The first 90 days are the most leveraged days of any role you will ever have. Every relationship is forming. Every assumption is being made. Every signal you send compounds. People who have been there for years are watching to see who you are, and the version of you they decide on in week three is the version they will keep for two years even after you have changed.
The standard advice for this period is some version of "listen first, build relationships, get a quick win, do not break anything." This is true and also useless. It is true at the same level that "eat less, move more" is true. The interesting question is not whether to listen but who to listen to, in what order, asking what, looking for what. The interesting question is not whether to get an early win but how to identify the win that signals the right thing about you to the right people without burning a bridge with the people who liked the way things were.
This is where structured AI rehearsal earns its keep. Not as a substitute for relationships, but as a thinking partner for the parts of the first 90 days that are too dense, too political, and too time pressured to figure out alone in real time. The LAUNCH Framework is fifty prompts for the prep, the audit, the decoding, and the planning that should happen between the visible activities of the first 90 days. Use it before the meeting, after the meeting, on the train home, on Sunday night. It will not replace the work. It will make the work meaningfully sharper.
This guide is LLM agnostic. The prompts work in ChatGPT, Claude, Gemini, and any approved enterprise model. Anonymise where data sensitivity requires it. Specificity in your inputs is the difference between generic onboarding advice and a real plan for your specific situation.
What is the LAUNCH Framework
A method for using AI as a structured thinking partner across the six categories of work that determine how a new role lands. Each pillar is one of the things you actually have to do in the first 90 days, and is rarely taught explicitly.
- L: Listen. Plan the listening tour. Decode what people are and are not saying. Find patterns across conversations.
- A: Audit. Inherit the situation honestly. Surface the technical, financial, team, and reputational realities you have walked into.
- U: Understand. Map the stakeholders, the politics, the unwritten rules, and the silent coalitions.
- N: Network. Build the relationships that will determine whether you ship or stall. Earn credibility deliberately.
- C: Choose. Decide what to focus on. Decide what to ignore. Decide what to deliberately delay.
- H: Hit. Bank early wins that signal the right thing to the right people without burning bridges.
Six pillars. The pillars run in rough sequence (you listen before you audit, audit before you understand, and so on) but the framework loops. By Day 60 you will be back in Listen mode for new questions that emerged from your Choose pillar work.
How to use this guide
Three principles. First, the first 90 days is not a performance. The temptation is to use AI to draft the impressive things you will say in your first all hands or the polished version of your first strategy memo. That is not where the leverage is. The leverage is in the thinking you do alone before those visible moments. Use AI for the audit, the map, the plan. Then go do the meeting yourself, in your own voice.
Second, the model can only help with what you give it. Paste the actual organisation chart, the actual quarter end numbers, the actual feedback from your boss, the actual three things your predecessor told you on your handover call. Generic input gets generic output. The specificity is the value.
Third, run the prompts twice. The first pass tells you what you have. The second pass, after a week of new information, tells you what you missed. Where the prompt says [paste context] or [insert names], do that literally.
L: Listen
L1: The Listening Tour Plan
I am [role] at [company / team]. Help me design my first 30 days
listening tour. List the categories of people I should meet (direct
reports, peers, cross functional partners, key customers, board /
exec sponsors, departed predecessors if reachable). For each category,
the right question to lead with, the wrong question to avoid, and the
order to do them in. Company and role context: [paste]
L2: The "What Are They Actually Saying" Decoder
Below is the transcript or notes from my conversation with [person /
role]. Tell me what they said directly, what they implied but did not
say, what they avoided saying that I should have noticed, and what
the most charitable and least charitable interpretations of each
section are. Flag the one sentence that is most worth re asking them
about. Notes: [paste]
L3: The Cross Conversation Pattern Finder
I have done five conversations with [paste roles or names]. Notes are
below. Find the patterns: what came up unprompted in three or more
conversations, what nobody mentioned that I would have expected,
where different people described the same situation in incompatible
ways. End with the three questions worth asking in my next round.
Notes: [paste]
L4: The "What Is Everyone Afraid to Say" Prompt
Based on the conversations and context below, generate the three
things that are most likely true about this team or company that
nobody has told me yet because the topic is politically expensive.
For each, tell me how I would find out without making it about me
being the new person who heard a rumour. Context: [paste]
L5: The Silence Reader
Below are my notes from the last two weeks of listening. Identify the
silences: what topics have not come up that should have, which
stakeholders have not engaged that should have, which standard
artefacts (roadmap, OKRs, retros, financials) I have not been shown.
For each silence, tell me what it most likely signals. Notes: [paste]
L6: The Cross Functional Friction Surfacer
Below is what I have heard about how [my team / function] works with
[adjacent team / function]. Identify the friction points implied by
what people are saying. For each friction point, tell me whether it
is structural (workflow design), interpersonal, or strategic. End
with the one friction I should not try to fix in my first 90 days
because the cost of touching it exceeds the benefit. Notes: [paste]
L7: The "What My Predecessor Would Not Tell Me" Prompt
My predecessor [left / was promoted / was fired]. They told me
[paste handover]. Generate the three things they probably did not
tell me about this role, this team, or this company, based on the
combination of how they left and what they did say. For each, tell me
how I would find out without making them look bad to people who still
work with them.
L8: The Manager Decode
Below is what my new manager has said to me in our first three
conversations and any other context I have about them. Tell me what
they care about, what they say they care about, how those differ,
what their tells are when they are unhappy, and what the version of
my role looks like that would make them feel like they made the right
call hiring me. Inputs: [paste]
A: Audit
A1: The Inherited State Audit
I have inherited [role / team / function]. Below is what I know about
the current state: financial, team, technical, customer, reputational.
Audit it. Rate each dimension on healthy, fragile, or actively broken.
For each broken or fragile area, tell me whether it must be addressed
in 90 days or whether it can wait until I have political capital.
State: [paste]
A2: The Technical and Process Debt Inventory
Below is what I have learned about how my team or function actually
gets work done. Inventory the debt. Process steps that exist for
historical reasons no one remembers. Tools that nobody likes that
nobody changes. Reports that get written that nobody reads. For each
debt item, score the cost of carrying versus the cost of removing.
Notes: [paste]
A3: The Promise Ledger
Audit the promises that exist in my new world. What has my
predecessor, my manager, my team, or the company committed to that I
am now responsible for delivering? Group into still credible, at
risk, and dead but undeclared. For the dead but undeclared, tell me
how to surface them without owning a failure I did not cause. Source
material: [paste]
A4: The Hidden Landmine Surfacer
Based on the context below, generate the five most likely landmines
in my new role. Not the obvious ones I have already been warned
about. The ones that detonate three to six months in: legacy
commitments, departed key people, contracts nearing renewal, audits,
regulatory exposure, talent retention risks. For each, tell me the
cheapest move this month to defuse it. Context: [paste]
A5: The "What Did the Last Person Actually Deliver" Audit
My predecessor was [paste their role and tenure]. Below is what they
were publicly known for delivering. Audit the gap between what they
were known for and what they actually delivered, based on the data I
have access to. Tell me which of their reputational wins were real,
which were narrative, and which were neither. Inputs: [paste]
A6: The Team Capability Audit
Below is what I know about my direct reports or team members: roles,
tenure, recent feedback, projects they own. For each person, tell me
what I can probably tell about their capability and engagement from
the data, what I cannot tell without a real conversation, and which
two people are most worth my first deep one on ones based on signal
versus noise in their profile. Team data: [paste]
A7: The Financial Reality Check
Below are the financial inputs I have for my new world (budget,
headcount, run rate, recent variance). Audit them. Tell me where the
numbers are saying something the narrative is not saying. Flag any
metric whose definition I should triple check before I commit to it
in a board or leadership conversation. Financials: [paste]
A8: The Reputation Inherited
What is the existing reputation of [my team / function / role] inside
this company? Based on the comments, asides, jokes, and direct
statements in my notes, characterise the prevailing view. Tell me
which parts of the reputation are deserved, which are stale, and
which one piece of reputation I should be deliberate about updating
in my first 90 days. Notes: [paste]
U: Understand
U1: The Stakeholder Power Map
Below are the stakeholders for my role. For each, score on formal
authority over my work, informal influence over my outcomes, current
disposition toward my role (champion, neutral, sceptic, blocker), and
the cost to me of having them at each disposition six months from
now. Recommend the three to invest in first. Stakeholders: [paste]
U2: The Decision Reality Map
On the question of [paste a decision relevant to my role], who
actually decides versus who has the title that says they decide?
Based on the context below, map the decision: who proposes, who
shapes, who signs off, who can veto. Tell me where my reading
diverges from the org chart and what that divergence implies for how
I work the decision. Context: [paste]
U3: The Boss's Boss Decode
My manager reports to [paste skip level]. Tell me what I should know
about that relationship: how my manager talks about their boss, how
the skip level talks about my function, what success for my role
looks like from the skip level's vantage. Identify the one thing my
manager probably filters out before it reaches their boss. Inputs:
[paste]
U4: The Coalition Mapper
Based on the conversations, meeting dynamics, and email patterns I
have observed, map the coalitions in [my team / division / company].
Who has lunch with whom, who CCs whom unnecessarily, whose names come
up together in unrelated contexts. Tell me which coalition I am
implicitly being slotted into and whether I should accept the slot or
position differently. Inputs: [paste]
U5: The Unwritten Rules Surfacer
Below is what I have observed about how decisions, meetings,
communications, and conflicts actually happen in this company.
Surface the unwritten rules. For each, tell me whether it is harmless
local custom or a load bearing norm I should not violate in my first
90 days. Observations: [paste]
U6: The Conflict History Reader
Based on the context below, what are the historical conflicts in my
new world that I am walking into? Person versus person, team versus
team, narrative versus narrative. For each, tell me what would be
naive of me to do, what would be sophisticated, and what would
deliberately blow up the existing equilibrium if I chose to.
Context: [paste]
U7: The "What Would Disappointment Look Like" Prompt
Three months from now, who in my new world would I most regret
disappointing? Based on the inputs below, generate a ranked list.
For each person, tell me what would constitute disappointing them in
their own words, not in mine, and the cheapest sustained behaviour
this quarter that would prevent it. Inputs: [paste]
U8: The Hidden Sponsorship Mapper
Who in this company has a quiet stake in my success that I have not
yet identified? Based on the context below (who pushed for my hire,
who attended my first meetings, who has reached out unprompted,
whose calendars open easily for me), map the sponsorship. Tell me
which sponsor I am underweighting and how to invest in that
relationship without being transactional. Context: [paste]
N: Network
N1: The First 30 Coffees Prioritisation
I have time for roughly 30 introductory conversations in my first
month. From the list below of people who have been offered or
suggested as introductions, rank them by combined value to my work
and difficulty of access. For each top 10, tell me the question
worth leading with and what success looks like from that
conversation. Candidates: [paste]
N2: The Cross Functional Ally Identifier
Across the functions I work with (paste functions), who in each is
most likely to become a real ally based on the inputs below: tenure,
prior projects, what people say about them, their stated priorities.
For each function, tell me the one ally to invest in and the one
person to be careful around. Inputs: [paste]
N3: The "Who Has Solved This Before" Finder
The problem I am most likely to face in the next 90 days is [paste].
Generate a list of likely archetypes of people who will have faced
the same problem (inside this company, in adjacent companies in the
industry, in the broader network). For each, tell me how to find
them without burning a favour and the most useful question to ask
when I do.
N4: The Outreach Drafter
I am reaching out to [paste person and their role]. Context on why:
[paste]. Draft three openers. First is warm and low pressure. Second
is specific and asks for a defined thing. Third is a question they
are uniquely positioned to answer. Tell me which is right for who
they are, and what to tighten before I send.
N5: The Peer Relationship Starter
My peers in this role at the same level are [paste names and roles].
Based on what I know about them, generate the right way to start
each relationship: who needs to be approached as a collaborator,
who as a friendly competitor, who as a sceptic to convert. For each,
the first concrete interaction I should propose.
N6: The Board or Exec Sponsorship Cultivation
I have access to [paste board members / exec sponsors]. For each,
generate the version of me they most want to see succeed (based on
why they backed my hire, their public statements, their other
investments). Tell me the cadence and content of updates that would
make each feel they made the right call, without becoming a
performance burden on me. Inputs: [paste]
N7: The Visibility Ladder
Build me a visibility ladder for my first 90 days. Where should I
be seen (which meetings, channels, forums), by whom, and saying
what. Rank by signal to noise ratio. Tell me the two visibility
moves that pay back most and the one I should specifically avoid
because the cost in attention exceeds the credibility gained.
Context: [paste]
N8: The Networking ROI Audit
Looking at my recent coffees, calls, and conversations, score each
on actual return: what did I learn, what did I commit to, what
follow up did I owe, who got value from me. Identify the pattern in
the conversations that are draining versus the ones that are
generative, and recommend the change to my approach for the next 30
days. Conversation log: [paste]
C: Choose
C1: The 90 Day Priority Filter
Below is the long list of things I could focus on in my first 90
days. Filter ruthlessly. For each candidate, score on (a) signal to
the people who matter, (b) reversibility if I am wrong, (c) cost in
time and political capital, (d) whether someone else is already
expected to own it. Recommend the three to commit to publicly and
the five to do quietly. Candidates: [paste]
C2: The "What to Say No To" Decider
In the next month I have been asked to take on or attend [paste].
For each request, score the cost of saying yes (time, signal,
precedent) and the cost of saying no (relationship, optics, missed
information). Recommend the no's and draft the version of no that
preserves the relationship for each.
C3: The 70 / 20 / 10 Allocator
Help me allocate my first 90 days across running what exists,
changing what should change, and building what does not yet exist.
A 70 / 20 / 10 split is the default. Based on the context below,
tell me whether my situation calls for that ratio or a different one,
and what each percentage should specifically translate to in
deliverables. Context: [paste]
C4: The Delegation Trigger
Below is the list of work currently on my plate or being handed to
me. For each, recommend keep, share, delegate, or kill. For each
delegate, name the most likely owner in my context and the framing
that hands it off without dropping it. List: [paste]
C5: The Strategic Narrowing Prompt
Below is the broad mandate I have been given. Narrow it. Generate
three credibly different narrower mandates that would still satisfy
the broad version. For each, tell me what work it commits me to,
what it lets me drop, and which version of me succeeds in that
narrower world. End with the recommended choice. Mandate: [paste]
C6: The Unfair Advantage Locator
Given what I bring to this role (paste background, skills, network,
prior wins) and the context I have walked into, locate my unfair
advantage. Where is the gap between what this role currently demands
and what I uniquely can offer. Tell me the one play that uses that
advantage that nobody else in my seat would have run. Inputs: [paste]
C7: The 90 Day Roadmap Drafter
Draft my 90 day roadmap based on the priorities, constraints, and
inputs below. Structure: 30 days listen and audit, 60 days plan and
align, 90 days commit and ship. For each phase, the deliverables, the
signals of success, the explicit risks, and the one thing I will not
do that I might be tempted to do. Inputs: [paste]
C8: The Deliberate Delay Planner
What should I deliberately not do in my first 90 days even though I
could. Below are candidates: hiring decisions, org changes, strategy
declarations, vendor changes, system migrations. For each, tell me
the right time to revisit and the cost of delaying. End with the one
delay that requires explicit communication to manage expectations.
Candidates: [paste]
H: Hit
H1: The Early Win Identifier
Generate a list of candidate early wins for my first 60 days. Each
should be: low cost to execute, visible to the right people, hard to
attribute to luck, and forward looking rather than cleanup. For each,
tell me what it signals about me to whom, and rank by signal to cost
ratio. Inputs: [paste]
H2: The 30 Day Deliverable Structure
Help me structure the deliverables for my first 30 days so that they
land as a coherent set rather than scattered activity. Below are my
in flight projects, observations, and obligations. Group them into
three to five themes a board update or skip level could absorb in
two minutes. Inputs: [paste]
H3: The First Announcement Drafter
Draft my first significant announcement (paste topic). Three versions:
maximally cautious, balanced, and high conviction. For each, predict
the reaction from my three most important stakeholders and how it
shapes their view of me. Recommend the version and the one sentence
to cut or rewrite.
H4: The Credibility Banking Checkpoint
Looking at the trajectory of my first 60 days, where am I banking
credibility and where am I burning it? Be specific. For each behaviour
or moment in my notes below, tell me which side of the ledger it
landed on for which audience. Recommend the one course correction
this week. Notes: [paste]
H5: The Over Promise Calibrator
Below is what I am preparing to commit to publicly (paste). For each
commitment, audit whether I have the inputs, the team, and the time
to deliver. Tell me which commitments I should silently downgrade,
which I should explicitly hedge, and which to remove. End with the
sentence that protects me without sounding like I am hedging.
H6: The First Quarter Review Prep
My first quarterly review with [my boss / the board] is in [paste
timeframe]. Help me prepare. What evidence I should bring of progress,
what to lead with given what I know about how they read situations,
what concession or open question to volunteer that signals honesty
without inviting attack. Context on the audience: [paste]
H7: The Early Win Sequencing Prompt
Below are my three to five candidate early wins for the first 90
days. Sequence them. Which to ship first to set the right tone,
which to land in week six to break a possible slump, which to hold
for the quarterly review. For each, the one risk that could turn the
win into a visible mistake.
H8: The Signal Audit
By Day 90, what is the story being told about me inside this
company? Based on my actions, decisions, and visible deliverables to
date, predict the three narratives that are currently competing for
prevalence. For each, tell me what behaviour this month would tip
the narrative in that direction. Recommend which to actively
cultivate and which to deliberately undercut. Inputs: [paste]
Bonus Prompt B1: The "What Would I Tell Myself at Day 1" Audit
Looking forward to Day 90 and back at Day 1, write me the note I
would want from my future self today. What did I overestimate. What
did I underestimate. Who did I underweight. Who did I overweight.
Which meeting did I dread that turned out to matter. Which meeting
did I prep for that did not. Be specific to my context below.
Context: [paste]
Bonus Prompt B2: The "Do I Actually Want This Role" Self Audit
Run a self audit on whether I actually want this role, not whether
I want to be seen succeeding in it. What was attractive about the
job description versus the job. What is harder than I expected.
What is more interesting than I expected. What would have to be
true at Day 90 for me to know this was the right call, and what
would have to be true for me to know it was not. Inputs: [paste]
The LAUNCH Framework in one image
L A U N C H
Listen Audit Understand Network Choose Hit
The pillars run in rough sequence but the framework loops. By Day 60 you are likely back in Listen for new questions surfaced by your Choose pillar work. The point is not strict order. The point is that when you sit down on a Sunday in Week 4 with a knot in your stomach about something you cannot yet name, you can find the pillar that fits and use the right prompt rather than spiralling.
How to combine LAUNCH with your model of choice
ChatGPT: 4 series models are strong for the Listen and Network pillars where the work is generative (drafting outreach, decoding conversations). o series models are stronger for the Audit and Choose pillars where the work is analytical (stakeholder mapping, priority filtering). A workable pattern is to draft with 4 series and stress test with o series.
Claude: works well across all six pillars, especially when you load your notes, org charts, and conversation transcripts into a Project. Claude tends to be honest about political dynamics and stakeholder reading without retreating into corporate hedging. Tell it explicitly to give you the version a sharp peer would deliver in private.
Gemini: deep research mode is useful for the Audit pillar when you are trying to triangulate the external reputation of your new team, function, or company against what you are hearing internally. Watch for context drift on the longest stakeholder mapping prompts.
The cross model habit that pays compounding dividends is to run the stakeholder map (U1) in two models. Where they agree on who matters, the map is robust. Where they diverge, the divergence reveals where your inputs were ambiguous and where you should ask better questions in your next round of Listen.
Frequently Asked Questions
Is the LAUNCH Framework only for new managers, or does it work for ICs and founders too?
It works across all of them. The pillars are the same. What changes is the cast: an IC's listening tour is shorter and the stakeholders are different from a CEO's, but the underlying work of listening, auditing, understanding, networking, choosing, and hitting is identical. Adjust the inputs you paste, not the framework.
How is this different from Michael Watkins' "First 90 Days"?
Watkins built the canonical map of what to do. The LAUNCH Framework is the operating system that uses AI to do the work he describes faster and more rigorously. Watkins gives you the questions to ask. The LAUNCH prompts help you actually structure the answers, decode the conversations, and pressure test your hypotheses. The two pair well. Read Watkins for the conceptual frame. Use LAUNCH for the weekly execution.
What if I am taking over from a popular predecessor versus a problematic one?
Different defaults. A popular predecessor means your Listen pillar is heavier (you need to understand what people loved before you change anything) and your Hit pillar is harder (early wins risk reading as criticism of the prior regime). A problematic predecessor means your Audit pillar is heavier (you need to surface what was actually broken without trashing them publicly) and your Hit pillar is easier (people are waiting to be helped). The framework is the same. The weighting changes.
Can I use these prompts with my actual notes and meeting transcripts?
Check your model tier's data retention policy first. Enterprise tiers of ChatGPT, Claude, and Gemini typically offer the data retention controls that make this acceptable. Consumer tiers usually do not. For sensitive personnel information, anonymise names and identifying details before pasting. The prompts work without the real names.
When do I run the prompts versus do the work?
Always before and after, never instead. Before a meeting: run U2 to map the decision dynamics. After a meeting: run L2 to decode what was said. Before the Sunday night planning: run C1 to filter priorities. After the first quarterly review: run H4 to audit credibility banked. The prompts are for the thinking you do alone. The work is what you do with people.
Is there a paid version of these prompts?
Yes. The fifty prompts above are the free version, in compact form. The LAUNCH Pro Pack includes 150 expanded prompts with example outputs, ready to load Claude Projects and custom GPT configurations for each pillar, and role specific onboarding sets for new CEOs, new heads of function, new ICs joining founder led teams, and new board directors. The Pro Pack is on the PromptLeadz Pro Collection at $29.
Where to go next
LAUNCH sits at the start of the role lifecycle. Once you are through the first 90 days, the role specific anchors in the PromptLeadz Free Vault take over. New product managers pair LAUNCH with the 7P Framework. New founders pair it with the FOUNDER Framework. New ops leaders pair it with OPS7. When the hard conversations start to arrive, HARDER takes over. When you need the AI to stop agreeing with you, CRITIC takes over.
The thing to internalise is that the first 90 days reward structure more than they reward charisma. The person who shows up at Day 90 with a clear map, a credible network, three banked wins, and a quiet sense of where the landmines are buried, beats the person who showed up at Day 1 with a memorable first impression every time.
PromptLeadz publishes battle tested AI prompt packs for founders, product, sales, marketing, operations, HR, finance, customer success, adversarial thinking, hard conversations, and new role launches. All prompts are LLM agnostic. Pricing is in USD.
Để lại bình luận: