I’ve Been Building the Future of Go-to-Market for 19 Years — and It’s Finally Here. Here’s How I Knew What Was Coming.

When we founded ZoomInfo in 2007, the conviction was that data is the prerequisite for everything in go-to-market (GTM). Assembling the architecture took the next 19 years, because the technology customers needed to consume the data the way they wanted was not yet built.

What changed is the environment around the data, not the data itself. GTM has hardened into a discipline that companies treat as a source of competitive advantage rather than a function their sales team carries out by feel. The infrastructure that makes AI-native execution possible — APIs any service can call, MCPs any agent can read — is finally available everywhere our customers do work.

The architecture customers pulled forward

Soon after we founded the business, customers wanted our data in places other than ZoomInfo. The first ask was the CRM. Then it was marketing automation tools. Then someone asked whether we had an API, and I had to figure out what an API was.

Because customers kept pulling the data into more surfaces, we kept rebuilding to meet them there. Over the past two years, we did that rebuild deliberately, re-architecting the underlying infrastructure to make the data available everywhere before the use cases that would justify it were obvious. When large language models arrived, and MCP followed, the data was already designed to be consumed somewhere other than our front-end application.

That sequence inverts the usual founder narrative about AI. We did not move toward AI-native infrastructure because of ChatGPT; the customer signal had been consistent enough to bet on for years. The recent leg of the journey is the part where the market caught up.

Why the discipline hardened

If you searched 2020 earnings call transcripts for “go-to-market,” you found a few mentions across maybe five calls. Since we changed our ticker from ZI to GTM, that number is up roughly 50 times. The term is in every job description and on every earnings call.

When we started ZoomInfo, we treated go-to-market as a real discipline inside the organization, and that created differentiation because most companies did not. Today, companies broadly recognize go-to-market as a discipline that can give them a competitive advantage. The corollary is that anyone who has not built that discipline is at a structural disadvantage.

This shift is happening now because the data required to make go-to-market scientific is finally available in a way it was not in 2007 or 2020. The discipline came first. The data caught up.

The 5-year test

Go-to-market is made up of roughly 20 tasks — prospecting, account research, territory planning, TAM identification, deal inspection, call coaching and so on. Historically, ZoomInfo was excellent at one or two of those boxes. The test we set when we changed our ticker was whether we could expand into the rest.

By the end of this year, the target is 8 boxes; by the end of 2027, fifteen. The jobs to be done in go-to-market have not changed since 2007. What has changed is where those jobs get done. Work that used to happen inside opinionated SaaS interfaces is now happening in Claude Code, in Codex and in custom internal applications that mirror specific revenue workflows. The infrastructure has to be flexible enough to plug into all of those places.

The layer, not the destination

Becoming the layer rather than the destination requires the underlying architecture to change. APIs and MCPs have to be flexible for any platform, not hard-coded for ours. They have to be readable by a developer and, more importantly, readable by an agent that can scan the docs and build on top without unique knowledge of our software.

Strategically, that geometry expands the surface area of the product from one functional area to 20. Go-to-market data has applicability in account management, marketing, product and sales development — every function that touches a customer. The application layer becomes one entry point among many. The data asset becomes the differentiator.

The unglamorous prerequisite is identity resolution. The same person might be Thomas in one platform, Tom in another, and T. in a third, with three different titles across the same records. Companies are harder. Cisco acquires AppDynamics and several other businesses, and dozens of company names should roll up to one entity. Disney owns ESPN, but ESPN runs its own business with its own buying centers. That lineage has to be managed precisely, because every downstream agent depends on resolving to one accurate entity.

Running at regulation

Working with 70% of the Fortune 50 requires trust, and trust in this category is earned at the statute level. Sixteen states in the U.S. have their own privacy laws, and every country has its own — GDPR across Europe, LGPD in Brazil, PIPEDA in Canada. Enterprises need a partner paying attention across that patchwork.

Our posture is to run toward regulation rather than away from it. When a new statute passes, we read it, understand what it requires and build the governance to meet it. Some countries are consent-only; others are opt-out. The infrastructure has to ingest data differently across those regimes because that is what gives enterprise customers confidence to consume it legally and compliantly.

The practitioner’s discipline

I meet with roughly 110 customers per quarter, talk to employees daily and carpool to work most days with our Chief Product Officer. None of that fully substitutes for being a practitioner in our own tools. When data reports a trend forming in a segment, I ask for an anecdote — one specific customer, one call, one workflow that exemplifies what the data shows. The anecdote is what grounds the abstraction.

Being a practitioner across our platform delivers the same grounding. I code in Claude, prototype in Vercel and use our products the way customers use them. That practice surfaces things customer meetings and internal dashboards report at a level of abstraction one or two layers too high. It is what tells me whether the next layer of the platform is ready before the market starts asking for it.