How Is Agentic AI Changing the GTM Stack and RevOps in 2027?
Agentic AI is changing the GTM stack by moving AI from suggesting to doing. The big sales and CRM vendors now ship named agents that prospect, qualify leads, book meetings, update opportunity fields and report on their own results inside the tools reps already use. For RevOps, the job shifts from administering fields and reports to governing the data, rules, permissions and measurement that agents act on.
The launches below are real and dated, but almost every performance number comes from the vendor selling the product. I could not read an independent analyst page that measures results, so this page does not quote analyst figures. Treat vendor numbers as claims, and test agents on your own data before you widen their scope.
What the vendors have shipped
Each item below comes from the vendor's own newsroom, documentation or release notes, with the date shown on the page.
- Salesforce. Agentforce reached general availability on October 29, 2024 with a Service Agent and Agent Builder (Salesforce). On March 16, 2026 Salesforce described a set of Agentforce Sales agents: prospecting, engagement, account research and meeting prep, pipeline management, quoting and partner success (Salesforce).
- HubSpot. Breeze Customer Agent and Prospecting Agent are sold to Pro and Enterprise customers, and HubSpot changed how both are priced effective April 14 (HubSpot).
- Gong. On April 16, 2025 Gong introduced over a dozen AI agents for revenue teams, including agents for deal monitoring and for deal and revenue prediction, included with Gong licenses at no extra cost (Gong).
- Outreach. Its release notes describe a Revenue Agent, a Deal Agent that suggests opportunity field updates after calls, and an AI Control Hub for switching agents on and off (Outreach, August 2026).
- Microsoft. Dynamics 365 Sales lists a Sales Qualification Agent, Sales Opportunity Agent, Sales Close Agent and Sales Research Agent, among others (Microsoft Learn).
- Clari and Salesloft. The two announced an agreement to merge on August 7, 2025, framed as the start of an "Autonomous Revenue System" in which humans and agents work together. The announcement gives no product specifics or results for agents (Salesloft newsroom).
Salesforce: from service agent to sales agents
The first Agentforce release was customer service: Service Agent pricing started at $2 per conversation, and Salesforce said customers had built over 10,000 agents at Dreamforce 2024 (Salesforce, October 2024). Sales agents followed. In the March 16, 2026 announcement the Engagement, Pipeline Management, Account Research and Meeting Prep, Quoting and Partner Success agents were described as generally available, with the Prospecting agent available March 30 (Salesforce).
The headline result is a Salesforce claim: its own president of sales said its agents contacted 130,000 leads and created 3,200 opportunities over four months, and Salesforce says sellers save up to 25 hours per week. Neither number has been independently checked.
Pricing is moving toward outcomes
HubSpot's April 2026 change is the clearest example. Customer Agent moved from $1.00 per conversation to $0.50 per resolved conversation, and Prospecting Agent moved from a monthly charge per enrolled contact to $1.00 per lead recommended for outreach, with a free 28-day trial (HubSpot). HubSpot also says Customer Agent resolves 65% of conversations; the page states no measurement period or method, so read that as a vendor claim.
Gong went the other way and bundled its agents into existing licenses (Gong). Either way, RevOps now has to model agent cost per outcome, not just seats. Our explainer on outcome-based pricing covers why vendors are doing this.
The stack is opening up and consolidating
Two patterns show up in the vendor pages. First, tools expose themselves to other agents: Microsoft says Dynamics 365 Sales supports the Model Context Protocol (MCP) so sales actions can surface through other agents and apps, and Outreach's August 2026 notes add an MCP server with 11 tools, limited to what the underlying user can do (Microsoft Learn; Outreach). Second, vendors are combining: Clari and Salesloft, whose merger was announced in August 2025 and completed on December 3, 2025 (Salesloft newsroom), argue for one revenue system where humans and agents work together, though it publishes no agent results (Salesloft newsroom).
The practical effect for a RevOps leader: the CRM is no longer the only place that writes to deal records. An agent in a sequencing tool, a call-intelligence tool and the CRM may all touch the same opportunity.
What RevOps teams actually change
The vendor pages show the controls agents need. The priorities below combine those sourced details with my own judgment, which I mark as such.
- Data hygiene. Agents read your fields, so unclear fields produce unclear actions. Outreach added AI descriptions for custom fields so its AI interprets them more accurately (Outreach). Writing those descriptions is now RevOps work.
- Routing and qualification. Qualification agents such as Microsoft's Sales Qualification Agent research a lead and decide whether to engage it, which means your ideal customer rules become the agent's rules (Microsoft Learn). My judgment: document routing and handoff criteria before an agent applies them at scale.
- Forecasting. Gong lists deal and revenue prediction agents, and Outreach's Deal Agent suggests opportunity updates that can be accepted or auto-accepted after calls (Gong; Outreach). My judgment: decide which agent-written fields feed the forecast, and who signs off.
- Measurement and comp. Outreach's May 2026 notes added filters that show meetings booked, opportunities created, pipeline influenced and revenue generated for Revenue Agent activity (Outreach). No vendor page I read says how to credit agent-sourced pipeline in commission plans; that is a decision for your sales and finance leaders.
- Governance. Outreach's AI Control Hub lets admins switch individual agents on or off and a change log shows old and new agent settings; Microsoft says agents are tested in sandbox or test environments before production (Outreach; Microsoft Learn). Use those controls, and keep an owner for each agent.
What is still unproven
Most performance claims here are self-reported: Gong says AI Briefer users saw a 19% higher overall win rate, HubSpot cites a 65% resolution rate, and Salesforce cites 130,000 leads contacted (Gong; HubSpot; Salesforce). Gartner and McKinsey both publish research on agents, but I could not load their pages for this article, so I am not repeating figures I have not read. Check their current publications yourself before citing them.
Questions people ask
What is agentic AI in a GTM stack?
In these vendors' own words, it means agents that act on tasks, such as researching leads, sending outreach, booking meetings or updating deal fields, instead of only suggesting. Examples: Salesforce's Agentforce Sales agents and Microsoft's Dynamics 365 Sales agents.
Do AI agents replace sales reps?
The pages I read do not claim that. They describe agents that work alongside sellers, for example Salesforce describing agents that prospect, engage leads and prepare meetings (Salesforce). Independent evidence on rep headcount was not part of what I could verify.
How are AI sales agents priced?
It varies. HubSpot moved to per-outcome pricing in April 2026, Salesforce started its Service Agent at $2 per conversation in October 2024, and Gong includes its agents in existing licenses (Gong).
What should RevOps do first?
My view, not a sourced rule: pick one narrow use case, fix the fields and routing rules it depends on, set a human owner, and measure it against a baseline before expanding.
Sources
- Salesforce: Agentforce general availability (Oct 29, 2024) · Launch date, Service Agent pricing, vendor claims
- Salesforce: Agentforce Sales (Mar 16, 2026) · Sales agent lineup, availability, vendor results
- HubSpot: Customer Agent and Prospecting Agent pricing (updated Apr 13, 2026) · Outcome-based pricing and vendor stats
- Gong: Gong Redefines AI Agents (Apr 16, 2025) · Agent lineup and vendor claims
- Outreach release notes, August 2026 · Deal Agent, AI Control Hub, MCP server
- Outreach release notes, May 2026 · Revenue Agent reporting
- Microsoft Learn: AI agents in Dynamics 365 Sales · Agent list, MCP support, test environments
- Salesloft newsroom: Clari and Salesloft merger agreement (Aug 7, 2025) · Merger announcement and vendor claims
- Salesloft newsroom: Clari and Salesloft complete merger (Dec 3, 2025) · Merger completed; Steve Cox named CEO of the combined company
Agentic AI in the GTM stack is real in shipped products, but proof of results is still mostly vendor-reported. The safest RevOps move is to treat each agent like a new team member: clear data, clear rules, a named owner and a measured trial.
Top photo: Alan Cleaver from Whitehaven, United Kingdom / CC BY 2.0 via Wikimedia Commons