We Used Claude AI Cowork to Prospect 12 Companies and Draft 48 Emails in One Session

I had a Friday afternoon with nothing on the calendar. We'd just finished the Orient Printing & Packaging deployment, had a case study with real numbers (49 use cases mapped, 11 projects deployed, 85% faster document generation), and I needed to get it in front of similar companies. Printing and packaging manufacturers, specifically.

Normally this would eat an entire week. A day researching companies, another day finding contacts, another day writing personalised emails, then manually scheduling follow-ups. I decided to try doing the whole thing inside one Claude AI Cowork session instead.

I honestly didn't think it would work end to end. It did.

Finding the right companies

I gave Coworka simple brief: find companies similar to Orient. Indian printing and packaging machinery manufacturers, 100–500+ employees, established, multiple departments, not yet AI-adopted. The kind of companies where the same use cases we deployed for Orient would immediately resonate.

Cowork searched the web, cross-referenced trade show exhibitor lists (Pamex, Printpack India, Labelexpo), pulled company data from corporate registries, and came back with 12 qualified prospects. Each one had a company profile, key decision-maker names, contact emails, and a rationale for why they fit.

Wait, did that actually work? I spot-checked three of them. The company details were accurate. The contact names matched LinkedIn. The rationale for each one made sense.

It also tiered them by conversion probability. Tier 1 were companies with nearly identical DNA to Orient, same products, same scale, same operational patterns. Tier 2 were larger companies in the broader packaging ecosystem. Tier 3 were adjacent industries with the same complexity profile.

Writing emails that don't sound like spam

This is where outreach usually falls apart. Generic emails get ignored. But writing truly personalised emails for 12 companies takes hours of research per prospect.

Cowork drafted personalised initial emails for all 12 prospects. Each one referenced something specific about the company (a recent trade show, their product range, their global footprint) and bridged it to the Orient case study. The hook wasn't “do you want AI?” It was “I did this for a company in your exact industry, here's what it looked like, want me to map your use cases?”

The first drafts were too salesy though. Words like “incredible pace” and “results were wild.” Have you noticed how AI defaults to that breathless marketing tone? I told Cowork to pull it back: understated, warm, let the numbers speak. It redrafted the entire batch with the corrected tone. Much better.

Building the full sequences

One email isn't a campaign. I had Cowork build a 4-touch sequence for each prospect:

• Day 1— Initial outreach with the Orient case study hook
• Day 3— Follow-up highlighting a specific deployed project (Offer Generator, BOM Generator, Troubleshooting Assistant) relevant to that prospect
• Day 7— New angle with a link to our blog article on AI deployment methodology
• Day 14— Graceful break-up email with the case study link, leaving the door open

That's 48 emails total. Each one different. Each one referencing something real about the prospect. I kept waiting for the quality to drop off as the volume went up. It didn't.

Straight into Gmail

This is the part that surprised me most. Cowork connected to Gmail and created all 48 emails as drafts, organised by prospect and sequence stage. I labelled them for visual clarity:Settle/1-Initial, Settle/2-Day 3, Settle/3-Day 7, Settle/4-Day 14.

Then Cowork built a send calendar. Tier 1 goes first on Monday, Tier 2 on Wednesday, Tier 3 on Friday, with follow-ups staggered across three weeks. It created scheduled reminders at 9 AM IST for each send date, so I get a notification, open Gmail, filter by label, and hit send.

Twelve prospects, 48 emails, a 3-week send calendar. One conversation. On a Friday afternoon.

Why this matters beyond outreach

This is exactly the kind of workflow I deploy for clients. Not a chatbot answering questions, but a structured system where AI does real operational work. The same approach that built this outreach campaign is what I use to build offer generators, RFQ systems, and troubleshooting assistants.

The pattern is the same every time: give the AI structured context (company brief, case study, prospect list), clearinstructions (tone, sequence structure, personalisation requirements), and the right tools (web search, Gmail integration). The output is production-quality work that would have taken days to produce manually.

That's what settling AI into a business actually looks like.