Reused proposal snippets carry stale claims into new pitches

The problem

Teams repeatedly search old decks and may reuse unsupported case results, expired offers or another client’s confidential details.

Agent job

Index reusable passages by service; attach source date, owner and permission status; quarantine conflicting numbers and confidential examples rather than picking a winner.

Expected output

Searchable claim library, reusable evidence-backed snippets and a review queue.

Why an agent / simpler alternative

Assessment

The agent can group differently phrased claims and identify apparent contradictions across proposals. A normal document search works better when a clean, approved library already exists. Humans must determine whether evidence, client permission and commercial claims are valid.

Setup requirements

Items

1. Redacted copies of agency-owned proposals in a bounded folder

2. Current offer descriptions and a named claim approver

3. A permission register for client names, logos and results

4. Spreadsheet or searchable local document format; no CRM connection needed for the pilot

Inputs and access

1. Name

   Redacted agency-owned proposal archive

   Source or owner

   Prospective consenting customer or specified official publisher

   Sensitivity

   Use synthetic, public or appropriately deidentified material; confirm actual sensitivity before work

   Permission needed

   Owner authorizes scoped access and processing; verify rights and retention before receiving files

2. Name

   Approved current service descriptions

   Source or owner

   Prospective consenting customer or specified official publisher

   Sensitivity

   Use synthetic, public or appropriately deidentified material; confirm actual sensitivity before work

   Permission needed

   Owner authorizes scoped access and processing; verify rights and retention before receiving files

3. Name

   Claim evidence and reuse permissions

   Source or owner

   Prospective consenting customer or specified official publisher

   Sensitivity

   Use synthetic, public or appropriately deidentified material; confirm actual sensitivity before work

   Permission needed

   Owner authorizes scoped access and processing; verify rights and retention before receiving files

Proposed steps

1. 1

   Inventory supplied proposals and classify confidential sections before indexing.

   Expected output: Document and permission manifest

   Approval gate: Owner approves processing; remove restricted client material.

2. 2

   Extract candidate reusable claims with document/page, date and exact supporting evidence location.

   Expected output: Claim-source table

   Approval gate: Do not turn a claim into an approved fact.

3. 3

   Group variants by topic and compare figures, scope, periods and definitions.

   Expected output: Conflict clusters

   Approval gate: Different client cohorts remain separate until a human reconciles them.

4. 4

   Ask the owner to mark each item approved, expired, restricted or unresolved.

   Expected output: Signed-off status register

   Approval gate: Unknown permission or evidence means no reuse.

5. 5

   Draft short reusable snippets only for approved claims and build search tags.

   Expected output: Draft snippet library

   Approval gate: Preserve caveats; no new success metrics or client endorsements.

6. 6

   Run five retrieval tasks and inspect every output for restricted text before delivery.

   Expected output: Retrieval test notes and final library

   Approval gate: Approve library recipients; sending any pitch remains separate.

Synthetic output illustration — not an executed result

Synthetic

Yes

Executed

No

Text

SYNTHETIC: Claim C014, “Reduced onboarding time by 30%,” appears in three decks with two baselines. Status: QUARANTINED. Evidence owner: agency lead. Reuse permission: unknown. Approved alternative: none. These claims and records are invented.

Human approval boundaries

1. Agency owner approves reusable claims, rights and confidentiality; separately approve any new pitch or external delivery.

2. Human service provider reviews quality, scope and quoted fee; customer approves contract, delivery recipients and payment terms. Separately approve purchases, credentials, publication and external sends. No commitments or payments are made in this research.

Failure modes

1. Old client permission may not cover reuse

2. Similar claims can refer to different populations

3. A polished snippet can hide weak evidence

Recovery and troubleshooting

Items

1. If source evidence is missing, keep only an unresolved index entry and remove the claim from reusable snippets.

2. If redaction is incomplete, stop processing and request a clean export rather than distributing the library.

3. If retrieval returns expired content, correct status filters and rerun the same five tasks.

Unexecuted validation plan

Sample setup

Index five invented proposals with expired prices, two conflicting claims and one confidential example.

Acceptance checks

1. Every claim has provenance and date

2. Conflicts and restricted content are quarantined

3. Owner retrieves all five target facts without opening every deck

Stop conditions

1. Stop if data authority or rights are unclear.

2. Stop if critical seeded failures are missed; do not claim readiness.

3. Stop if manual correction time exceeds the pilot effort budget without customer approval.

Executed

No

Costs and effort — unknown or hypothetical

Status

unknown_or_estimate

Currency

CAD

Items

1. Agent subscription/API usage: not quoted; verify current plan and commercial data terms.

2. Human intake, rights review, QA and revisions: unknown until timed pilot.

3. Storage, hosting, integration and payment-processing costs: excluded unless separately agreed.

4. Customer acquisition and support time: not measured.

Caveat

No margin or income estimate. Price hypotheses are editorial offer tests, not observed market rates; taxes, scope and effort require a real quote.

Proposed service concept — earnings not verified or guaranteed

Buyer

Small B2B agency with a folder of old proposals

Deliverable

Searchable claim library, reusable evidence-backed snippets and a review queue.

Charging basis

One-off proposal-library cleanup plus separately agreed quarterly refresh.

Acquisition channel

Permission-based discovery with existing professional contacts, referrals or an opt-in sample page; none executed.

Assumptions

1. Customer has this unresolved problem and values a reviewed deliverable above existing built-in alternatives.

2. A consenting pilot confirms data access, achievable quality and economically tolerable review time.

3. Provider can deliver within a written scope and supports the result.

Revenue guarantee

No

Canadian adaptation — proposed

Proposed Canadian pilot: separate CAD/USD amounts and English/French claims; preserve geography and date scope of Canadian client results. Have the agency confirm reuse rights and local client permissions before a snippet becomes reusable. No inferred Canadian deployment or legal assurance.

Limits

1. Original proposed plan; untested.

2. Problem and capability references are evidence of context only, not demonstrated end-to-end success, customer demand, willingness to pay or profitability.

3. Compatibility, workload, delivery effort and availability must be verified in a real authorized pilot.

4. Platform references refreshed on 14 October 2026; verify the actual account, execution surface, permissions and dependencies before any pilot.

Candidate platforms — compatibility untested

1. Name

   Claude (Cowork capabilities)

   Required capabilities

   1. Read supplied files and produce editable documents and spreadsheets.

   Capability source urls

   1. https://support.claude.com/en/articles/13345190-get-started-with-claude-cowork

   2. https://support.claude.com/en/articles/16761823-claude-cowork-and-chat-are-one-claude

   Workflow compatibility

   Unknown; verify in the proposed first test.

   Historical candidate name

   Claude Cowork

   Selection basis

   Editorial candidate from general capabilities; exact end-to-end fit untested.