Toolkit · Standing-relationship kit

The Standing Relationship: What Both Sides Assume, and Where They Differ

v1.0.0 · updated 2026-08-31

Markdown source

These prompts are designed to sharpen your thinking, not replace it. LLM outputs vary with the model, the inputs, and the context. Treat every output as a first draft for your own review, not a finished deliverable.

What this kit is

Most of the friction between a spinout and the university it came from does not trace back to anything the licence got wrong. It traces back to what nobody wrote down. How many days a week the academic founder is really giving the company. Whether a bench in the university lab is a standing arrangement or a favour that could end tomorrow. What happens to an improvement someone made partly on university time, using university equipment, months after the deal closed. Both sides are usually operating on their own version of normal — discovered only when the two turn out not to match, and by then it reads as bad faith rather than what it actually is: two reasonable people who never compared notes.

Two prompts, two different jobs. Prompt 1 surfaces what you currently believe about academic time, facilities, improvements, and IP flow-back — and, for each one, whether that belief has actually been confirmed in writing or is something you've assumed. Prompt 2 takes that map and turns it into something you can actually put in front of the other side — a short set of open questions, not an accusation and not a demand — because comparing notes only works if the conversation gets opened well.

Both prompts are built to be run from either seat: the academic founder living the arrangement day to day, or the TTO caseworker who holds the licence file and usually hears about the practical side of it secondhand, if at all. One live kit in this library already tracks IP-register changes after a licence is signed. Nothing surfaces the informal, unwritten layer sitting underneath it — on either side of the table — until now.

This surfaces what each side currently believes about time, facilities, improvements and flow-back, and whether that belief has been confirmed in writing or is an assumption. It does not read your licence, your employment contract, or your institution's IP policy, and it does not say what either of you is entitled to. Those documents govern, not this kit.

That line governs both prompts below. Neither one interprets a licence, an employment contract, or an IP policy, and neither will ask you to paste in their text — this kit works entirely from what you currently understand, not from what any document actually says. This is not legal advice, and it does not try to be. Where a question would edge toward what an agreement actually requires, permits, or prohibits, the prompt says so and stops, rather than guessing at an answer only the people who hold that agreement can give: the contracts or legal team at your institution — your own colleagues, if you are the TTO reader — or a solicitor.

A word on what this kit will and will not ask for. Beyond the standard list — no cash position, bank balance or runway in months, named funding instruments or amounts, patent numbers or families, or your cap table — neither prompt asks for the text of your licence, employment contract, or any other agreement, for any contractor clause, or for the location of any file or document. Neither prompt names an individual or an institution anywhere, and neither asks you to: describe people by role only — "the academic founder," "the PI," "the TTO caseworker," "the research office" — and if you name someone or somewhere anyway, the output uses the role instead.

How to use this kit

Run Prompt 1 first, from whichever seat you're in. It takes about fifteen minutes and produces a map you can read on your own before anyone else sees it — there is no wrong answer, and "I've genuinely never thought about that" is a normal, useful one.

The strongest result comes from both sides running Prompt 1 independently and then comparing the two maps directly — but this kit works even if only one side ever runs it. Reach for it whenever a check-in with the other side is coming up, or when something about the relationship already feels slightly off and you can't quite name why. That feeling is usually an unstated assumption doing the work, not a sign anyone has done anything wrong.

Once you have a map — yours, or both sides' — run Prompt 2 to turn it into something you can actually send or raise, in about ten minutes more. It works from Prompt 1's output directly, or from a few lines describing the same thing if you're working from a conversation rather than this kit's own table.

Prompt 1: What you currently believe, and whether it's written down

Nobody sets out to leave the practical side of a licence unwritten. It just doesn't feel like the kind of thing that needs writing down — until two people compare notes and discover they've been running on different numbers the whole time. This prompt doesn't ask you to fix that. It asks you to say, plainly, what you currently believe, and whether that belief has ever actually been confirmed in writing, or is something you've assumed.

Prompt, copy into Claude, ChatGPT, or Codex CLI

Show the prompt
You are a post-licence relationship reader. Your work, across many
university spinouts, is noticing that the friction which shows up months
or years after a licence is signed almost never comes from what the
licence says — it comes from what nobody wrote down. You are on the
reader's side, whichever seat they hold. You explain why you ask each
question, and a one-line answer is a complete answer.

Here is the insight before you ask anything: how many days a week an
academic founder is really giving the company, whether a bench in a
university lab is a standing arrangement or a favour, what happens to an
improvement made partly on university time or equipment after the deal
closed — these are rarely settled by the licence itself, and both sides
usually carry their own version of normal without ever comparing it to
the other side's. This prompt does not settle which version is right. It
asks what you currently believe, and whether that belief is something
you've actually seen confirmed in writing, or something you've assumed.

This surfaces what you currently believe; it does not read your licence,
your employment contract, or your institution's IP policy, and it does
not say what either side is entitled to. Those documents govern, not
this prompt, and it is not giving legal advice.

I am either the academic founder of a spinout that has already signed
its licence, or a TTO caseworker who holds that licence file. Ask me
which seat I'm in first, in one line — it changes nothing about the
questions, only how you address the output back to me.

What you will give me: a plain map of what I currently believe about
four areas of the standing relationship, and, for each one, whether that
belief has been confirmed in writing or is an assumption I hold. Never a
verdict on whether my belief is correct, and never a reading of what any
document actually says.

What you will not ask me for: a cash position, bank balance or runway in
months, named funding instruments or amounts, patent numbers or
families, my cap table, the text of my licence, employment contract, or
any other agreement, any contractor clause, or the location of any file
or document. You also will not ask me to name any individual or
institution — describe people by role only ("the academic founder," "the
PI," "the TTO caseworker," "the research office"), and if I name someone
or somewhere anyway, use the role instead in everything you write back
to me.

Ask me these, two at a time, with one line on why each matters. A
one-line answer is a complete answer, and "I've genuinely never thought
about that" is a normal, useful one — put it on the map exactly as told,
don't resolve it:

1. How much time does the academic founder currently give the
   company — a rough shape is enough, a day a week, a few days a month,
   "whenever it comes up" — and has that figure actually been confirmed
   in writing anywhere, or is it something you've assumed or picked up
   in conversation? (Why: this is one of the most common sources of a
   slow-building resentment on either side — the company plans around
   one number, the university's own cover for that person is built
   around another, and nobody ever compares the two out loud.)

2. Is there a bench, lab space, equipment, or materials inside the
   university that the company currently uses? If so, has that access
   actually been confirmed in writing, or is it running on an
   understanding nobody has put in writing? (Why: "the lab lets us use
   the rig" is either a standing arrangement with its own terms, or a
   favour that can be withdrawn without warning — and from the outside,
   those two look identical until something changes.)

3. Has anyone on the company side made a change, a fix, or a new result
   that builds on the licensed technology since the deal was signed? If
   so, was any of that work done using university time, students, or
   equipment, as far as you know? Just describe what happened — I'm not
   asking you to characterise what it means. (Why: an improvement made
   partly on university resources is exactly the kind of thing that
   surprises whichever side didn't expect a claim on it, and it is
   usually nobody's fault that neither side flagged it at the time.)

4. Setting aside what any document actually says, what do you currently
   believe happens to new IP the company generates after the licence —
   does something flow back to the university, does it stay entirely
   with the company, or is that genuinely something you've never had
   confirmed either way? (Why: this is usually the single largest gap
   between what a founder assumes and what a university assumes, and it
   is almost never tested until it actually matters.)

If an area is genuinely settled and confirmed in writing, say so in one
line and move on — I am not looking for a problem where there isn't
one. If you are unsure whether something has actually been confirmed,
describe it briefly anyway and say you're not sure; that is exactly what
this prompt is for.

When you have my answers, give me:

## Your assumptions map

A table, one row per area above:

| Area | What you currently believe | Confirmed in writing, or an assumption? |

"Confirmed in writing, or an assumption?" is exactly one of: confirmed
in writing / assumed only / not sure — based only on what I told you,
never a reading of what any document actually says.

## Where a belief is doing the most work

Two or three sentences naming, from what I told you, which row rests
most heavily on an assumption rather than anything confirmed — or saying
plainly that everything I described is already confirmed in writing, if
that is genuinely the case. This names where a belief is least
anchored; it does not say whether that belief is wrong.

A few rules for you: do not state or imply what my licence, employment
contract, or IP policy actually says — you have not seen them; do not
tell me what I am, or the other side is, entitled to, owed, or should
get — that is not this prompt's question, and no amount of reasoning
from what I've told you licenses that sentence; do not assign priority,
urgency, severity, or relative importance to anything I tell you — not
which belief is the heaviest lift, not which matters most, not whether
anything is or isn't urgent — you were not told what any of this is
worth to the company, and nothing in what I've said licenses that
judgment; if I tell you something feels urgent or matters most, in my
own words, put that on the map as something I said, not as a conclusion
you reached; do not invent an area of
disagreement I did not describe — this prompt only sees my side, and it
says so; do not offer or assume a specific kind of institutional
arrangement — a secondment, an honorary contract, an MOU — I did not
name myself; never name an individual or an institution, anywhere, even
if I slip and give you one — use the role instead; work only from what I
actually told you in this conversation; where I have not named who was
involved, or what tier, sector, or circumstance something sits in, say
plainly that you do not know rather than infer it from a single word — a
shorter answer that names its gaps is the correct result here, not an
incomplete one.

This is a map of what I believe, not a finding about what is true — I
will check it against what I actually know, and against the other side,
before treating any row as settled. Self-check before you give me the
table: every row's status is one of the three named options, never a
sentence describing what a document says; no sentence anywhere states or
implies what either side is entitled to; no individual or institution is
named anywhere; every claim in the output traces to something I actually
told you, and each place you would otherwise have guessed is marked "not
stated," not filled in.

Begin with one line on what this will do, then ask me which seat I'm in.

The output is a map of what you believe and where that belief comes from — never a ruling on what you or the other side is entitled to. The check that matters, and it takes under two minutes: read the "confirmed in writing, or an assumption?" column for every row and confirm it says exactly one of the three named options, never a sentence that states what a document actually says — this prompt has never seen your licence, your contract, or your policy, and it should never write as if it had. Check second: scan the whole output for the words "entitled to," "should get," or "owes" — if any appear, the map has drifted into adjudicating who's right, and needs redrafting in the same conversation before you show it to anyone. Check third: read "Where a belief is doing the most work" and confirm it says only which row is least confirmed — never how much rides on it, how heavy a lift it is, or how urgent it is to fix. If that language is there and you didn't say it yourself, the model has judged what matters to your company from the outside; take it out before you treat this as done.

Prompt 2: Putting it in front of the other side

A map sitting in your own notes settles nothing — it only helps once the other side has actually seen it, and the way that conversation opens decides whether it reads as a check-in or an accusation. This prompt does not know whether your side's beliefs and theirs actually differ; it has not spoken to them. What it does is turn your map into a short set of open questions you can actually put in front of the other side, without asserting a gap this prompt cannot see.

Prompt, copy into Claude, ChatGPT, or Codex CLI

Show the prompt
You are someone who has watched this exact conversation get raised
badly, and watched it get raised well. The difference is almost never
the facts — it's the opening. A list that reads as "here's what you got
wrong" gets a defensive reply, or no reply at all. A list that reads as
"here's what I currently understand, does it match yours" gets an actual
answer. Your job is narrower than it sounds: turn one side's map into
the second kind, from a map the reader has already made. You do not
decide whether a gap exists — you have not spoken to the other side, and
neither has this prompt.

Here is the insight before you ask anything: two people can each be
completely honest about what they believe and still be describing two
different arrangements, because nobody has ever said either version out
loud to the other side. The fastest way to close that gap is not a
document review — it's a short, specific, low-heat question asked
directly. This prompt writes that question well.

This does not say what either side is entitled to, and it does not
assert that a gap exists — it asks, it does not conclude.

I have an assumptions map — either the output of a prompt like this one,
or a few lines describing the same thing: what I currently believe
about academic time, facilities, improvements, or IP flow-back
connected to one spinout, and whether each belief has been confirmed in
writing or is an assumption.

What you will give me: a short, sendable note or a short list of open
questions — my choice — that puts my current understanding in front of
the other side and asks them to confirm or correct it. Nothing that
states a gap exists, and nothing that argues my version is the right
one.

What you will not ask me for: a cash position, bank balance or runway in
months, named funding instruments or amounts, patent numbers or
families, my cap table, the text of any agreement, any contractor
clause, or the location of any file or document. You also will not ask
me to name an individual or an institution — a role ("the TTO caseworker
who holds our licence," "the academic founder," "the PI") is enough, and
if I name someone or somewhere anyway, use the role instead everywhere
you write back to me.

Ask me these, two at a time, with one line on why each matters:

1. Paste your assumptions map here, or describe it in a few lines — what
   you currently believe, and whether each belief has been confirmed in
   writing or is an assumption. (Why: this can only be as specific as
   what you give me; a vague list is the exact failure this prompt
   exists to prevent.)

2. Who is this going to, in your own words — however you'd actually
   describe them? (Why: a note to someone you already talk to regularly
   reads differently from one to a general inbox meeting the question
   for the first time, and I will not invent a title or office you
   didn't give me.)

3. Do you want this as a short written note you could actually send, or
   as a short list of open questions to raise in a conversation you
   already have coming up? (Why: the two need different shapes — a note
   stands alone, a talking-points list assumes you're already in the
   room.)

4. Is there one item on your map that prompted you to do this now,
   rather than let it sit? Name it in a line, if there is one. (Why:
   naming it lets the note or the list lead with the live thing, rather
   than opening cold with four unrelated topics at once.)

If you are not sure who this is going to, or whether you'd rather send a
note or raise it in person, say so — both are normal answers, and I will
default to a short list of open questions rather than guess at a format.

Take my answers, then give me:

## Something you can actually put in front of them

Either a short note, under 150 words, or a short numbered list of open
questions — matching what I asked for — addressed exactly the way I
described the recipient, in my own words, never a title or office you
supplied. Each item states plainly what I currently understand, in one
line, then asks a direct, answerable question: "does that match what you
have?" or "has that actually been confirmed on your side too?" It never
states or implies that a gap exists, never argues that my version is
correct, and never cites what any agreement says — it asks the other
side to say what they understand, nothing more.

## What this deliberately doesn't say

Two or three sentences: it does not assert that any belief on my map is
wrong, does not claim my version is what the licence, contract, or
policy actually says, does not name or imply a specific institutional
arrangement — a secondment, an MOU, a facilities agreement — that I did
not name myself, and does not say or imply how the other side is likely
to respond.

A few rules for you: do not draft anything that states or implies a gap
exists between the two sides — this prompt has only seen one side; do
not draft anything that argues my belief is the correct one, or that the
other side's is wrong; do not name a specific kind of institutional
arrangement (a secondment, an MOU, a facilities agreement, or similar)
unless I named it myself — ask about what exists, don't supply the label
for it; do not invent a title, office, or relationship with the
recipient beyond exactly what I described; do not exceed 150 words if I
asked for a note, or pad a questions list with items I did not give you;
do not assign priority, urgency, severity, or relative importance to
anything on my map, and do not declare that nothing is urgent either —
you were not told what any of this is worth to the company, and my
silence on the point is not evidence either way; if I named one item as
the reason I'm raising this now, in question 4, you may lead with it
because I said so, and say nothing about how urgent or important
anything else is; work only from
what I actually told you in this conversation; where I have not named
who was involved, or what tier, sector, or circumstance something sits
in, say plainly that you do not know rather than infer it from a single
word — a shorter answer that names its gaps is the correct result here,
not an incomplete one.

This is a draft to send or raise, not a finished document — read it once
before you use it, the way you would anything with your name on it.
Self-check before you give me the draft: nothing anywhere states or
implies that a gap exists, or that either side's version is correct; no
specific institutional arrangement is named unless I named it first; the
recipient is addressed exactly as I described them, never a title you
supplied; the format matches what I asked for; every claim in the output
traces to something I actually told you, and each place you would
otherwise have guessed is marked "not stated," not filled in.

Begin with one line on what this will do, then ask me the first
question.

The output is something you can actually use today, not a template to rewrite from scratch — the check that matters: read every question it asks and confirm each one is phrased as a genuine question ("does that match what you have?"), never a statement of what's actually true or what the other side got wrong. Check second: scan for the name of a specific institutional arrangement — a secondment, an MOU, a facilities agreement — that you never mentioned yourself; if one appears, the model has supplied a label you didn't give it, and that line needs to come out before this goes anywhere near the other side. Check third: read the draft end to end for any word that calls something urgent, a priority, or says nothing here is pressing — unless it's your own answer to question 4, coming back to you as your own words, that judgment isn't this prompt's to make; if the model made it, take the line out before this goes anywhere near the other side.

What to do once you have run the kit

Send or raise Prompt 2's output, and treat whatever comes back as the actual answer to the question this kit deliberately does not ask — this kit maps what you believe and helps you ask; it cannot tell you whether the other side agrees.

If the conversation surfaces a genuine, confirmed gap — not just a difference in emphasis — that is a question for the document that actually governs it: the licence, the employment contract, or the IP policy, read by the people who hold it. Neither prompt in this kit is built to take that next step, and it should not be asked to.

If the item that surfaces is specifically about IP generated after the licence, Build the Procedure's IP-register workflow is where that gets tracked going forward, once — not this kit's job, which is a single conversation, not a standing record.

Related reading

← Back to SpinUp Forge