The offer, the price and the close

When a product has no guide of its own, pick the offer below that matches the prospect's problem, or let the founder pick.

What you sell

What the client receives

• Their own API key, organisation and history (the public demo is shared and limited).
• Their own rules, tools and agents wired in, with the founder's help on a call.
• Every decision signed, so at the end of 30 days they hold a record of what was checked.
• A 30-minute review at the end: what it caught, what it did not, and what to do next.

What the offer itself says about its limits

• A pilot is a trial of a working product, not a production contract or a service-level agreement.
• No uptime guarantee; the hosted service runs on our own servers.

Price source: Homepage LinkedIn products hub ($500 30-day pilot checkout). Prices are the founder's proposed list prices, not validated market rates; any discount needs the founder's OK.

$500 pilot checkout Send this link after the call; then email vincent@firmeai.com with the client name.

Also offer

What the client receives

• Tool-schema scan: injection in descriptions, secrets, unconstrained exec/sql/path parameters.
• Tool-poisoning, typosquat and authentication findings.
• Authority boundary and action-firewall gap analysis.
• Ranked remediation with the smallest patch, plus a signed bundle.

What the offer itself says about its limits

• Heuristic static analysis of definitions; a clean scan is not a guarantee.
• We do not probe a live production server without written scope.
• Findings are ranked by blast radius, not by count.

Price source: cain_audit/products.py → mcp-review. Prices are the founder's proposed list prices, not validated market rates; any discount needs the founder's OK.

Also offer

What the client receives

• report.md: every module graded REAL / PARTIAL / STUB / UNKNOWN, with the exact lines that caused each grade.
• findings.json: the same result for the client's engineers, machine readable.
• evidence.zip: a signed record plus a small checker that imports none of our code.
• A list of the files that were clean, and a section of notes from the person who reviewed it.

What the offer itself says about its limits

• The detector is precise but narrow: it misses many kinds of fake code. A clean grade is not a guarantee.
• A person reviews every finding before the report goes out.

Price source: Operator playbook stream 1 (docs/strategy/CAIN-AS-A-BUSINESS-OPERATOR-PLAYBOOK.md); delivered by cain_audit. Prices are the founder's proposed list prices, not validated market rates; any discount needs the founder's OK.

Also offer

What the client receives

• Nine buyer requests answered against this product, each with evidence or an explicit gap.
• An exportable record of what the agent can do and what it did on the test run.
• A signed bundle the buyer's security team verifies themselves.
• A short list of the answers that cannot be supported today.

What the offer itself says about its limits

• Screenshots and slideware are scored as a fail by buyers; this pack is evidence, not marketing.
• A control we could not observe is recorded as NOT_EVIDENCED, never as present.
• This is not SOC 2, not a certification, and not legal advice.

Price source: cain_audit/products.py → safety-pack. Prices are the founder's proposed list prices, not validated market rates; any discount needs the founder's OK.

What you need from the client

• What the prospect's problem is, in their words.
• Which product section they looked at.

How to close

1. Ask which problem matters most this month.
2. Match it to one offer below.
3. Hand to the founder with the lead details.

Copy you can send

First email

Subject: a quick question about your AI agents

Hi [first name],

Companies that let AI agents do real work are starting to get a new question from
their customers: "what controls do you have on those agents?" We help answer it: we stop the dangerous
actions before they happen and give you evidence your customer can check themselves.

Fixed price, starting at $500. Worth a 15-minute call?

[your name], working with FIRME AI CORPORATION

LinkedIn connection note (under 300 characters)

Hi [name], I work with a team that helps companies control what their AI agents are allowed to do, and prove it to customers. Would a 15-minute look be useful?

Follow-up, three days later

Hi [first name], one concrete thing: if any of your AI agents can touch money or customer data, we can show you in
five minutes what it could be tricked into doing. Worth it?

Fill in the brackets by hand for each person. Never paste the same message to a list.

Hand-off to the founder

Email vincent@firmeai.com with: company, person and role, what they said in their own words, which offer, the price you quoted, and the agreed next date. The founder handles scoping, written authorisation to test, invoices and delivery.

Next: the rules, the glossary and the quiz →

FIRME AI CORPORATION · CAIN-42 is a Byzantine trust fabric. Numbers marked "our own check" were run by us, not by an independent party. Nothing on these pages is a certification.