How Can Marketing Agencies and Consultants Use AI to Write Client Proposals Faster? 2 Real Setups (2026)

How do agencies draft proposals with AI after a discovery or sales call?

Both agencies follow the same basic loop. The call gets recorded. The AI reads the transcript along with material the owner already wrote. It drafts the proposal, and a person edits it before it goes out.

At GrowthSpree, co-founder Ishan Manchanda needed to send a proposal after every sales call. A properly researched plan took him four to five days, so for a long time he just emailed a description of the agency's services instead. Now Fireflies records his sales calls and its MCP server connects the transcript to Claude. He keeps a separate Claude skill for each offering: AI services, paid media and ABM. The matching skill runs competitor research and an audit, picks the relevant parts of the service and drafts a proposal for the prospect's industry. It uses the strategy he covered on the call plus any nuances he adds. He described the setup on the Closing Time podcast. Full case: GrowthSpree's proposal setup.

Adebukola Ajao runs B.D.Y. CONSULT, a boutique marketing agency, on her own and has for a decade. She handles client strategy, creative direction, marketing execution, partnerships and operations herself. A proposal for a new client used to take more than a week. Now Gemini goes through her discovery-call notes and recordings, work emails and Google Docs on a prospect and drafts a first proposal from templates she wrote. Read AI transcribes the calls and shows where the prospect was engaged or tuned out, so she can rework those sections before sending. Source: Business Insider. Full case: B.D.Y. CONSULT's proposal setup.

What tools do these agencies use for proposals?

Each agency uses two tools: one to capture the call and one to write the draft. For more agency setups beyond proposals, see AI automations for marketing agencies and consultants.

• Fireflies (GrowthSpree): records sales calls. Its MCP server connects the transcript to Claude.
• Claude (GrowthSpree): one skill per service line runs the research and writes the proposal draft.
• Read AI (B.D.Y. CONSULT): transcribes calls and shows where the prospect was engaged or tuned out.
• Google Gemini (B.D.Y. CONSULT): drafts the first proposal from her own templates, using notes, recordings, emails and Google Docs.

How much faster are proposals with AI?

Manchanda says the custom proposal is ready about an hour after the call. It used to cost him an extra day and sometimes four to five days. Ajao says a proposal that took more than a week now takes a day or less.

Ajao also points to the emotional side: "If a client says no, I don't feel burned out from having spent a week doing their proposal. It's just onto the next one."

Keep two limits in mind. Both figures are self-reported, and neither agency gives win-rate figures. And Manchanda is clear that AI made the agency faster, not better. Delivery quality still comes down to the people running each account.

What should a person still check before sending?

Both setups keep a person between the draft and the client. These are the checks from the two cases.

• Read the draft against the call transcript (GrowthSpree).
• Verify prices and any competitor claims (GrowthSpree).
• Check scope and price against what was actually said on the call (B.D.Y. CONSULT).
• Expand the sections Read AI shows the prospect reacted to, and cut the parts they skipped (B.D.Y. CONSULT).
• Send the proposal from your own email (both).

What about the same pattern outside marketing agencies?

The firms below are not marketing agencies, so they are not in the table above. They use the same call-to-proposal pattern. More examples are on our AI for proposals and estimates page.

Element Design Build, a design-build remodeler in Hudson, Ohio, uses Fireflies and Claude to turn recorded sales calls into proposals. Owner Vlatko Demrovski says initial proposals that took 60 to 90 minutes now take 10 to 15. "I read every single word," he says. Source: Qualified Remodeler. Case: Element Design Build.

Iconoclastic Capital, a financial planning firm in Rochester, New York, turns meeting transcripts into an interactive proposal web page using a v0 template. CEO Christopher Haigh says custom proposals took three to four hours even with templates. Now the template drafts the priorities and he only tweaks them. He gave no new time figure. Source: Kitces Financial Advisor Success podcast. Case: Iconoclastic Capital.

Evergreen Insurance in Washington is an insurance agency. Its leadership uses ChatGPT for Business for proposals among other jobs. No time figures were given. Source: IndependentAgent.com. Case: Evergreen Insurance.

What does a first DIY setup look like?

Pick the setup that matches the tools you already use. These steps come from the two agency cases. If you want a smaller first project, start with our quick win guide.

1. Claude and Fireflies route: connect Fireflies to Claude via Settings > Connectors, and have Fireflies join your sales calls.
2. Build one Claude skill per service (a SKILL.md file, then upload it).
3. After each call, ask Claude to use the matching skill to draft the proposal.
4. Gemini and Read AI route: write one proposal template per service in Google Docs, then create a Gemini Gem with your templates and price list.
5. Connect Read AI to your calendar. After each call, paste the summary, transcript, your notes and the email thread into the Gem.
6. Either route: in your prompt, ask the AI to mark anything not supported by the transcript or your notes as [CHECK] or [ASK].
7. Before sending, check the draft against the call, verify prices and competitor claims, and send it from your own email.