AIforBC source-backed guide
Check an AI-assisted quote before it becomes a promise
A source-checking checklist for quantities, units, price, timing, exclusions, and approval, with a fictional flawed quote.
By AIforBC, operated by Boost Commerce Group. AI-assisted educational material; no individual expert review is claimed. Editorial policy and corrections →
Practical answer
Treat a generated quote as an unapproved draft. Check every commercial statement against the current price record, customer request, and approval rules. Keep estimates and conditions visible; do not let fluent wording invent a commitment.
A fictional example you can check
Fictional source: 12 boxes, 20 fasteners per box; approved list price $38 per box. Delivery is estimated Tuesday, subject to dispatch confirmation. Freight is not yet priced. No discount is authorized.
Reviewable resultFlawed draft: ‘240 boxes at $38, including freight, guaranteed Tuesday. We have applied your 10% loyalty discount.’ Checked position: 12 boxes / 240 fasteners; $456 before applicable taxes and unconfirmed freight; timing estimated, not guaranteed; no discount approved.
The $456 subtotal is simple fictional arithmetic: 12 × $38. It is not a real quote. Units, freight, certainty, and discount authority all need separate checks; correcting just the multiplication leaves serious errors behind.
Original AIforBC teaching example. No customer record, field test, or measured business result is represented.
Practice / compare with the source above
What would you change before using this?
Read the draft before the corrections. These are deliberately flawed authored examples, not model output. Use the corrected draft for practice. It does not authorize operational use.
Deliberately flawed draft
Your order is 240 boxes at $38 per box. Freight is included, Tuesday delivery is guaranteed, and your 10% loyalty discount has been applied.
Trace each correction
“240 boxes”
Why it fails: The request is 12 boxes with 20 fasteners in each. The draft changes the purchasing unit.
Checked position: 12 boxes = 240 fasteners. Fictional subtotal: 12 × $38 = $456 before applicable taxes and unconfirmed freight.
“Freight is included”
Why it fails: Freight is explicitly not yet priced. ‘Included’ invents a cost commitment.
Checked position: Freight: unconfirmed; obtain the approved amount and terms before release.
“Tuesday delivery is guaranteed”
Why it fails: The source describes an estimate subject to dispatch confirmation, not a guarantee.
Checked position: Tuesday is an estimate only; dispatch confirmation remains open.
“10% loyalty discount”
Why it fails: No discount is authorized. A polished customer-friendly sentence cannot create pricing authority.
Checked position: No discount applied. Ask the designated approver about any requested adjustment.
A corrected practice draft
Internal review draft — not an approved quote: 12 boxes containing 20 fasteners each, 240 fasteners total. At the fictional approved list price of $38 per box, the subtotal is $456 before applicable taxes and unconfirmed freight. No discount is authorized. Tuesday delivery is estimated and subject to dispatch confirmation. Resolve freight, applicable tax treatment and dispatch details through the normal approval process before issuing a quote.
Review before handing it on
- Check quantities and packaging units against the request.
- Recalculate the subtotal independently from the current price record.
- Confirm discount authority; do not infer it from customer history.
- Make freight, applicable tax treatment and timing conditions visible.
- Record who approved the final version before it is released.
Put the method to work
1. Freeze the reference set
Use the current customer scope, approved product specification, price version, freight information, and internal terms. Note which records remain unconfirmed. A draft should not combine a current request with an obsolete price sheet.
2. Check the numbers independently
Compare quantity, unit, packaging, currency, subtotal, discounts, and tax treatment with the approved records. Recalculate arithmetic outside the prose generator. Flag any number that has no source, rather than treating it as a reasonable approximation.
3. Underline every promise
Inspect words such as guaranteed, included, approved, compliant, available, and waived. For each, identify the approving source and person. Change an estimate back to an estimate and remove unsupported inclusions. Use the organization's established terms rather than drafting new legal protections.
4. Check exclusions and open conditions
For the example, freight and dispatch confirmation remain open. Make these visible in the review record and resolve them before issuing an approved quote when required. A hidden caveat at the end cannot repair an incorrect guarantee in the opening sentence.
5. Record approval and keep the trail
Use the ordinary quote-approval process, including who checked and released it. Save the final version and its references in an approved system. During a trial, record which defects were caught and whether a second reviewer was needed.
Reusable worksheet
Copy this into an approved workspace. Start with fictional information; do not paste confidential records into an unapproved AI tool.
Quote version / source versions: Quantity and unit verified: Arithmetic and currency checked: Price / discount authority: Freight / taxes / exclusions: Timing and stock confirmation: Unsupported promises removed: Approved by / date / release status:
Where this method stops
This checklist is not legal, tax, engineering, or procurement advice. It does not determine enforceability or product fitness. It is a draft-review aid, not authorization to quote or change your organization's terms.
Your next action
Source trail
Primary sources used in this guide
- Generative artificial intelligence — ITSAP.00.041Canadian Centre for Cyber Security
Context for protecting sensitive information and checking generated output. It does not validate these fictional examples or promise business results.
- AI Risk Management FrameworkNational Institute of Standards and Technology
Voluntary risk-management context. Our pilot worksheet is an original educational example, not a NIST certification or endorsed method.
AIforBC uses official and primary sources where practical. This guide provides general operational education, not legal, privacy, cybersecurity, or financial advice.
Try the method
