Email to Quote in 30 Seconds — and Zero, When You're Ready
Most quote requests start with an email. "Hey, can you give me a price on..." followed by a description that's either perfectly clear or barely decipherable. Your current process: read it, interpret it, open a template, fill in line items, format, attach, send. Thirty to forty-five minutes for one quote.
There's a faster version. In watch mode, an agent reads the email, drafts the quote from your pricing history, and shows you the result. You review. You hit send. About 30 seconds.
But 30 seconds is the default — the mode you start in. As your trust grows, the same workflow gets faster:
- Watch mode (default): ~30 seconds. The agent drafts; you review every send.
- Trust mode: ~zero seconds of your time. The agent drafts and sends; you see what went out in tomorrow's morning brief.
- Auto mode: the agent runs the whole funnel — draft, send, follow up, close — and surfaces a Monday brief: "Two quotes signed overnight. One stale. Three new inbounds drafted."
Why email is the right starting point
Email is where quote requests live. Your customers don't log into a portal to submit structured job requests. They email you the way they email everyone — informally, with varying levels of detail, sometimes with attachments.
The traditional response is to manually translate that email into a structured quote. The translation work — reading, interpreting, structuring, pricing — is where most of the time goes.
The vertical-agent response is to let the agent do the translation: read the email, identify the work, map it to your shop's pricing history, draft a structured quote. You decide whether to review it (watch), spot-check tomorrow (trust), or read about it Monday (auto). The translation work doesn't disappear; it just isn't yours anymore.
How the agent drafts from email
The loop has four parts, and it's the same loop in every mode — the only thing that changes is where you sit on it.
1. Email capture. The quote request hits your Gmail. The agent picks it up automatically — or you click one button in the Setell Chrome extension to push it through. Same agent, same canonical state, no matter which way you trigger it. 2. Parsing. The agent reads the email and extracts what's being requested — work type, quantities, specs, deadlines, anything else the customer left for you. Attached drawings or specs get pulled in too. 3. Drafting from your pricing history. This is the vertical-brain part. The agent uses your actual QuickBooks invoice history, your past quotes, and the per-vertical defaults for anything it hasn't seen, and builds a structured quote with line items, quantities, and prices. Not a guess from a generic model. A draft from what your shop has actually charged. 4. Review and send — or hand it off. In watch mode, you see the draft, check the line items, hit send. In trust mode, the agent sends drafts for routine work automatically and shows you what went out. In auto mode, the agent owns the funnel — including the follow-up — and you read the recap when you want to.The customer receives a professional PDF, brand-matched, on your letterhead, in their inbox. They don't see watch or trust or auto. They see a quote that came back faster and cleaner than the three other shops they emailed.
A worked example
Real RFQ shape — the kind a machinist would get:
"Need a price on 10,000 mounting brackets, A36 carbon steel, 1/8" thick, laser-cut with a 0.005" kerf tolerance, black-oxide finish, 4-week delivery. Drawing attached."
In watch mode, here's the wall-clock breakdown:
- Customer email arrives: 0 seconds
- Agent picks it up from Gmail or you click "Send to Setell": 0–5 seconds
- Draft appears — material at your current A36 vendor rate, setup amortized at the 5,000-piece break your shop uses, run-time at your laser's actual cycle, black-oxide finish line at your past charge, freight estimate, your margin: ~15 seconds
- You review the draft: 15 seconds for a familiar spec
- You hit send: 1 click
In trust mode, you've graduated this kind of RFQ (repeat customer, familiar material, quantity break the agent has seen before). The agent drafts and sends. You see "Quote sent to ACME Manufacturing — A36 brackets, 10k qty, $X" in tomorrow's brief. If something looked off you spot-check it. If nothing looks off, you don't think about it.
In auto mode, you've graduated stale-quote follow-ups too. ACME doesn't respond in three days. The agent nudges. Doesn't respond in seven. The agent sends a polite "still interested?" Doesn't respond in two weeks. The agent closes the quote out and surfaces it as "stale, no response" in your Monday brief, freeing your pipeline view of dead weight.
The work didn't change. Your relationship to the work changed.
Three peer surfaces, one canonical state
The 30-second draft happens whether you're in the app, in your Gmail tab, or driving the agent from your stack — Claude Code, Claude.ai, ChatGPT, or your bookkeeper's MCP-aware tools.
- The app at app.setell.ai: pipeline view, quote review, customer history, Boxx (the in-app agent surface).
- Your Gmail: the Setell Chrome extension surfaces drafts and lets you push outbound through your inbox with brand-matched PDFs.
- Your agent stack: when Setell-MCP is live (waitlist open), you'll call the same agent from your existing stack — "draft three quotes from yesterday's inbox" from Claude Code, or have your bookkeeper run the loop from their workspace.
What makes the draft accurate
The accuracy of the draft depends on the quality of the data the agent draws from. A vertical agent does two things a generic AI tool can't:
Reads your QuickBooks history deeply — every past invoice line, customer-matched, item-cataloged. Not a separate price list you maintain in parallel. Your actual billed work, becoming the basis for the next draft. Knows your vertical's pricing patterns. The defaults are vertical-aware (machine shop ≠ landscaping ≠ HVAC), and they tighten every time you confirm or correct a draft. The first quote might need adjustment. The hundredth will need almost none for standard work, because the agent's pricing memory is now your shop's pricing memory.What changes when the draft takes 30 seconds (or zero)
A few things move when the speed shifts:
You quote more jobs. Work that wasn't worth 30 minutes is worth 30 seconds. You stop filtering out small or uncertain opportunities — they all get quoted. You respond first. Customers email three or four shops. The first professional response sets the anchor. Everything else gets compared to you. Pending quotes disappear. No more drafts sitting in your folder for three days because Tuesday got busy. Either it went out (watch mode), or the agent already handled it (trust mode), or you'll see it in the brief (auto mode). Revision loops stop hurting. Customer wants 20,000 instead of 10,000. You tell the agent (or it picks it up from the reply). The revision engine applies a structured patch — quantity-break recalc, run-time scaled, setup amortization adjusted — and produces a versioned quote. Not a new file. Not a rebuild.Getting started
Connect Gmail and QuickBooks — about 5 minutes, both. The free tier gives you 10 quotes to test the watch-mode loop end-to-end; no credit card. Send a handful, see how the drafts compare to what you'd have built by hand, and start graduating individual action classes to trust mode when you're ready.If you'd rather drive the agent from your stack — Claude Code, Claude.ai, ChatGPT, your bookkeeper's tools — join the Setell-MCP waitlist. Same agent, same canonical state, called from wherever you already work.
30 seconds is the entry point. Zero is the destination — at the pace you decide.
Ready to quote faster?
Start your 14-day free trial. See your first AI-drafted quote in minutes.