Your reps are no longer the only ones using your enablement content. AI agents now draft follow-up emails, prepare call briefs, assemble deal rooms, and answer questions in the middle of a deal. When they do, they draw on the same competitive intelligence, pricing, messaging, and process guidance your reps use.
Bad GTM knowledge affects both users, but agents increase the reach of each error. An outdated claim that a rep might use in one or two deals can appear in hundreds of buyer interactions when an agent reuses it.
Revenue enablement teams now need to maintain one source of GTM knowledge that people and agents can both trust.
Many companies still rely on systems built around human search. Legacy enablement platforms store content in portals, where old files can remain available long after the information changes. Reps have to stop working to find an answer or ask for help. Horizontal AI platforms can search across a much wider set of content, but they often lack a reliable way to identify what is current, accurate, or approved. When an agent uses outdated or unapproved material, it can repeat the error across many workflows before anyone finds the source.
Gartner predicts that AI agents will outnumber human sellers ten to one by 2028, while fewer than 40% of sellers will say agents improved their productivity. It also predicts that more than 40% of agentic AI projects will be canceled by the end of 2027. Across the two releases, Gartner points to data quality, inadequate risk controls, and unclear business value as barriers to useful deployment.
Many companies focus on model choice even though the quality of the underlying knowledge often determines whether the output is useful.
A platform that serves reps and agents must meet five requirements.
1. Treat GTM knowledge as governed infrastructure
Governed GTM knowledge needs an owner, a version, a permission level, an approval state, and a review cycle. Files stored in a content portal often lack this context, which makes it difficult to tell whether a person or agent should use them.
Once agents begin taking action, an old pricing deck can become the source for a quote, a follow-up email, or a buyer-facing page before anyone notices the mistake.
The system underneath your reps and agents should help teams create knowledge, approve it, control who can see it, track where it is used, and flag it when it goes stale. Whatever reaches a person or an agent should be current, accurate, and appropriate for that user.
During an evaluation, pick an outdated competitive claim and ask the vendor to show who owns it, when it was last approved, where it appears, and what happens when it changes.
2. Put AI to work maintaining the knowledge
Manual maintenance often falls behind changes to GTM knowledge. A pricing update, product release, new competitor, or messaging change can require dozens of edits across playbooks, battlecards, training, and buyer-facing content.
AI can reduce the manual work required to keep that knowledge accurate. The platform should help draft, update, classify, and curate content while preserving ownership and approval controls. A change at the source should flow to every place that uses it, including the agents grounded in that knowledge.
During a demo, ask the vendor to change a source, show every affected asset, and prove that the corrected information reaches both a rep and an agent. The exercise shows whether the platform can maintain knowledge after it generates the first draft.
3. Govern the brand at the point of creation
As reps use agents to create more buyer-facing content, brand controls have to cover every follow-up, deck, executive summary, and deal-room page.
Brand controls have to work inside the creation workflow. Approved messaging, templates, design rules, and reusable components should guide the output as it is made. Human review should remain part of high-stakes work, but routine assets also need controls that operate before the content reaches a buyer.
During an evaluation, have a rep and an agent create the same asset from approved knowledge. Both versions should look and sound like your company.
4. Connect knowledge to coaching and buyer execution
Knowledge, coaching, and buyer-facing execution should share context. Separate sources increase the risk that the rep hears one message during coaching, the agent drafts another, and the buyer reads a third in the deal room.
Vendors should prove that connection by tracing one approved message from its source to a coaching recommendation, a rep answer, an agent-generated follow-up, and a buyer-facing experience. Any step that relies on a copied file or a separate source creates another version that someone has to maintain.
5. Require MCP access, then test it live
Your knowledge system should connect to the AI tools and agents your company chooses. Model Context Protocol, or MCP, provides an open way to make governed knowledge available across that stack.
A content export or search endpoint gives an agent more material to sift through without showing which material it should trust. An MCP connection should preserve the permissions, ownership, freshness, and source context needed to choose the right information.
During an evaluation, connect an agent to the platform and ask it the same question you ask a rep. Then change a permission or correct a stale entry and repeat the test. The new answer should reflect the update for both users and continue to respect their access permissions.
How Spekit supports both users
Spekit launched GTM Knowledge Engine 2.0 this year to provide this foundation for agentic work. It organizes knowledge into modular, reusable units with ownership, permissions, governance, and source context. AI Content Builder helps teams create and maintain content from that approved knowledge, while Brand Studio applies company standards during creation. Spekit MCP makes the same governed knowledge available to ChatGPT, Claude, Copilot, Glean, Gemini, and custom agents.
Reps receive accurate answers and content inside the tools where they work, and agents draw from the same source. A change to the source reaches both users without requiring teams to update separate copies.
Clari and Salesloft and Showpad and Bigtincan have completed their combinations, while Seismic and Highspot have agreed to merge. These companies now face the work of joining platforms, roadmaps, and customer experiences, and buyers should expect that work to affect product priorities.
At your next renewal, ask your current provider to demonstrate these five capabilities and show which knowledge is influencing revenue, what has gone stale, and how quickly a correction reaches every rep and agent. The answers will show whether the current platform can support both users and whether a migration belongs on the roadmap.
