In June 2025, UserGems ran a survey with Wynter on the state of AI adoption across sales and marketing. The result was pretty shocking with only 7% of GTM teams reporting clear ROI from AI.

Ten months later, the models have leveled up and the agents are multiplying. So why is GTM lagging behind, and what does the stack look like when you actually fix it?

Why coding and law cracked AI before GTM did

Look at the categories where AI has obviously worked like coding, legal. They share a structural advantage: the data needed to make a good decision is self-contained.

For coding, you need your own codebase. For legal, you need the contract in front of you. Point an LLM at a bounded, high-quality dataset and the magic happens.

GTM is fundamentally harder. To make a good outbound or ABM decision, you need:

• First-party data (CRM, product usage, call recordings, past champions, closed-lost notes), dispersed across a dozen systems
• Third-party data and signals (job changes, intent, hiring, funding, buying committee changes), constantly changing and often cluttered

Stack AI on top of bad or fragmented data and you get automated bad decisions at scale. That's exactly what the first wave of AI SDRs delivered, spray-and-pray v2, dressed up as innovation.

Your stack has a system of records and a system of actions. The brain is missing.

Here's the picture of a typical GTM stack in 2026:

• System of records on the left, your CRM, where information is stored
• System of actions on the right, your sales engagement platform, dialer, marketing automation, ad platforms

What's in the middle? Each individual rep's brain, deciding on the fly which accounts to work, which signals matter, and what to say.

That's how you end up with inconsistent results, buried signals, and reps spending two-thirds of their day on research instead of conversations.

The missing layer is a GTM brain. Something that sits between records and actions, continuously processes first- and third-party signals, scores what matters, and tells your team who to target, what to say, and when to act.

Where the stack is headed: 2025 to 2027

The shift is easier to see on a timeline:

• 2025: Bolted on AI. Writing assistants glued onto existing tools. AI SDRs optimized for quantity, not quality. Humans still doing all the operational work.
• 2026: Intelligence-led. AI starts identifying signals, driving prioritization, and handling the lower-value tasks automatically. Humans reclaim time for the high-leverage work: phone calls, key messages, strategy.
• 2027: Autonomous GTM. Humans set the strategy and the guardrails. Agents run the motion, learn from outcomes, and iterate autonomously. The rep's job is to review and redirect, not to execute.

The direction of AI takes over the operational load so humans can focus on strategy and the conversations that actually close deals.

A field guide to the vendor landscape

Choosing a vendor in this category is overwhelming, everyone is promising the world. A more useful question than "what does it do?" is "what's its philosophy?" Because philosophy is what determines which functionality gets built and how it gets used.

Here's the overview of the current map:

• ZoomInfo: Data coverage and quantity. Full suite for SDRs. Freedom to build any workflow. Philosophy: more data = more options.
• 6sense / Demandbase: Built from intent-based advertising. Sales is an afterthought. Strong fit if advertising is your primary motion.
• Nooks, UniFi, Regie, Common Room: Breadth of signals. Quantity often outweighs quality. Heavy lift on the end user to decide what matters.
• Clay: Extremely versatile. Can do almost anything, if you have the nerves, time, and resources to build and maintain it.
• UserGems: Data quality as the foundation (it's where we came from with champion tracking), sales and marketing orchestrated together, and the brain managed centrally instead of by each individual rep.

None of these are wrong philosophies. They're different bets. Pick the one that matches how you actually want to run GTM.

What UserGems is building

UserGems positions itself as the AI command center for outbound and ABM, the brain in the middle.

So what does that “brain in the middle” actually look like in practice?

Two components do the work:

• Data Agents capture first- and third-party signals, run custom research, and keep your account and contact data clean
• Intelligence Agents score accounts and contacts, build prioritized lists, personalize outreach, and orchestrate sales and marketing execution

Critically, UserGems is not a system of action. We don't build sequences, dialers, or ad platforms. The brain sends its outputs into whatever you already use, Salesforce, Outreach, Salesloft, HubSpot, LinkedIn Ads, Meta, Google. You keep your stack. We make it smarter.

That's the architecture decision we think GTM teams need to make in 2026. Not "which tool replaces everything?" but "where does the intelligence layer live, and who owns it?"

Because whoever ends up filling that brain role, your reps, a vendor, or nothing at all, is the one deciding whether AI actually shows up in your pipeline, or stays stuck at 7%.