97% of revenue teams are investing in AI. Only 7% are seeing measurable ROI. Episode 2 goes inside that 7%.

‍[San Francisco, CA] — UserGems, the AI command center for outbound and ABM, today released the second episode of Brains of GTM, its ongoing webinar series putting revenue leaders who are generating real pipeline from AI in front of an audience of practitioners. The episode, titled Inside the 7%: AI that Actually Drives Pipeline, featured Caden King, Acquisition ABM Manager at Snowflake, and Julien Sauvage, CMO at Cordial.

The full recording is available now on YouTube.

The 7% problem

Last year, UserGems surveyed 100+ revenue leaders for its State of AI Adoption report. The finding that shaped this episode: only 7% of teams are seeing measurable ROI from AI — while 97% plan to keep investing anyway. Brains of GTM Episode 2 tracked down two of those 7% and asked them to show their work.

Snowflake: agents that change how sales makes decisions.

Caden King leads acquisition ABM at Snowflake, where his team has built a system of AI agents that picks accounts, drafts sales emails, writes ad copy, and launches campaigns. His core thesis: most teams build AI to automate current work or scale campaigns. The real pipeline lift comes from agents that directly influence how sales prioritizes accounts.

Key plays from Caden's segment:

• Rep-specific narratives over dashboards. Snowflake's weekly Sales Insights Emails give every AE a different story based on what's happening in their patch, with next steps spelled out — removing ambiguity so reps actually act.
• Account selection that shifts with the sales problem. When the problem is "we're getting meetings but not converting," the model weights ICP fit and buying signals. When the problem is "we can't get meetings," it weights engagement. Same input style, different outputs.
• Target where you're under-covered and under-pipelined. The instinct is to fix wherever coverage is lowest. The actual move: find where sales has capacity and new data can create pipeline that didn't exist before.
• One persona gap, all-in. Pick one persona group where the gap is meaningful, then align enablement, messaging, and outreach around it. Trying to fix everything at once is a focus problem, not a budget problem.

Watch Caden's full session here.

Cordial: a layered AI content engine built like a French pastry.

Julien Sauvage has spent 10+ years leading marketing at Salesforce, Gong, Clari, and now Cordial. His approach to AI systems: they're intelligently layered, what he calls the croque-monsieur framework. AI does the research. Humans create the POV: judgment, taste, originality, the emotional hook. AI measures what worked. Skip a layer and the system breaks. At Clari, this approach drove a 40% increase in inbound pipeline.

Key plays from Julien's segment:

• Research has to be broad and continuous. Most teams stop at Gong calls —roughly 5% of the available signal. Julien's stack spans voice of customer, competitive intelligence, internal teams, and market analysts.
• Brand is measurable, if you account for lag and interaction effects. His Marketing Impact Analysis framework does correlation analysis that most teams skip: brand pays back in months, not days, and brand alone rarely drives pipeline, brand plus media share is what works.
• The POV sniff test. An agent that grades any positioning doc: Is there a villain? Is it emotional enough? Where does it go flat?

Watch Julien's full session here.

What both approaches share

Caden and Julien came in with different stacks, different industries, and different motions of agentic ABM at enterprise scale versus a layered content engine for a fast-moving CMO. They landed on the same conclusion: AI drives pipeline when it's built around how sales actually sells and how humans actually think.

As Caden put it in the Q&A: without agents handling the campaign layer, his team would never have had the bandwidth for the high-touch, physical-world work — direct mail, video, the outreach that actually breaks through.

About Brains of GTM

Brains of GTM is a webinar series from UserGems featuring revenue leaders who have built and operationalized signal-based, AI-powered GTM systems. The series was created in response to a gap UserGems identified in the market: an abundance of theoretical content about AI and GTM, and a shortage of practical evidence of what these systems actually look like in production.

Episode 1 featured Mark Kosoglow (CRO, Docebo) and Kevin Driscoll (Head of Global ABM, Datadog). The full series is available on the Brains of GTM YouTube playlist.

The next episode will be announced shortly.

‍About UserGems

UserGems is the AI Command Center for B2B revenue teams, the brain that sits between your CRM and execution tools. We unify signals, prioritize buyers using transparent AI scoring, and deploy AI agents (Gem-E) to automate outbound and ABM execution. Customers include Mimecast, Workday, Salesloft, Sendoso. Learn more at www.usergems.com.