Flagship synthetic workflow
A deal team uses AI to draft an investment memo for a synthetic company. The draft includes revenue growth, margin expansion, market leadership, customer concentration, ESG, non-GAAP, and forecast claims. QEL checks which statements can survive against filings, model tabs, reconciliations, market excerpts, and source assumptions.
The draft
The memo says Atlas grew revenue 18%, improved gross margin, has 21% adjusted EBITDA margin, is the market leader, has no meaningful customer concentration, is ESG-compliant, and offers downside-protected returns.
The risk
A polished memo can carry unsupported financial metrics, misleading superlatives, stale market claims, unqualified non-GAAP measures, or forecast language that should require review. QEL makes the admission record visible before those claims become final work.
What QEL controls
QEL extracts claims, maps evidence, applies the active rule pack, builds a registry, and compiles survived output only from admitted or caveated records.
Why QEL is needed
A polished memo can carry unsupported financial metrics, misleading superlatives, stale market claims, unqualified non-GAAP measures, or forecast language that should require review. QEL makes the admission record visible before those claims become final work.
The draft reads polished before its material claims are qualified.
Some claims need exact source spans, not just a credible nearby document.
Unsupported or review-required claims can be copied forward unless a firewall stops them.
Synthetic Draft Preview
The memo says Atlas grew revenue 18%, improved gross margin, has 21% adjusted EBITDA margin, is the market leader, has no meaningful customer concentration, is ESG-compliant, and offers downside-protected returns.
Full synthetic source document
Synthetic source document inspected by QEL.
Synthetic QEL demonstration material - not investment advice - not securities offering material - not compliance certification.
Evidence Packet
Type: synthetic financial table
Supports: Revenue growth, margin movement, debt line.
Cannot support: Market leadership, ESG methodology, or investment recommendation.
Type: calculation support
Supports: Adjusted EBITDA margin and adjustment line.
Cannot support: Unqualified audited/GAAP claims.
Type: market source
Supports: Market size, source date, rank.
Cannot support: Current leadership without caveat.
Type: risk support
Supports: Top customer and top-three concentration.
Cannot support: Diversification or no-concentration claims.
Configurable Rule Pack
The active synthetic investment memo pack requires metric-level source support, non-GAAP reconciliation, market-source dates, peer-set caveats, and review gates for projections and recommendations.
Synthetic sample rule pack only. Not official, not endorsed, not securities guidance, not accounting advice, and not a substitute for qualified human review.
{
"pack_id": "synthetic_sec_investment_memo_claim_support_pack_v0",
"display_name": "Synthetic SEC / Investment Memo Claim Support Pack",
"version": "0.1.0",
"domain": "financial_services",
"workflow": "investment_memo_pitch_book_claim_review",
"status": "synthetic_demo_only",
"not_official": true,
"synthetic_demo_only": true,
"intended_users": [
"investment banking",
"private equity",
"corporate development",
"investor relations",
"research compliance",
"audit"
],
"compatible_outputs": [
"pitch book",
"investment memo",
"transaction deck",
"ESG summary",
"IC memo"
],
"required_evidence_types": [
"filing excerpt",
"financial statement table",
"model extract",
"non-GAAP reconciliation",
"market report",
"peer comps",
"ESG policy"
],
"decision_states": [
"admitted",
"admitted_with_caveat",
"blocked",
"review_required"
],
"rule_categories": [
"GAAP metric claims",
"non-GAAP metric claims",
"market claims",
"customer concentration claims",
"peer comparisons",
"ESG claims",
"forecast claims",
"investment recommendation language",
"promotional claims"
],
"human_review_gates": [
"forecast/projection claim",
"ESG methodology claim",
"investment recommendation",
"non-GAAP presentation"
],
"blocked_language_patterns": [
"guaranteed returns",
"downside protected",
"market leader",
"best-in-class",
"no meaningful concentration"
],
"caveat_triggers": [
"derived calculation",
"non-GAAP reconciliation",
"outdated source date",
"partial peer set"
],
"freshness_requirements": [
"current filing period",
"model date",
"market report date",
"peer comp date"
],
"source_span_requirements": [
"source name",
"period",
"unit/currency",
"line item",
"calculation formula",
"assumption"
],
"audit_artifacts_generated": [
"claim map",
"evidence/source-span map",
"admission decision table",
"registry view",
"blocked/review appendix",
"survived output",
"provenance/audit summary"
],
"final_output_policy": "Final output allowed only for admitted or admitted_with_caveat registry records.",
"disclaimer": "Synthetic sample rule pack only. Not official, not endorsed, not securities guidance, not accounting advice, and not a substitute for qualified human review."
}Claim Firewall Process
Results dashboard
QEL separated admitted, caveated, blocked, and review-required claims before compiling survived output.
Top findings
admitted
Why it matters: Core financial metrics can survive when values, period, and calculation match.
Evidence / rule reason: FY25 118.0 versus FY24 100.0 supports the calculation.
Survived final output: yes
admitted_with_caveat
Why it matters: Non-GAAP statements need reconciliation and scope language.
Evidence / rule reason: Reconciliation exists, but the rule pack requires a caveat.
Survived final output: yes, with caveat
blocked
Why it matters: Unsupported superlatives can create buyer, board, and review risk.
Evidence / rule reason: Market report ranks Atlas third.
Survived final output: no
blocked
Why it matters: Concentration risk should not be erased by confident memo language.
Evidence / rule reason: Top customer is 31 percent of revenue.
Survived final output: no
review_required
Why it matters: Forecast claims need assumptions, sensitivity context, and reviewer gates.
Evidence / rule reason: Model input exists but final forecast language requires review.
Survived final output: no
blocked
Why it matters: Guarantee-like investment language should not survive a governed memo.
Evidence / rule reason: Blocked language pattern in the active rule pack.
Survived final output: no
Claim Admission Report
QEL does not compile final output from raw AI text, unchecked draft language, or a final model polish step. Survived output is compiled only from registry records with admitted or admitted_with_caveat decisions and final_output_allowed = true.
Rule Pack Marketplace Preview
This synthetic pack previews how QEL could support industry-specific admission controls. In a future rule-pack ecosystem, teams could configure, review, version, and govern packs for their own workflows.
This is a synthetic preview only. The pack is not official, certified, endorsed, or a production claim.