Case Study · BluePrintNJ AI

A sales call agent that preps,
recaps, and drafts the proposal.

We built an AI agent in Claude that handles the work around every sales call. It researches the prospect, writes the briefing, processes the recording, and drafts the recap email and the proposal. People run the call and approve what goes out.

Published October 2, 2026By Parul KhoslaBuilt on Claude and Claude Cowork
5 minprep per sales call, down from about 3 hours
Same dayproposal turnaround, down from 2 to 3 days
7 stepscovered end to end, 2 of them done by a person
0drafts reach a prospect without a person reading them first

At a glance

The hours around each sales call went back to selling

The agent runs on Claude and Claude Cowork. It is used by Sanjay Khosla, CEO, and Akash Minocha, Revenue Growth Manager.

Before the agentWith the agent
Prep per sales callAbout 3 hours5 minutes
Recap emailWritten by hand after the callDrafted from the recording
Proposal turnaround2 to 3 daysSame day

The challenge

Every sales call came with hours of work on both sides of it

We are a small team, and the owner still runs sales meetings himself.

Before a call, someone had to research the prospect. That meant reading their website, looking up the people in the meeting, understanding their industry, and working out what to ask. It took about 3 hours per call.

After the call, the work started again. Notes had to become a recap email. The recap had to become a proposal. Writing a proposal took 2 to 3 days, while the prospect waited.

That time came straight out of the hours Sanjay and Akash could spend talking to the next prospect.

What the agent does

The agent covers every step around the call

People run the call and approve what goes out. Steps 3 and 7 belong to a person.

  1. Research the prospectBefore the call, the agent studies the prospect's website, the people in the meeting, and their industry.
  2. Write the briefingIt returns a detailed report with talking points and a suggested flow for the meeting.
  3. Run the call · personSanjay or Akash leads the meeting, prepared. The call is recorded.
  4. Process the recordingThe agent pulls out the prospect's needs, concerns, and agreed next steps.
  5. Draft the recap emailA follow-up email is drafted from what was actually said.
  6. Draft the proposalThe proposal is drafted from the call, matched to what the prospect asked for.
  7. Approve and send · personA person reviews the recap and the proposal before they go out.

Where people stay in the loop

The agent drafts. People decide.

Nothing reaches a prospect without a person reading it first.

  • The briefing is a starting point. Sanjay or Akash reads the report before the call and adds what only they know, like a past conversation or a referral.
  • The recap is checked against the call. The draft is reviewed for accuracy before it is sent, so the prospect never gets a summary of something they did not say.
  • The proposal is approved by a person. We confirm scope and pricing before the proposal goes out.

This is the same rule we build into every client project. AI handles the volume. People own the judgment.

Results

Prepared for every call, proposals while it is still fresh

  • Prep drops from about 3 hours to 5 minutes. Sanjay and Akash walk into every call with a full briefing and talking points.
  • Proposals go out the same day, not in 2 to 3 days. The prospect gets a proposal while the conversation is still fresh.
  • No notes lost. The recap comes from the recording, so it reflects what was actually said.
  • More time for the next prospect. The hours saved on prep and writing go back into selling.

What this means for your business

Every business has work that happens around the real work

For us, it was the research and writing around every sales call. For an accounting firm, it might be client intake. For a law firm, engagement letters. For a manufacturer, quotes.

The best first automation is not the flashiest one. It is a job your team:

01Repeats every weekFrequent enough that the time saved adds up fast.
02Does the same way each timeSame steps, same inputs, so an agent can follow them.
03Would rather not doHours someone would rather spend with clients or prospects.

We build agents like this for clients through our Automation and Agent Builds service. We start with your real workflows, rank them by impact against effort, and build the one that pays back first. Governance comes built in, with a person reviewing anything that reaches a client.

← All case studies

Take the first safe step

See where your business
stands with AI.

Take the AI readiness quiz to see where you stand, or book a free Lunch and Learn for your whole team.