My talk last week at ACS Fall 2025 in Washington, DC was 25 years in the making.
When I was a PhD chemistry student presenting research posters, I dreamed of speaking here someday. Back then, I imagined I’d be talking about carbon-carbon bond catalysis. Instead, I found myself speaking about a very different kind of bond, the one between me and my AI assistant, Atlas.
It works best when the baton is passed back and forth between human and artificial intelligences.
And nowhere did that become clearer than on the exhibition floor, in conversations with marketers and scientists all navigating the same tension: between what we can do with AI… and what’s actually working.
Ahead of my talk, I walked the exhibition floor with a simple goal: to learn, not sell. No QR codes. No branded stress balls. Just real conversations and a survey in hand.
Jasmine Gruia-Gray and I ran a simple on-the-floor survey with exhibitors and asked about sales and marketing, and how they are using AI.
63% of companies aren’t piloting AI at all, which means they’re either deeply cautious, heroically understaffed, or not yet seeing how to unlock the value it can unlock..
Less surprising: the most common AI use cases were content creation, marketing automation, and basic personalisation, solid, safe plays that reflect where most teams are in the adoption curve.
The hall was busy, but the intent behind the busyness was clear. Exhibitors were there to meet new customers, deepen relationships, and give air cover to launches. Brand building mattered, but as a supporting act, not the headliner.
When we asked what actually moves people to act post-show, the answers were refreshingly clear:
Live demos. Tight case studies. A short path to a real person. These beat a generic nurture sequence every single time.
Cutting through the exhibition floor noise is hard, capturing the right signals at the booth and getting them to the right person fast. Near-term priorities reflected that reality: enablement for the field and launches that actually land.
Success was judged by:
When asked how marketing could improve, the themes were:
Most teams are still figuring things out. A few are experimenting in pockets. Only a handful are running anything at scale.
But when AI did help, it wasn’t the dramatic, future-of-work kind of help. It was the helpful-intern kind of help. Quietly competent, efficient, and fast, as long as you gave it a good brief.
You’ll get more out of your post-show follow-up if everyone agrees on one simple thing: what qualifies as a follow-up.
Keep it light. Keep it consistent. Keep moving.
Speed without story rarely persuades.
In complex buying cycles, nuance and evidence win. If your inputs are average, AI will just help you produce average faster. The sharper play? Invest in one powerful proof point per segment, a clear result, a real quote, a short demo that shows rather than tells.
Then? Let AI help you scale that.
Create great inputs. Then automate.
That’s the real chemistry.