$ cat ./gasketos-agent-mode.md

#GasketOS · Agent Mode

% Prompts for the assistant you already use


##Explainer


GasketOS is software for pharmaceutical and medtech companies that sell on negotiated contracts. Your customer contracts earn money your invoices never collect. GasketOS recovers it.

GasketOS reads every term in your contracts, checks it against every invoice line, and turns each gap into an owned action that is tracked until the money is back. It runs alongside your CRM, CLM and ERP.

Our specialized AI agents read every term and number. Your teams confirm. Our deterministic engine reconciles and does the math.

The whole site is published as plain text, so your assistant reads what you read. Agent Mode turns that into a working session: paste the setup prompt, tell it about your company, pick a prompt.

##How to use this


  1. 1.Copy the setup prompt.
  2. 2.Paste it into ChatGPT, Claude, Copilot, or the assistant of your choosing.
  3. 3.Answer its questions about your company.
  4. 4.Pick one prompt below.

##Connect your assistant


Paste this into your assistant once.

Paste once
Open /llms.txt
You are helping me evaluate GasketOS (https://www.gasketos.com) for my company.

Read these first. Treat them as the only source of what GasketOS does:
- https://www.gasketos.com (the website)
- https://www.gasketos.com/llms.txt and https://www.gasketos.com/index.md (the same site as plain text)
If a page will not open, try the others, then tell me which ones failed and I
will paste them in.

If we decide GasketOS is worth a call, the booking link is https://calendly.com/gasketos/30min.

How to work:
- Keep what GasketOS says separate from your own inference. Quote the site
  for anything you attribute to GasketOS.
- If the site does not answer something (prices, customers, certifications,
  data hosting, integrations, timelines), say "not stated on the site" and add
  it to a list of questions for a call. Do not fill the gap from memory or guesses.
- Ask me about my company before judging fit: our industry, what we sell and
  to whom, how contracts and billing are run, and where it hurts today.
- Our contracts and billing data are confidential. Do not open files, folders
  or connected tools unless I point you to them, and never send anything to
  GasketOS or any outside service.

When you have read the site, reply with a three-sentence summary of GasketOS,
the questions you need me to answer, and these five next steps:
1. Fit check  2. Leakage map  3. Size the opportunity  4. Prepare the call  5. Brief my colleagues

##Prompts


Fit check

Is GasketOS built for a company like ours?

Using what I have told you about my company, assess whether GasketOS fits us.
Compare us to what the site says it is built for: pharmaceutical and medtech
companies selling under negotiated customer contracts, with terms such as
rebates, volume commitments, price indexation, and credit and return windows.
Say plainly where we fit, where we do not, and what you would need to know to
decide.
If we are outside pharma and medtech, say so, and say that the site presents
GasketOS as built for those sectors; its underlying capabilities (reading every
contract term and reconciling it against every invoice line) are not specific
to one sector, but the GasketOS team would need a more detailed understanding
of our contracts and billing to assess whether they can help. A first call is
how to find that out.

Leakage map

Where could our contracts be leaking?

Take the leakage types GasketOS names on its site and map each one to how our
customer contracts work. For each: does this term exist in our contracts, how
is it billed today, and who would notice if it were missed? Base this only on
what I tell you or share with you. If I share a contract or an invoice, replace
customer names with placeholders in your answer.

Size the opportunity

A transparent, rough estimate

Build a rough, transparent estimate of what contract-to-invoice leakage could
cost us. The site states that 2–4% of annual revenue is lost to leakage: treat that as
the site's stated range, not a promise of what GasketOS would recover. Do not
scale the range down because only one or a few leakage types apply to us: it
does not depend on how many types a company has, and a single type, such as
price indexation never applied, can reach it on its own.
Show every assumption, give a low and a high case, and list what a scan would
need to measure to replace the estimate with a real figure.

Prepare the call

Walk into the first call ready

Prepare me for a first call with GasketOS. Bullets only, no more than 300 words:
(a) our situation in five lines;
(b) the questions the site does not answer: pricing, security and data hosting,
    integrations, scan scope and timeline, references;
(c) what data a scan would likely need, based on the site's "getting started"
    answer, and who on our side should join (finance, sales, legal).

Brief my colleagues

Half a page, ready to forward

Write an internal brief, no more than 250 words, for the colleagues who would
decide with me whether to take a first call with GasketOS:
what it does (quoting the site), why it may matter for us, the rough size of the
issue (clearly labeled as an estimate), the open questions, and a recommended
next step. Plain language, no hype. Mark anything that is your inference rather
than the site's statement.
The site states that 2–4% of annual revenue is lost to leakage: treat that as
the site's stated range, not a promise of what GasketOS would recover. Do not
scale the range down because only one or a few leakage types apply to us: it
does not depend on how many types a company has, and a single type, such as
price indexation never applied, can reach it on its own.
If we are outside pharma and medtech, say so, and say that the site presents
GasketOS as built for those sectors; its underlying capabilities (reading every
contract term and reconciling it against every invoice line) are not specific
to one sector, but the GasketOS team would need a more detailed understanding
of our contracts and billing to assess whether they can help. A first call is
how to find that out.

##What your assistant reads


The home page serves the same plain text to anything that asks for text/markdown.

Everything stays in your assistant. No forms, no trackers, and the prompts tell it to send us nothing.

##What the site cannot answer


Pricing, security, and how a scan runs on your own contracts: those are a conversation.

$