The shape of a good one
Day 01 of #GPTEPD — [what you tried] What I tried: What worked: What didn't: What surprised me: What I'll try next:
Then put it up with
#GPTEPDAnywhere you already post. Most of this crowd is on LinkedIn, which is also where the live sessions run. On Instagram those five lines won't paste — talk them through to camera instead, which is what VPD was in the first place.
Post the experiment, never the file. Nothing that identifies a client goes in a public post — and if you had to anonymise it to run the experiment, keep it anonymised when you write it up.
Five lines is a post. You already answered them.
These are the same questions as the field note in the workbooks, which is deliberate — write them down at the end of an experiment and the post is already finished.
- 01What I triedThe experiment, and what you gave it to work with.
- 02What workedBe specific. “Quite good” is not a finding.
- 03What didn'tThe useful half.
- 04What surprised meUsually the thing worth telling someone.
- 05What I'll try nextTomorrow's experiment, in one line.
Two rules, and they matter more here than most places
- 01Keep client information fictional or anonymised, unless your firm has approved both the AI environment and this particular use of that data.
- 02Check every output before it touches professional advice, a decision or a record. The model is a draft, never a reviewer.
Stop watching. Start trying.
Nobody needs to look polished on day one.
Pick a chain. Do day one. Post what happened. That is the whole commitment — there is nothing to buy, nothing to sign up for, and no list you end up on.
#GPTEPDEvery attempt anyone posts is under here.