Day 21 — AI Engineering
You spend 4 hours writing a proposal. The client spends 4 minutes reading it.
An MCP server that drafts SOWs, scopes, and pricing from a brief. Give it a client and what they need — get back a full proposal with sections, line items, and a timeline. You edit and send. Done.
Example Queries
Tools
| Tool | What It Does |
|---|---|
| generate_proposal | Generate a new proposal from a client brief. Auto-selects template, recommends services, estimates pricing, and drafts every section. |
| list_proposals | List and search proposals by status, client, or date. Shows pipeline overview with total values at each stage. |
| get_proposal | Get the full proposal document — all sections, line items, client details, and totals. Ready to review or edit. |
| list_services | Browse service offerings with rates, typical hours, and deliverables. Filter by category to build the right scope. |
| list_templates | View available proposal templates — SOW, quick proposal, discovery, retainer. Shows sections, tone, and when to use each. |
| estimate_scope | Estimate hours, cost, and timeline from requirements. Returns low/mid/high estimates with budget and timeline feasibility. |
Pipeline
Whitfield & Associates · Legal
NovaByte · SaaS
Verdant Wellness · Health & Wellness
Bright Futures Foundation · Nonprofit
Coastal Realty Partners · Real Estate
Summit Construction Group · Construction
Why This Build
Every consultancy has the same bottleneck: writing proposals. You finish a great discovery call, and then you spend 4 hours formatting a document the client will skim in 4 minutes. Meanwhile, two other leads go cold because you didn't follow up fast enough.
Most proposal tools are either glorified templates (you still write everything) or fully automated garbage (the client can tell). The sweet spot is in between — generate the structure, the pricing, the sections, and the scope automatically, then let a human refine the parts that matter.
This server manages six clients, eight service offerings, four proposal templates, and a pipeline of proposals at every stage. Ask it to generate a proposal from a brief and it'll pick the right template, recommend services, estimate hours, calculate pricing, and draft every section. You edit, you send, you close.
Demo Data
6
Clients
8
Services
4
Templates
6
Proposals
12
Sections
14
Line Items
Six clients across legal, SaaS, health & wellness, construction, nonprofit, and real estate. Eight service offerings from MCP development to strategy consulting. Four proposal templates (SOW, quick proposal, discovery, retainer). A realistic pipeline with accepted, sent, draft, and declined proposals — including one client who went with a cheaper competitor.
Setup
cd day21-ai-proposal-writer
npm install && npm run seed && npm run buildThen add the config to Claude Desktop and restart.
If your proposal process is a bottleneck between “great call” and “closed deal,” let's talk about automating the tedious parts.