agentclaw

workflow: proposal writing

Three days to write a proposal is three days for the deal to go cold

The hours are the visible cost: a senior person digging through old decks, rewriting scope, hunting for the current rate card. The invisible cost is worse. Every day between 'send me a proposal' and 'sent', the prospect's urgency fades and someone else's document lands first. We install a system that assembles a first draft from your past winning proposals, your pricing rules, and the notes from the sales call — minutes after the call ends. Your team edits, approves, and sends.

the manual version

Where the time actually goes

Follow one proposal from 'great call' to 'sent' and count the steps that involve actual selling. Someone digs through shared drives for the last proposal that resembled this deal. Copies it. Swaps the client name and hopes they caught every instance. Rewrites the scope section from memory of a call that happened four days ago. Messages a delivery lead to confirm estimates, then waits. Hunts for the current rate card, or asks the founder what this should cost, then formats, proofreads, and routes it for internal review.

Do your own math on it. If a proposal takes six hours of combined effort across the people who touch it, and your firm sends eight a month, price those hours at loaded cost. Then add the part no timesheet captures: the deal that stalled while the draft sat half-finished, and the senior person who spent Tuesday assembling a document instead of running the next call.

Almost none of this work is judgment. It's retrieval, transcription, and arithmetic — finding what you've already written, restating what the prospect already said, and applying prices you've already set. That is exactly the work software should be doing.

  • The archaeology: searching drives and email for the closest past proposal to start from
  • The Frankenstein edit: pasting sections from three old documents and hoping the previous client's name doesn't survive
  • The pricing scramble: asking the founder what this service costs now, because the rate card lives in their head
  • The estimate chase: waiting on a delivery lead to confirm scope they already described on the call
agentclaw · workflow run

$ claw run invoice-intake

→ 47 documents queued

→ extracted · matched · posted

✓ done in 3m 12s · 0 exceptions escalated

the automated version

From call notes to a priced draft in your template

Three stages, wired into the tools you already sell with. The first draft is ready before the prospect's thank-you email.

  1. 01

    Read the deal

    An agent reads the call transcript or notes, the CRM record, and the email thread. It extracts what the prospect actually asked for: the problem, the scope hints, the timeline, the budget signals, the names of the people who will decide. Anything load-bearing that's missing doesn't get invented — the agent sends the rep two or three pointed questions instead of guessing.

  2. 02

    Assemble from past wins and pricing rules

    The agent searches your proposal library for the closest winning proposals by service, industry, and deal size, then pulls approved scope language, case descriptions, and team bios from them. Pricing is not left to the model's imagination: your rate card, minimums, tiers, and discount thresholds are encoded as rules and applied deterministically. The draft comes out in your template, in your voice, with your numbers.

  3. 03

    Flag, review, send

    The draft lands in your doc tool with flags on everything that needs a human call: pricing outside the rules, scope you've never delivered, terms your standard contract doesn't cover. Each flag routes to the right approver with the reason attached. A person edits, approves, and sends. Wins and losses get logged against the library, so next month's drafts start from stronger material.

Signs it's time to automate this

You don't need all six. Two or three usually means the math already works.

  • Proposals take days to go out even when the prospect was ready to buy on the call
  • Your most expensive people write proposals from scratch instead of selling
  • A proposal has gone out with another client's name still in it
  • Pricing varies by who wrote the document, not by what the deal deserves
  • Deals visibly stall in the gap between 'send me a proposal' and delivery
  • Your best proposal language lives on individual laptops, not in a shared library

Straight answers

Won't the drafts sound generic?+

Not if the system is built the way we build it. The drafts are assembled from your own past winners — your scope language, your case descriptions, your phrasing — not from internet boilerplate. If your proposal library is thin or scattered, the first weeks of the engagement include collecting and tagging it, because the library is what makes the output sound like you. A model writing from nothing produces generic proposals. A model assembling from a hundred of yours does not.

What stops it from quoting the wrong price?+

Pricing never comes from the model guessing. Your rate card, minimums, and discount rules are encoded as deterministic logic: the agent looks prices up and applies rules, the same way a spreadsheet would. When a deal falls outside the rules (an unusual scope, a discount past your threshold), the draft ships with that section flagged and routed to whoever owns pricing, not filled in with a plausible-looking number. The model writes the words. The rules set the prices.

Can I do this myself?+

A light version, yes. Our free proposal draft workflow shows you how to turn call notes plus a past proposal into a solid first draft with off-the-shelf AI tools, and the rest of the free skills library covers the surrounding pieces. Where DIY runs out of road is the infrastructure: retrieval across a real proposal library, pricing rules the model can't override, CRM integration, and flags that route to the right approver. Pasting into a chat window gets you a draft. It doesn't get you a system your whole team can trust with pricing.

What does it cost?+

Engagements start at $5,000/month, which covers building the workflow and running it: maintaining the library, tuning the pricing rules, and fixing things when your CRM or templates change. Run your own numbers before booking. Count the hours your team puts into proposals each month, price them at loaded cost, and weigh what faster turnaround would mean for the deals currently stalling in your pipeline. If that total sits well under the fee, you don't need us yet — start with the free resources and come back when volume grows.

Find out what your proposal turnaround actually costs

The free AI opportunity audit maps how proposals get built in your shop today, and tells you whether this is the workflow to automate first — or something else is.

We take on companies ready to invest $5,000+/month. Not there yet? Our free resources are genuinely free.