More Conversations, Same Team: Building AI Email Programs That Actually Get Replies
How Gilson Uses an AI Email Agent to Start Real Conversations With Scientists
Sales teams don’t want another tool, and buyers don’t want another marketing email. Counterintuitively, an AI email agent can serve both of those desires while producing better outcomes for both groups.
Most email programs can report opens and clicks but have little to say about what the buyer is thinking. Stephan Cherek, Account Based Marketing Manager at Gilson, a family-owned maker of laboratory equipment since 1957, calls that gap the black hole. In this Blueprint webinar, he shows how his team turned automated email into a consistent two-way conversation.
Scientists now write back with paragraphs about their workflows. Sales reps receive leads dripping with context about pain points and needs.
Ben Funkhouser, Senior Product Marketing Manager at 6sense, hosts the session and the Q&A. Press play to see what Gilson changed after its first attempt fell flat, and what you can borrow for your own program.
The biggest lessons from Gilson
Stephan’s team reported a 91% reach rate across their target list, a 54% open rate compared with about 26% for standard CRM emails, and about $250K in new direct pipeline across instruments and consumables.
Three insights sit behind those numbers:
- AI email works when it stops sounding like marketing email. Gilson’s first attempt pointed readers to a web page and felt transactional. Replies climbed once the emails asked scientists about their own work.
- Replies bring the voice of the customer to marketing. Scientists wrote paragraphs about their workflows, which gave the team message insight that clicks and opens never did.
- Sales adoption comes from framing and handoffs. Positioning the agent as an assistant, and passing leads to reps with context, turned a skeptical team into advocates.
Highlights (and timestamps)
1. Position the agent as an assistant, not a replacement
Gilson’s sales team asked the obvious question: “Is this going to replace me?” Stephan’s answer shaped the whole program.
“That’s why we went down the sales assistant route. We’re not replacing you. We’re giving you an assistant to help give you more leads so that you can close more and filter through all the noise to get to the people who are actually ready to have a conversation with you.” Watch at 15:12
2. Hand off with context, then close the loop
Reps get more than a name and a reply. Stephan’s team shares what the buyer was researching before they responded, then checks every month on whether each lead turned into an opportunity or a quote.
“Something that’s helped our sales reps is putting context around the handoffs. We can give them information about where the buyer was in the cycle, whether they were looking up keywords or topics or searching on our website. That can help drive the conversation.” Watch at 18:00
3. Start with one agent
Gilson began with a single agent in the US, added a second for more coverage in high-priority areas, then a third for the UK and Benelux (a political and economic union of Belgium, the Netherlands, and Luxembourg).
“The one recommendation I would always make when starting AI email is to start small and build trust. Until you have your workflow solidified and the handoff issues worked out, start with one agent and figure out which processes can be automated and which stay manual.” Watch at 19:39
4. Don’t let AI email sound like marketing email
Gilson’s first attempt was a friendly email with a link to the website. It read as transactional, and the results showed it.
“The biggest lesson is that conversational email should never sound like a marketing email. Let your CRM and automation platforms handle those. Let conversational email be the personalization and the specifics, where you can ask questions you couldn’t ask in a marketing email.” Watch at 20:52
5. Ask customers about their work and leave your product out
The fix was to change the subject of the email from Gilson to the scientist.
“We talked to the scientist about their own work, because scientists love talking about themselves, just like any normal human does. They love talking about their research and their workflows. So in our first couple of emails, let’s not talk about the product we want them to use. Let’s not talk about us. Let’s talk about them.” Watch at 22:36
6. Replies deliver the voice of the customer
Marketing rarely talks to buyers directly. Replies changed that. Stephan uses AI tools to group the responses into a handful of themes, then adjusts messaging every month or two.
“We were getting responses back on our first or second emails, paragraphs long, where they would outline their entire workflow. This was great information for marketing, because we don’t typically talk to the customer. Now we’re getting customer voice.” Watch at 23:04
7. An off-topic reply can still become pipeline
This is part of the pipeline increase Stephan reported.
“The deal actually came from a Pipette Max email. It was talking about automation and workflows, and what came out of it was a six-figure consumable deal... Just because we’re not hitting the customer with the specific thing they’re looking for, they’re still responsive to the emails because we’ve kept them conversational.” Watch at 28:54
8. Treat every reply as valuable data
Positive or critical, each reply is a chance to learn something or to fix a problem before it grows. Stephan also recommends pairing ROI with customer voice when reporting to leadership.
“Response rates are what’s most important, and replies are your most valuable data. Any type of reply, we treat it like gold, whether it’s good or bad. Sometimes it’s feedback we love. Sometimes it’s critical, and that gives someone on the sales or service team an opportunity to reach out and ask, ‘How can we fix this problem for you?’” Watch at 35:24
9. One rep’s results won over the team
Gilson started with one inside sales super user, built an agent for her territory, and coached her to treat it as an assistant.
“She was seeing an increase in her close rate. She was seeing better performance in her numbers, and she became a very strong advocate for this product internally.” Watch at 40:55
10. Build a four-email sequence that gets more direct as it goes
After the sequence, Gilson typically suppresses contacts for about three months so the agent doesn’t crowd the rest of their omnichannel outreach.
“We use a four-email series. In the first one, we don’t really talk about the product. We do a brief intro and put it on the customer to talk about themselves if they’re willing to. In the second, we talk briefly about the product. The third has a more qualifying call to action, like requesting a demo. On the last one, the subject line says something like ‘final email’ or ‘last call.’” Watch at 46:46


