How to Build a Quote/Proposal Assistant That Writes Like You

Outcome: on-brand proposals in minutes

The playbook below turns your proposals into a fast, consistent system that speaks in your voice, pulls the right blocks, prices correctly, routes for approval, and ships as a PDF or email—without you babysitting every draft.

1) Create your Brand Voice File (the “how we talk” layer)

This is the compass the assistant uses before it writes a single sentence.

What to include

• Tone & style: direct, friendly, punchy; avoid buzzwords; use second person (“you”)
• Positioning: who we serve, primary value, proof points
• Phrases to use/avoid: “plain-English ROI,” avoid “synergy,” etc.
• Formatting rules: headings, bullet density, callouts, signature block
• Examples: 2–3 redlined proposals (good vs. not-us)

Example (YAML or JSON works):

brand_voice:
  tone: "plain-spoken, confident, Gen X concise"
  persona: "ops-minded AI consultancy for small businesses"
  value_props: ["faster ops", "lower costs", "practical AI adoption"]
  do_say: ["playbook", "guardrails", "measurable"]
  dont_say: ["cutting-edge synergy", "paradigm-shifting"]
  formatting:
    headings: "H2 > H3"
    bullets_per_section: 3-6
    callouts: "bold + italic for key numbers"
  examples:
    - good_url: "internal://proposal-good.pdf"
    - bad_url: "internal://proposal-bad.pdf"

2) Build Modular Blocks (plug-and-play content)

Think Lego, not novels. Each block is short, single-purpose, and versioned.

Core block types

• Scope: what’s included/excluded (per package)
• Approach/Method: how you do it (1–2 paragraphs)
• Timeline: start → milestones → delivery
• Proof: case blurb, metrics, testimonial
• Risks & Assumptions: what must be true
• Legal Lite: MSA reference, IP notes, termination

Example block

{
  "id": "scope-ai-inbox-v1",
  "title": "Scope: AI Inbox Pilot",
  "use_when": ["lead_gen", "support_inbox"],
  "includes": [
    "Set up shared inbox assistant with guardrails",
    "10-day pilot, 100 sample messages",
    "Weekly review + tune prompts"
  ],
  "excludes": ["Custom integrations beyond Gmail/Outlook"]
}

3) Define Pricing Variables (no more manual math)

Turn pricing into parameters your assistant can calculate safely.

Inputs to capture

• Package (Starter/Pro/Custom)
• Seats / volumes / SKUs
• One-time vs monthly
• Discounts & approval thresholds
• Taxes/fees rules by region

Example

{
  "pricing_model": {
    "starter": {"monthly": 1500, "setup": 1000, "limit_messages": 1000},
    "pro": {"monthly": 3500, "setup": 2500, "limit_messages": 5000}
  },
  "variables": {
    "messages": 3200,
    "chosen_plan": "pro",
    "discount_pct": 10
  },
  "rules": {
    "auto_discount_max_pct": 10,
    "manager_required_over": 20000
  }
}

4) Approvals & Guardrails (the “don’t let me get sued” layer)

Automate checks before anything leaves your house.

Automated gates

• Price floor: block drafts below margin X%
• Scope mismatch: flag if block set conflicts (e.g., “pilot” + “24/7 SLA”)
• Legal tags: add data-processing addendum when PII present
• Brand voice lint: reject banned phrases

Human approvals

• Sales: pricing > $20K or discount > 10%
• Legal: new logo/likeness, custom IP clauses
• Ops: timeline under 2 weeks

5) Draft → PDF/Email (push-button packaging)

The assistant assembles the blocks, applies voice, fills pricing, inserts your visuals, and exports to your preferred wrappers.

Outputs

• Proposal PDF: cover → summary → scope → timeline → price → next steps
• Email draft: tight summary + CTA + calendar link
• Editable doc: for last-mile tweaks

Email skeleton

Subject: Proposal: <Outcome> in <Timeframe>

Hi <Name> — here’s the plan to <business outcome>.
Highlights:
• Start <date>; first win by <milestone>.
• Fixed monthly: $X; setup: $Y; cancel anytime after month 1.
• Guardrails: privacy-first, human-in-the-loop.

PDF attached. Want me to pencil in <Day/Time> for a walkthrough?
— Alex, BoostMyAI

Putting It Together (flow)

1. Intake: client type, goals, constraints → auto-select recommended blocks
2. Compose: assistant drafts with brand voice + merges blocks
3. Price: compute from variables; apply rules/discounts
4. Check: run guardrails; route approvals if triggered
5. Export: PDF + email; log to CRM

Minimum Viable Build (1–2 days)

• Day 1: voice file v1, 10 core blocks, pricing JSON, simple approval rules
• Day 2: generate 3 proposal templates, wire CRM sync, PDF/email export

Prompts That Work (steal these)

• “Using brand_voice.yaml and these blocks: scope-ai-inbox-v1, timeline-standard-v2, write a 1-page proposal for a 10-day pilot. Keep bullets tight, avoid buzzwords, include a measurable next step.”
• “Draft the email summary in 110 words or less, with a single CTA to book a call. Do not summarize internal pricing mechanics.”
• “Run the price engine with plan=pro, messages=3200, discount=10%. Flag any rule violations, then render a pricing table.”

KPIs to Track

• Time to draft (target: <10 minutes)
• Close rate vs baseline
• Discount % by segment
• Revisions per deal
• Proposal → meeting booking rate

Common Pitfalls (and fixes)

• Bloated blocks: Keep them short and versioned; retire low-performers.
• Voice drift: Re-train the voice file monthly using winners only.
• Pricing chaos: Lock variables; require approvals for custom math.
• PDF purgatory: Always keep an editable source (DOCX/HTML) behind the PDF.

Ready-to-Use Section Outline (copy/paste)

1. Executive Summary – outcome in one paragraph
2. What You Get – 5–7 bullets (scope)
3. How We Deliver – short method + timeline
4. Investment – table with setup, monthly, options
5. Assumptions & Guardrails – 5 bullets
6. Next Step – CTA with calendar link

Why This Works

You remove creativity from the parts that shouldn’t vary (voice, structure, math) and keep creativity where it closes (fit, proof, CTA). The result: every proposal sounds like your best one—just faster.

Want it built?

We can stand this up for your team, wire it into your CRM, and have you shipping on-brand proposals this week. Contact us today to get started!