Eighteen prompts. Paste them into your AI. It does the work.

AI USE CASE INVENTORY — RUN THIS PROMPT IN YOUR AI TOOL OF CHOICE
Zero → Frontier — free, ungated, no attribution required

HOW TO USE THIS: Copy everything below the line into ChatGPT, Claude,
Gemini, or any chat AI. Fill in the bracketed sections with your own
information first. The AI will generate a completed inventory table —
you don't fill this out by hand.

---

I want you to build me an AI Use Case Inventory for my work. Here is
my situation:

MY ROLE / BUSINESS: [describe what you do, e.g. "solo consultant,
marketing services for local businesses"]

MY RECURRING TASKS (list as many as you can, rough is fine):
[e.g. "weekly client status emails, monthly research on industry
trends, meeting notes 3x/week, drafting social posts, invoicing"]

For EACH task I listed, build a table with these columns:
1. Task name
2. How often it happens
3. What makes it slow (drafting bottleneck vs. research bottleneck
   vs. decision bottleneck — explain the difference if it's unclear
   which one applies)
4. Confidentiality level (Public / Internal / Confidential / Highly
   sensitive) — flag anything above "Internal" as needing a data-
   handling check before using AI on it
5. Reversibility (fully reversible draft vs. hard-to-reverse action
   like a sent email or filed document) — flag hard-to-reverse tasks
   as always needing human review before acting
6. Suggested AI approach (which kind of tool fits: general chat
   assistant, a research tool, a meeting-notes tool, a coding tool)
7. Worth trying now? (yes / no / not yet) with a one-line reason

After the table, give me:
- The single task I should try AI on FIRST, and why it's the best
  starting point (highest time savings + lowest risk)
- Any task where you'd recommend NOT using AI at all, and why

Be honest and specific — don't rate everything "yes, worth trying."
Some of my tasks may genuinely not be good AI candidates, and I want
you to tell me which ones and why.

---

Free to copy, adapt, and reuse. Part of the Zero → Frontier project.

AI TOOL EVALUATION — RUN THIS PROMPT IN YOUR AI TOOL OF CHOICE
Zero → Frontier — free, ungated, no attribution required

HOW TO USE THIS: Find the pricing page and privacy policy of the AI
tool you're considering (copy the URL, or paste the actual text of
those pages if the AI can't browse). Paste them into the prompt
below along with your own situation. The AI will produce a structured
evaluation — you don't fill out a blank form yourself.

---

I'm evaluating whether to adopt a new AI tool. Here is the
information:

TOOL NAME: [name]
WHAT I'D USE IT FOR: [your specific recurring task from your Use
Case Inventory, if you have one]
PRICING PAGE CONTENT: [paste the text, or the URL if you can browse]
PRIVACY POLICY CONTENT: [paste the text, or the URL if you can browse]

Evaluate this tool against these six criteria, and for each one,
tell me what you found and flag anything unclear or not publicly
stated (don't guess — say "not stated" if the page doesn't say):

1. FIT — does this tool's actual features match my stated task, or
   is it overkill / underkill for what I need?
2. DATA HANDLING — is my input used to train the tool's models by
   default? Is there an opt-out, and is it self-serve or does it
   require contacting support?
3. COST STRUCTURE — what exactly is included free, and what's the
   hard cap (messages/day, minutes/month, credits)? Is there a
   surprise-overage risk?
4. RELIABILITY — does the tool disclose known limitations or
   failure modes anywhere on its own pages?
5. EXIT PLAN — can I export my data if I stop using this? Is there
   vendor lock-in risk (e.g., proprietary hosting, no data export)?
6. HUMAN OVERSIGHT FIT — does adopting this tool require me to add
   any new review step before I trust its output for real use?

End with a one-paragraph verdict: adopt, adopt with a specific
caution, or don't adopt yet — and the single biggest reason why.

---

Free to copy, adapt, and reuse. Part of the Zero → Frontier project.

AI IMPLEMENTATION ROADMAP — RUN THIS PROMPT IN YOUR AI TOOL OF CHOICE
Zero → Frontier — free, ungated, no attribution required

HOW TO USE THIS: Run this after you've completed the AI Use Case
Inventory Prompt. Paste in your ranked list of tasks, and the AI
sequences them into a realistic rollout plan instead of trying
everything at once.

---

I have this AI Use Case Inventory ranked from my earlier work:
[paste your inventory table, or just list your tasks with rough
priority]

TEAM SIZE: [number of people]
TIME AVAILABLE PER WEEK FOR THIS: [be realistic, e.g. "2-3 hours"]

Build me a phased rollout roadmap:

PHASE 1 (first 2-4 weeks): The single highest-value, lowest-risk use
case to start with. Explain why this one goes first.

PHASE 2 (next 4-8 weeks): 2-3 more use cases to add once Phase 1 is
working smoothly — sequenced by a mix of value and ease of adoption.

PHASE 3 (beyond that): Anything higher-risk or higher-effort (e.g.
tasks touching confidential data, or requiring a governance policy
first) that should wait until the basics are solid.

For each phase, tell me:
- What "done" looks like before moving to the next phase
- The one thing most likely to derail this phase, and how to avoid it

Do not recommend trying to implement everything simultaneously — the
whole point of this roadmap is sequencing, not a wish list.

---

Free to copy, adapt, and reuse. Part of the Zero → Frontier project.

AI VENDOR REVIEW — RUN THIS PROMPT IN YOUR AI TOOL OF CHOICE
Zero → Frontier — free, ungated, no attribution required

HOW TO USE THIS: Use this when comparing two or more AI vendors/tools
side by side, not just evaluating one in isolation (that's the AI
Tool Evaluation Prompt). Paste in details for each candidate.

---

I'm choosing between multiple AI vendors for [use case]. Here are the
candidates:

VENDOR A: [name] — [paste pricing/privacy page content or URL]
VENDOR B: [name] — [paste pricing/privacy page content or URL]
VENDOR C (optional): [name] — [paste pricing/privacy page content or URL]

Build me a side-by-side comparison table with these rows:
1. Core capability fit for [use case]
2. Free tier or trial terms
3. Ongoing cost at realistic usage volume: [describe your expected volume]
4. Data handling / training-data default
5. Data export / exit difficulty if I later switch vendors
6. Anything each vendor discloses about known limitations or risks

Then give me a recommendation with the single deciding factor — not
just "it depends," but which one and why, given what I told you about
my use case and volume. If the vendors are genuinely too close to
call on the facts provided, say so explicitly rather than picking one
arbitrarily.

---

Free to copy, adapt, and reuse. Part of the Zero → Frontier project.

MEETING NOTES & FOLLOW-UP — RUN THIS PROMPT IN YOUR AI TOOL OF CHOICE
Zero → Frontier — free, ungated, no attribution required

HOW TO USE THIS: Paste your raw meeting transcript or rough notes
where indicated. Works in any chat AI (Claude, ChatGPT, Gemini) or
a dedicated meeting-notes tool — see the Tools > Meetings & Knowledge
page for a comparison of those.

---

Here are my raw notes/transcript from today's meeting:
[paste your unedited notes or transcript here]

Turn this into three sections:

1. DECISIONS MADE — only things the group actually agreed on, not
   things that were merely discussed.

2. ACTION ITEMS — each with an owner and a rough deadline if one was
   mentioned. If an item's owner is unclear from the notes, list it
   under "UNASSIGNED" instead of guessing who's responsible.

3. OPEN QUESTIONS — anything raised but not resolved, so it doesn't
   silently disappear.

Then draft a short follow-up message (3-5 sentences) I could send to
the group summarizing the above, in a [formal / casual — pick one]
tone.

Flag anything in my notes that seems ambiguous or contradictory
rather than silently resolving it one way.

---

Free to copy, adapt, and reuse. Part of the Zero → Frontier project.

SOP DRAFTING — RUN THIS PROMPT IN YOUR AI TOOL OF CHOICE
Zero → Frontier — free, ungated, no attribution required

HOW TO USE THIS: Describe your process in plain, messy language — the
AI will structure it into a proper Standard Operating Procedure. You
don't need to already know SOP format.

---

I need a Standard Operating Procedure (SOP) written for this process:

PROCESS NAME: [e.g. "Onboarding a new client"]
WHO DOES THIS: [role/person]
HOW OFTEN: [frequency]
ROUGH DESCRIPTION OF WHAT HAPPENS, IN ORDER (messy is fine, just get
it all down):
[describe every step you can think of, in whatever order they come
to mind — the AI will organize it]

TOOLS/SYSTEMS INVOLVED: [list any software, accounts, or tools used]
THINGS THAT COMMONLY GO WRONG: [list any known failure points]

Turn this into a structured SOP with these sections:
1. Purpose — one sentence on why this process exists
2. Who's responsible
3. Step-by-step instructions, numbered, in the correct order (fix
   any ordering issues from my rough description)
4. Common mistakes and how to avoid each one
5. What "done" looks like — how to confirm the process was completed
   correctly
6. Related processes or SOPs this connects to, if any

If my description has a gap or an unclear step, ask me a clarifying
question rather than inventing a plausible-sounding step.

---

Free to copy, adapt, and reuse. Part of the Zero → Frontier project.

INTERNAL FAQ BUILDER — RUN THIS PROMPT IN YOUR AI TOOL OF CHOICE
Zero → Frontier — free, ungated, no attribution required

HOW TO USE THIS: Paste in the recurring questions your team actually
asks (from Slack, email, or memory) and the AI structures them into
a proper internal FAQ document.

---

I want to build an internal FAQ for [team/topic, e.g. "our expense
reimbursement process" or "how we handle client requests"].

HERE ARE THE QUESTIONS PEOPLE ACTUALLY ASK, roughly, with whatever
answer I know off the top of my head (some may be incomplete —
flag those):
[list as many real recurring questions as you can, with rough answers]

Turn this into a structured FAQ:
1. Group related questions together under clear headings.
2. Write each answer clearly and completely — if my rough answer was
   incomplete, flag exactly what's missing rather than inventing the
   missing part.
3. Order questions by how often they likely come up, most common first.
4. Add a "last updated" placeholder and a note on who to ask if the
   FAQ doesn't cover a new question.

Flag any question where you don't have enough information from me to
give a confident answer, rather than guessing.

---

Free to copy, adapt, and reuse. Part of the Zero → Frontier project.

PROJECT UPDATE — RUN THIS PROMPT IN YOUR AI TOOL OF CHOICE
Zero → Frontier — free, ungated, no attribution required

HOW TO USE THIS: Paste in your raw project notes/status and get a
clean status update in the format your stakeholders actually want —
without spending the usual 20+ minutes formatting it yourself.

---

I need a project status update. Here's the raw information:

PROJECT: [name]
WHAT'S HAPPENED SINCE THE LAST UPDATE: [messy notes are fine]
WHAT'S BLOCKING PROGRESS, IF ANYTHING: [list it honestly]
WHAT'S NEXT: [rough plan]
AUDIENCE: [e.g. "my manager," "the client," "the whole team"]

Turn this into a status update with:
1. Overall status: On track / At risk / Blocked (pick the honest one
   based on what I told you — don't default to "on track" if what I
   described sounds risky)
2. What's done since last update (2-4 bullet points, specific not vague)
3. What's blocking, if anything, and what's needed to unblock it
4. What's next
5. One sentence a busy reader could read alone and understand the
   whole status from

Match the tone to my audience — more detail and context for the team,
more concise and outcome-focused for a client or executive.

---

Free to copy, adapt, and reuse. Part of the Zero → Frontier project.

CONTENT REPURPOSING — RUN THIS PROMPT IN YOUR AI TOOL OF CHOICE
Zero → Frontier — free, ungated, no attribution required

HOW TO USE THIS: Paste one piece of content you've already made
(a blog post, a talk transcript, a long email). The AI turns it into
several other formats, so you're not starting from scratch each time.

---

Here is a piece of content I already created:
[paste your blog post, transcript, article, or long-form piece]

Repurpose it into the following, each as its own clearly labeled
section:

1. FIVE SOCIAL POSTS (for [platform, e.g. LinkedIn/Twitter/Instagram])
   — each one a different angle or takeaway from the piece, not the
   same summary repeated five times.

2. ONE EMAIL NEWSLETTER VERSION — shorter, with a single clear call
   to action at the end.

3. TEN PULL QUOTES — the most quotable, specific sentences from the
   original (not paraphrased — pull the actual sentences).

4. ONE SHORT-FORM VIDEO/REEL SCRIPT (60-90 seconds) — a hook in the
   first line, the core idea, and a closing line.

Constraints:
- Keep every fact, number, and claim exactly as it appears in the
  original — do not introduce new statistics or claims.
- Match my original tone; don't make it more generic or "salesy"
  than the source material.
- Flag if the original doesn't have enough substance for any of the
  four formats above, rather than padding it with filler.

---

Free to copy, adapt, and reuse. Part of the Zero → Frontier project.

NEWSLETTER DRAFTING — RUN THIS PROMPT IN YOUR AI TOOL OF CHOICE
Zero → Frontier — free, ungated, no attribution required

HOW TO USE THIS: Paste in your raw update items or a longer piece of
content, and get a structured newsletter draft — not a generic
template, one built from what you actually have to say this issue.

---

I need to draft this issue of my newsletter. Here's what I have to
share:
[list your raw updates, links, or paste a longer piece of content
to draw from]

AUDIENCE: [who reads this newsletter]
TYPICAL LENGTH: [short/scannable vs. longer-form]
TONE: [your usual voice — casual, professional, personal]

Draft the issue with:
1. A subject line — give me 3 options, each testing a different
   angle (curiosity, benefit-forward, direct).
2. An opening line that isn't generic ("Hope this finds you well" —
   avoid anything that could be copy-pasted into any newsletter).
3. The body, organized by the items I gave you, each with a clear
   one-line takeaway before the detail.
4. One clear call to action at the end — not three competing ones.

Keep every fact and number exactly as I stated it — don't add
statistics or claims I didn't give you.

---

Free to copy, adapt, and reuse. Part of the Zero → Frontier project.

LEAD MAGNET PLANNER — RUN THIS PROMPT IN YOUR AI TOOL OF CHOICE
Zero → Frontier — free, ungated, no attribution required

HOW TO USE THIS: Describe your audience and expertise — the AI
proposes concrete lead magnet ideas and outlines the strongest one,
rather than you starting from a blank page.

---

I want to create a lead magnet (a free resource that attracts my
ideal audience in exchange for their attention/email). Here's my
context:

WHAT I DO: [your business/expertise]
MY AUDIENCE: [who you're trying to attract, be specific]
THEIR BIGGEST PAIN POINT RIGHT NOW: [what they struggle with]
FORMAT PREFERENCES: [e.g. "I can write, but can't design," "I could
record a short video," — be honest about what you can actually produce]

Give me:
1. FIVE lead magnet concepts specific to my audience's actual pain
   point — not generic ideas like "ultimate guide to X."
2. For each, one sentence on why it would genuinely appeal to my
   specific audience, not audiences in general.
3. Pick the single strongest concept given my format constraints, and
   outline it in full: a working title, 5-7 section headers, and
   what goes in each section.
4. One sentence describing the follow-up next step this lead magnet
   should point toward (not a hard sell — a natural next step).

Be honest if any of my five ideas is weak — don't pad the list with
filler options just to hit five.

---

Free to copy, adapt, and reuse. Part of the Zero → Frontier project.

CLIENT ONBOARDING & PROPOSAL DRAFTING — RUN THIS PROMPT
Zero → Frontier — free, ungated, no attribution required

HOW TO USE THIS: Fill in your real details about the client/project.
The AI drafts either an onboarding sequence or a proposal outline —
pick which one you need, or ask for both.

---

I need help drafting client-facing material. Here's the context:

WHAT I DO: [your business/service]
CLIENT/PROJECT: [who this is for, and what they need]
KEY DETAILS: [scope, timeline, price if relevant, anything unusual
about this client]
TONE: [formal / warm-professional / casual — pick one]

I need ONE of the following (tell me which, or ask for both):

OPTION A — A 3-EMAIL ONBOARDING SEQUENCE:
1. Welcome email — sets expectations, confirms next steps
2. Information-request email — asks for what you need from them to
   start, as a clear checklist, not a vague ask
3. Kickoff-confirmation email — confirms start date and what happens
   next

OPTION B — A PROPOSAL OUTLINE:
1. Problem statement (in their words, based on what I told you)
2. What's included / scope
3. What's explicitly NOT included (to prevent scope creep later)
4. Timeline with milestones
5. Investment/pricing section (leave exact numbers as placeholders
   if I haven't given them)
6. Next steps / how to say yes

Constraints:
- Don't invent scope, pricing, or timeline details I haven't given
  you — use placeholders like [PRICE] if I left something out.
- Keep it specific to what I told you, not generic template language.

---

Free to copy, adapt, and reuse. Part of the Zero → Frontier project.

RESEARCH BRIEF FOR A CLIENT — RUN THIS PROMPT IN YOUR AI TOOL OF CHOICE
Zero → Frontier — free, ungated, no attribution required

HOW TO USE THIS: After you've done real research (see the Research
Workflow playbook), paste your verified notes and sources in here —
this turns your research into a polished, client-ready brief, it
does not do the research or verification for you.

---

I've done research on [topic] for a client. Here are my verified
notes and sources:
[paste your own research notes, with sources — do not skip
verification and paste an unverified AI summary here]

CLIENT CONTEXT: [who they are, why they need this]
FORMAT NEEDED: [e.g. "one-page brief," "detailed report with sections"]

Turn my notes into a client-ready research brief:
1. Executive summary — 3-4 sentences, the key takeaway up front.
2. Key findings, organized by theme, each citing which source it
   came from (use my sources, don't introduce new claims).
3. What this means for the client specifically, given their context.
4. Open questions or areas needing further research, stated honestly.

Do not add any claim, statistic, or source that I didn't provide —
if you think something is missing, flag it as a gap for me to
research, not something to fill in yourself.

---

Free to copy, adapt, and reuse. Part of the Zero → Frontier project.

DELIVERABLE REVIEW CHECKLIST — RUN THIS PROMPT BEFORE SENDING CLIENT WORK
Zero → Frontier — free, ungated, no attribution required

HOW TO USE THIS: Paste in a finished client deliverable (a report,
a proposal, a design brief) before you send it — this is a final
quality pass, run separately from the Human Review Self-Check prompt,
which focuses on AI-specific risks rather than deliverable quality.

---

I'm about to send this deliverable to a client. Review it critically
before I do:

DELIVERABLE: [paste the full deliverable]
WHAT THE CLIENT ASKED FOR: [the original brief or request]
CLIENT CONTEXT: [anything specific about this client's expectations,
sensitivities, or prior feedback]

Check it against:
1. Does it actually answer/deliver what the client asked for — not
   a well-written document that drifted from the original ask?
2. Are there any internal inconsistencies — numbers, dates, or names
   that don't match elsewhere in the document?
3. Is the tone appropriate for this specific client, based on what
   I told you about them?
4. Is anything overpromised — a guarantee, timeline, or outcome
   stated more confidently than it should be?
5. Is there anything that reads as generic or template-like that
   should be more specific to this client?
6. One thing that would make this deliverable noticeably stronger
   before it goes out.

Be specific about what to fix — not just "looks good" or vague praise.

---

Free to copy, adapt, and reuse. Part of the Zero → Frontier project.

AI USE POLICY STARTER — RUN THIS PROMPT IN YOUR AI TOOL OF CHOICE
Zero → Frontier — free, ungated, no attribution required

GROUNDING: This prompt is structured around the U.S. National
Institute of Standards and Technology's AI Risk Management Framework
(AI RMF 1.0) — a free, voluntary framework, not a law or vendor
product. Official document: https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.100-1.pdf
This is a STARTING DRAFT for a small business with no dedicated legal
team — have an actual lawyer review anything you plan to enforce as
policy, especially in a regulated industry.

HOW TO USE THIS: Fill in your business details, then paste the whole
prompt into an AI tool. It will draft a starter policy organized
around NIST's four functions: Govern, Map, Measure, Manage.

---

I need a starter AI use policy for my business. Here's the context:

BUSINESS TYPE: [e.g. "5-person marketing consultancy," "solo law
practice," "10-person nonprofit"]
DO WE HANDLE SENSITIVE DATA? [client financials, health info, legal
matters, personal data — describe what, if anything]
CURRENT AI TOOLS IN USE: [list what your team already uses]
INDUSTRY REGULATIONS THAT APPLY, IF ANY: [e.g. HIPAA, attorney
ethics rules, financial compliance — leave blank if none apply]

Structure the policy around these four sections (based on the NIST
AI Risk Management Framework's four functions):

1. GOVERN — who is responsible for AI decisions at our organization,
   and what's the basic ground rule for using AI here (e.g., "AI can
   assist drafting, but a human must review before anything is sent
   or filed").

2. MAP — where AI is actually likely to touch our work (list the
   realistic use cases based on what I described above) and what
   could go wrong in each (data exposure, inaccurate output, bias).

3. MEASURE — simple, non-technical ways we'll check whether AI
   output is accurate and safe to use (e.g., "any AI-drafted client
   communication is read in full by a human before sending").

4. MANAGE — for each risk identified in MAP, state whether we: fix
   it (add a review step), monitor it, accept it, or ban that use
   case entirely.

Also include:
- A short "Prohibited Uses" list specific to what I described (e.g.,
  no pasting client PII into consumer AI tools without a data
  processing agreement in place)
- A one-line disclosure statement for client-facing work, if
  relevant to my business type

Keep it in plain English a non-lawyer can actually read and follow —
not compliance jargon. Remind me at the end that this is a starting
draft, not legal advice, and should be reviewed by an actual attorney
before being adopted as real policy — especially if I mentioned any
regulated industry above.

---

For regulated professions specifically, see also: ABA Formal Opinion
512 on generative AI (lawyers): https://www.americanbar.org/content/dam/aba/administrative/professional_responsibility/ethics-opinions/aba-formal-opinion-512.pdf

Free to copy, adapt, and reuse. Part of the Zero → Frontier project.

HUMAN REVIEW SELF-CHECK — RUN THIS PROMPT ON AI OUTPUT BEFORE YOU USE IT
Zero → Frontier — free, ungated, no attribution required

HOW TO USE THIS: After an AI tool drafts something for you, paste
that draft BACK into the same or a different AI tool with this
prompt. You're using the AI to help audit its own (or another
model's) output before you trust it — this does not replace your own
judgment, it structures it.

---

I want you to review this AI-generated output critically, as if you
were checking someone else's work, not defending it:

OUTPUT TO REVIEW: [paste the AI-generated draft/output]
WHAT IT WAS SUPPOSED TO ACCOMPLISH: [state the original goal]
WHERE IT WILL BE USED: [e.g. "sent to a client," "published
publicly," "internal only"]

Check it against these seven points and flag anything that fails,
rather than assuming it passes:

1. FACTUAL CLAIMS — does it contain any number, date, name, or
   statistic I should independently verify rather than assume is
   correct?

2. ATTRIBUTION — does it attribute anything to a specific person,
   company, or study? If so, flag it as needing verification.

3. TONE AND VOICE — does this sound generic/AI-written rather than
   like an actual person wrote it? Point out specific phrases that
   feel like AI filler or hedging.

4. IRREVERSIBLE ACTIONS — does using this output as-is trigger
   anything that can't be undone (a send, a payment, a public post,
   a legal or medical claim)? If so, say explicitly that a human
   must confirm before that action happens.

5. PRIVACY — does this reveal or reference anything that looks like
   confidential, proprietary, or personal information that shouldn't
   be here?

6. BIAS OR FAIRNESS — is there any stereotype, unfounded assumption,
   or exclusionary language, even subtle?

7. FINAL VERDICT — would you be comfortable if a human put their
   name on this exactly as written? If not, say specifically what
   needs to change before it should go out.

Be direct about problems — don't soften a real issue into a vague
suggestion.

---

Free to copy, adapt, and reuse. Part of the Zero → Frontier project.

APPROVED / PROHIBITED USE MATRIX — RUN THIS PROMPT
Zero → Frontier — free, ungated, no attribution required

GROUNDING: This pairs with the AI Use Policy Starter Prompt — that
one drafts your overall policy narrative; this one produces the
specific approved/prohibited task matrix that policy should reference.
Grounded in the same NIST AI RMF framework:
https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.100-1.pdf
This is a starting draft, not legal advice — have it reviewed by an
attorney before adoption, especially in a regulated industry.

HOW TO USE THIS: List your team's actual AI use cases (from your Use
Case Inventory if you have one) and let the AI sort them into a clear
matrix.

---

Help me build an approved/prohibited AI use matrix. Here are the
AI use cases my team actually does or might do:
[list every realistic use case — drafting emails, research, code,
client communications, HR decisions, financial analysis, etc.]

DO WE HANDLE ANY OF THE FOLLOWING? [check all that apply, or describe]
- Client/customer personal data
- Financial or payment information
- Health information
- Legal documents or advice
- Hiring/HR decisions
- Anything covered by a specific regulation (name it if so)

For each use case I listed, classify it into one of three columns:
1. APPROVED, NO RESTRICTIONS — low-risk, no sensitive data involved
2. APPROVED WITH HUMAN REVIEW REQUIRED — useful but needs a check
   before anything goes out or gets acted on
3. PROHIBITED — should not be done with AI tools at our organization,
   and explain specifically why (what's the risk)

Be conservative with category 3 given what I told you about sensitive
data — when in doubt between "review required" and "prohibited," flag
it as prohibited pending an attorney's input rather than guessing.

---

Free to copy, adapt, and reuse. Part of the Zero → Frontier project.

AI EXPERIMENTATION LOG — RUN THIS PROMPT
Zero → Frontier — free, ungated, no attribution required

HOW TO USE THIS: Use this after trying a new AI tool or workflow for
the first time. Paste in what happened, and the AI structures it into
a proper log entry — building a record of what's actually been tried,
so your team doesn't repeat failed experiments or lose track of wins.

---

I just tried a new AI tool or workflow. Help me log this properly:

WHAT I TRIED: [tool/workflow name and what you were attempting]
WHAT HAPPENED: [messy notes are fine — what worked, what didn't]
TIME SPENT: [rough estimate]
WOULD YOU DO THIS AGAIN?: [yes/no/with changes]

Structure this into a log entry with:
1. Date and summary (one line)
2. What was tried and why
3. Outcome — be honest if it didn't work, don't spin a failure into
   a false success
4. Time invested vs. time saved (or lost)
5. Recommendation: adopt, adopt with changes, or abandon — and why
6. Tags for searchability (e.g. tool name, task category, team)

If I described a failure, don't soften it — an honest log of what
doesn't work is exactly as valuable as a log of what does, and softening
it defeats the purpose of keeping this record at all.

---

Free to copy, adapt, and reuse. Part of the Zero → Frontier project.