Enterprise proposal generator

Enterprise Proposal Generator

A step-by-step guided skill that produces a tailored proposal for enterprise and large-merchant prospects. Architecture is hybrid — this SKILL.md guides Claude through gathering inputs and writing merchant-specific content; a deterministic Python renderer (shared/outputs/proposal_deck.py) then builds the 12-slide PPTX with consistent FinVerge branding.

Direction of the proposal

Enterprise proposals go FinVerge → Merchant. FinVerge is the vendor offering payments infrastructure; the merchant is evaluating us against alternatives (Razorpay, Cashfree, PayU, etc.). The cover tagline is "Built for Enterprise" (not "Banking Collaborations" — that's for banks).

When to use

Use whenever the user wants a proposal / pitch / deck for a merchant prospect:

• "Make a proposal for BeYoung"
• "Enterprise pitch for [SaaS company]"
• "Draft a deck for our D2C lead tomorrow"
• "Proposal for [NBFC] on NACH and PG"

Do NOT use this skill for:

• Bank proposals → use bank-proposal-generator
• Single-product decks (PG demo, NACH FAQ, etc.) → use the respective product skill (pg-product, nach-product, collect-product, orchestrate-product)
• Referral partner proposals → (future: referral-partner-proposal-generator)

Core design principle

Conversational, step-by-step. Ask one thing at a time. If the user provides some inputs upfront ("proposal for BeYoung, D2C, ₹5 Cr/mo GMV, PG+Collect, CFO audience, PPTX"), skip those steps and only gather what's missing.

Step 1: Merchant name + website domain

Ask:

Who's the merchant (display name), and what's their website domain?

The domain is used to fetch their logo via Clearbit. Example: "BeYoung, beyoung.in".

Normalize the merchant name to a lookup key (lowercase alphanumeric — "BeYoung" → "beyoung"). Check whether a logo is already cached by inspecting shared-assets/logos/merchants/INDEX.md. If not cached, note this for Step 7.

Step 2: Segment

Ask:

Which segment does this merchant fall into?

1. E-commerce — online shopping, marketplaces
2. SaaS — subscription software, cloud services
3. D2C — direct-to-consumer brands (apparel, beauty, FMCG)
4. NBFC — lending, EMI collections, financial services
5. Education — schools, ed-tech, coaching, upskilling
6. Insurance — premium collections, claims
7. Subscription — any recurring-billing business
8. B2B Marketplace — multi-seller platforms, payouts
9. Government / PSU — fee collection, tax, BBPS

Read the corresponding file: assets/proposal-kb/segments/<segment-key>.md (e.g., d2c.md). This gives segment-specific positioning, pain points, recommended product mix, and typical objections.

Valid segment keys: ecommerce, saas, d2c, nbfc, education, insurance, subscription, b2b-marketplace, government.

Step 3: GMV range

Ask:

What's the merchant's approximate monthly GMV range?

• Under ₹50 L/month — Starter tier
• ₹50 L – ₹5 Cr/month — Growth tier
• ₹5 Cr – ₹50 Cr/month — Scale tier
• ₹50 Cr+ / month — Enterprise tier

GMV drives pricing in assets/proposal-kb/pricing-tiers-by-gmv.md. Map to a gmv_range key: under_50l, 50l_to_5cr, 5_to_50_cr, above_50cr.

Step 4: Products to propose

Ask:

Which FinVerge products should this proposal cover? Here's what's customer-sellable:

• PG (Payment Gateway 2.0) — digital + offline rails, 27+ banks
• NACH & Recurring — subscription, EMI, mandate lifecycle
• Collect (No-Code Platform) — Links, Pages, Buttons, Forms
• Orchestrate (Payment Orchestration) — multi-gateway routing, failover

Pick one or multiple. You can also include internal-only products (KYC, MerchantOS, Partner Dashboard) with a gap warning, or placeholder products (Collections, Connected Banking, BBPS COU, PA-P) — these will have [TO FILL] markers in the output.

Product keys used with the KB loader: pg, nach, collect, orchestrate, kyc, merchantos, partner-dashboard, collections, connected-banking, bbps-cou, pa-p.

Step 5: Audience / stakeholder

Ask:

Who's the primary audience for this proposal?

• CFO / Finance — cost structure, settlement, reconciliation
• COO / Operations — ops efficiency, reporting, SLAs
• VP Engineering / CTO — API design, integration effort, observability
• VP Growth / CMO — checkout conversion, customer experience
• Founder / CEO — strategic differentiation, roadmap fit
• Mixed / Not sure — balanced executive framing

Read references/enterprise-stakeholder-profiles.md for what each role cares about. This shapes the emphasis in the Executive Summary and Why-FinVerge-for-Segment sections.

Map to primary_role key: cfo, coo, vp_eng, vp_growth, founder, mixed.

Step 6: Specific pain points / context (optional)

Ask:

Any specific context for this merchant? (Optional — say "none" to skip)

Examples:

• "Currently on Razorpay, considering multi-gateway to improve success rate"
• "Growing fast, need better mandate handling for subscription revenue"
• "Manual reconciliation across 3 gateways is killing the finance team"
• "Need white-label checkout for their D2C brand"

This feeds the Merchant Context section of the proposal.

Step 7: Output format

For now, only PPTX is supported (DOCX + Canva come in v0.8.0). If the user asks for DOCX/Canva, tell them we'll ship PPTX this time and the other formats are coming.

Step 8: Confirm inputs, then generate

Summarise and confirm before running any tools:

Here's what I'll generate:

• Merchant: BeYoung (beyoung.in)
• Segment: D2C Brands
• GMV range: ₹5 Cr – ₹50 Cr/month (Scale tier)
• Products: PG, Collect
• Audience: CFO
• Pain point: Manual reconciliation across 3 gateways
• Format: PPTX

Proceed?

Generation flow (run once confirmed)

1. Resolve merchant logo

If the merchant logo is already cached (shared-assets/logos/merchants/<key>.png), skip. Otherwise run:

python3 shared/fetch_merchant_logo.py "BeYoung" --domain beyoung.in

The fetcher tries Clearbit first (works for most merchants with a website), falls back to Google favicon. If both fail, the proposal still renders without the merchant logo and a gap is flagged.

2. Load reference content

Read (only as needed):

• assets/proposal-kb/finverge-about.md — canonical company profile
• assets/proposal-kb/segments/<segment>.md — segment-specific positioning + pain points + recommended product mix
• assets/proposal-kb/pricing-tiers-by-gmv.md — pricing tier for the chosen GMV range
• references/enterprise-language.md — B2B tone guide (no RBI-heavy language)
• references/enterprise-stakeholder-profiles.md — per-role framing
• references/proposal-sections.md — section structure + per-stakeholder emphasis
• references/compliance-clauses.md — relevant compliance content for selected products
• references/deck-patterns.md — FinVerge house deck style

Use kb_loader.load_public(product, facet) for per-product content. DO NOT call load_internal() — this is a customer-facing output.

3. Write section content as a JSON blob

Assemble the following content, specifically tailored to the merchant, segment, and audience. Each field is plain prose (no markdown), 2-6 sentences unless otherwise noted.

{
  "merchant": {
    "display_name": "BeYoung",
    "domain": "beyoung.in",
    "segment": "d2c",
    "gmv_range": "5_to_50_cr"
  },
  "audience": {
    "primary_role": "cfo",
    "secondary_roles": []
  },
  "products": ["pg", "collect"],
  "sections": {
    "executive_summary": "<3-5 sentence summary: who FinVerge is, why it fits this merchant, expected outcome. Tailored to the CFO audience and the merchant's stated pain.>",
    "merchant_context": "<4-6 sentence framing of the merchant's business situation: their segment's common dynamics, the specific pain they raised in step 6, how FinVerge fits. No fluff.>",
    "why_finverge_for_segment": "<4-6 sentence, segment-specific differentiators. Pull from segments/<segment>.md + FinVerge's proven capabilities. Concrete, not generic.>",
    "next_steps": [
      "<Concrete action item 1 — something FinVerge commits to>",
      "<Concrete action item 2 — something the merchant needs to do>",
      "<Concrete action item 3 — commercial / timeline alignment>"
    ]
  },
  "include_pricing": true,
  "include_settlement_timelines": false,
  "include_revenue_sharing": false,
  "output_format": "pptx"
}

Save this blob as proposal_meta.json in the output directory (so the user can regenerate with tweaks later).

4. Invoke the renderer

python3 skills/enterprise-proposal-generator/scripts/generate_proposal.py \
    --meta /path/to/proposal_meta.json \
    --outdir ~/Downloads/enterprise-proposals/<merchant-key>-<YYYY-MM-DD>

The renderer:

• Validates the JSON against assets/templates/proposal_meta_schema.json
• Loads per-product KBs via load_public()
• Calls shared/outputs/proposal_deck.py which produces a 12-slide PPTX
• Surfaces any [TO FILL] markers in KB content as gaps
• Writes output + gaps.md + preserves the proposal_meta.json for future regeneration

5. Report back to the user

Summarise:

Generated: ~/Downloads/enterprise-proposals/beyoung-2026-04-17/Proposal.pptx

Gaps to fill (N items): [list them]

What's in the deck (12 slides):

1. Cover (BeYoung logo + FinVerge logo)
2. Executive Summary
3. About FinVerge (credentials)
4. BeYoung Context
5. Proposed Solution (PG + Collect overview)
6. PG Deep-Dive
7. Collect Deep-Dive
8. Integration & Architecture
9. Commercial Terms (if included)
10. Compliance & Security
11. Why FinVerge for D2C
12. Next Steps

Want me to adjust any section or regenerate with different inputs?

Handling edge cases

• Merchant logo fetch fails: proceed without logo, flag as gap
• Product KB has many [TO FILL] markers: surface all of them in gaps.md so sales can edit the deck before sending
• Internal-only product requested (KYC, MerchantOS, Partner Dashboard): warn that public content is [TO FILL] placeholder; still include at user's request
• Segment not recognised: ask user to pick from the 9 supported segments OR proceed with the "enterprise" generic segment (less tailored)

Reference files

• references/enterprise-language.md — B2B enterprise tone (no RBI fluff)
• references/enterprise-stakeholder-profiles.md — CFO, COO, VP Eng, VP Growth, Founder
• references/proposal-sections.md — 12-slide structure + per-stakeholder emphasis
• references/compliance-clauses.md — universal clauses (copied from bank-proposal-generator)
• references/deck-patterns.md — FinVerge house deck style

Asset files

• assets/proposal-kb/finverge-about.md — company profile
• assets/proposal-kb/pricing-tiers-by-gmv.md — GMV slab → MDR mapping
• assets/proposal-kb/segments/<segment>.md — 9 segment-specific positioning files
• assets/templates/proposal-template.md — 12-section content skeleton
• assets/templates/proposal_meta_schema.json — JSON schema for AI content

Scripts

• scripts/generate_proposal.py — orchestrator that validates meta + calls the shared renderer

Shared code reused

• shared/kb_loader.py — load_public(product, facet) for product content
• shared/brand.py — merchant_logo(name) + design tokens
• shared/layout.py — all slide primitives
• shared/slides/{cover, about, product, cta}.py — generic slide builders
• shared/outputs/proposal_deck.py — the multi-product deck renderer
• shared/fetch_merchant_logo.py — Clearbit-based logo fetcher
• shared/version_check.py — stale-version auto-notify