The go-to-market services category is being rebuilt around AI agents, and the deals closing this month show the destination: not software a revenue team buys, nor an agency it hires, but a single operating layer where human specialists and autonomous agents run marketing, sales, and customer success together. The June acquisition of agentic-AI firm Knownwell by B2B services company 2X, which values the combined business at more than $400 million, is the clearest signal yet that the outsourced GTM model and the agentic AI stack are collapsing into one offering.
From tools and teams to an operating system
For a decade the revenue stack and the revenue workforce were bought separately. Companies licensed CRM, sales engagement, and intelligence platforms on one track, and hired agencies, SDR shops, and RevOps contractors on another. The two rarely shared a system of record, and the seams between them, the handoffs, the duplicated data, the manual reconciliation, were where pipeline quietly leaked.
The agentic model erases that separation by making the workflow itself the product. 2X describes its combined business as a GTM operating system that unifies specialist talent, AI-activated workflows, and revenue operations across marketing, sales, and customer success. Knownwell contributes the intelligence layer: semantic analysis of conversations, emails, and project activity that turns scattered commercial signals into actions an agent can execute. David DeWolf, Knownwell’s founder, becomes CEO of the merged company; 2X founder Dom Colasante stays on the leadership team. The combined organization claims more than 1,200 specialists and over 200 enterprise clients.
Why the timing is not a coincidence
The capital and the adoption curve have arrived at the same moment. GTM-AI funding in 2026 is tracking past $2.7 billion, and Gartner projects that by the end of the year 40 percent of enterprise applications will ship with task-specific AI agents. Agents that can prospect, draft, route, and update records are no longer demos; they are line items in next year’s plan.
That changes the economics of who does the work. The case for a large outsourced team doing repetitive, rules-based tasks weakens the moment an agent can do those tasks at marginal cost. The durable value moves to the smaller group of people who design the workflows, set the guardrails, and intervene on the judgment calls. A services company that owns both the people and the agents can reprice that labor in a way a pure agency or a pure software vendor cannot.
The bellwether question
One $400 million deal does not make a category, but it rhymes with what is happening around it. Enterprise AI agent platforms are racing to become the execution layer beneath GTM work, and Microsoft is curating that race through programs that give agentic startups Azure architecture and co-sell access. The pattern is consistent: whoever owns the layer where agents actually act, with approvals and audit trails intact, owns the relationship. 2X is betting that layer is best sold bundled with the humans who run it.
What it means for the sales and RevOps leader
The first decision this forces is a make-versus-buy question that did not exist 18 months ago. The choice is no longer between building an internal RevOps team and hiring an agency. It is whether your revenue operating system, the place where data, workflows, and governed AI live, is something you own or something you rent from a services partner that runs it with its own agents.
Renting it is faster and removes the burden of integrating agents into legacy systems, where most of the difficulty actually sits. Industry practitioners note that roughly three-quarters of AI implementation problems trace to change management and data architecture, not the models themselves. A partner that absorbs that work is selling time. The cost is dependence: your pipeline logic, your customer signals, and your institutional knowledge increasingly live inside someone else’s operating system.
The evaluation, then, is less about feature lists and more about control. Ask where the system of record sits and whether you can leave with your data and your workflow definitions intact. Ask which decisions the agents are authorized to make without a human, and who is accountable when one acts wrongly against a named account. Ask whether governance, the approval rules and audit trails, is yours to configure or the vendor’s to enforce.
The skeptic’s case
Agentic architectures are still early, and the open questions are not trivial. Best practice treats an agent as software that has been granted authority to act, which means data access, approval rights, evaluation tests, and security limits should be defined before deployment, not after. The disciplined path starts with bounded workflows that have clear inputs, established rules, and measurable outputs, then widens scope as the agents earn trust. A combined human-plus-agent services model can deliver that discipline, or it can obscure it behind a managed-service contract that makes the agents harder, not easier, to inspect.
What to do now
Revenue leaders do not need to pick a GTM operating system this quarter. They do need to decide, deliberately, how much of their operating model they are willing to run on someone else’s agents, and to write that decision into procurement criteria before a vendor writes it for them. Start by mapping which GTM workflows are genuinely rules-based and agent-ready versus which depend on judgment that should stay in-house. The companies that consolidate fastest will be the ones that knew, going in, which parts of the revenue engine they were prepared to hand over.
Source: 2X via GlobeNewswire
Related: the revenue stack being rebuilt for AI agents | GTM services layer going human-agentic
