Gartner's November 2025 prediction that AI agents will outnumber human sellers by 10 to 1 by 2028 has been cited widely, often without the second half of the finding: fewer than 40 percent of sellers will report that those AI agents improved their productivity. Those two data points together say something more specific than either says alone. AI will be everywhere in sales. Most of it will not do what buyers hoped. The more useful question is not whether AI replaces sales, but which parts of the sales function it actually displaces versus which parts it augments, and what the evidence base looks like for each.
What McKinsey's research says is automatable
McKinsey's September 2024 report "An Unconstrained Future: How Generative AI Could Reshape B2B Sales" projected that generative AI could eventually automate activities consuming up to 80 percent of a sales rep's current time. The activities McKinsey identified as automatable are specifically the research and administrative tasks that fill rep schedules: prospect research, contact data enrichment, first-draft outreach messages, call preparation, post-call CRM updates, follow-up email drafts, and pipeline status reporting.
The same report found that companies using AI to assist sales teams, rather than fully automate their selling motions, reported efficiency gains of 10 to 15 percent concentrated in these administrative categories. The documented gains are not in closing rates, deal sizes, or the quality of complex negotiations. They are in the time cost of the preparation and documentation work that surrounds selling. That distinction is important for interpreting what "AI replacing sales" actually means in production environments today.
| AI handles today | AI does not replace |
|---|---|
| Prospect research and contact enrichment | Stakeholder political dynamics |
| First-draft outreach emails | Complex multi-party negotiations |
| Call preparation summaries | Building credibility with skeptical buyers |
| Post-call CRM updates | Reading when to push vs. pull back |
| Follow-up email drafts | Managing procurement (legal, finance, security) |
| Pipeline status reporting | Judgment calls on deal prioritization |
| Routine objection-handling templates | Champion coaching inside the account |
The tasks AI is not replacing
The McKinsey framework draws a consistent line between AI handling tasks that do not require human judgment and AI replacing tasks that do. Prospect research does not require judgment about a specific account relationship. CRM updates do not require judgment about when to push for a close. Writing a first-draft outreach email does not require judgment about the political dynamics inside a buying committee. These tasks are time-consuming but judgment-light, which is exactly where current AI is strong.
The tasks that drive enterprise deal outcomes do require judgment: reading stakeholder dynamics to identify who actually controls a budget decision, managing procurement processes that involve legal, finance, and security reviews simultaneously, building credibility with a skeptical technical buyer who has been through three failed implementations, knowing when to apply pressure and when to pull back in a negotiation. The Ebsta x Pavilion 2025 GTM Benchmarks report, analyzing $48 billion in pipeline data, found that early involvement of economic decision-makers increased win rates by 55 percent and that stalled engagement after initial contact dropped win rates by 113 percent. The skill of getting the right people engaged at the right time is not decomposable into the pattern-matching tasks that current AI systems handle well.
Where AI SDRs fit in the displacement picture
The most concrete case for AI replacing a sales role is the AI SDR category, which markets itself as automating outbound prospecting at scale. The evidence on AI SDR performance is mixed enough to warrant careful review before treating this category as validated. Our analysis of AI SDR performance found that tools like 11x and AiSDR produce results that vary significantly based on how they are used: high-volume, low-ACV outbound to well-defined prospect lists can work, while complex sales motions with multi-stakeholder buying processes show weaker results because the AI cannot adapt to signals that require judgment.
Gartner's June 2025 analysis finding that only approximately 130 of the thousands of vendors marketing AI agent capabilities have real agentic AI functionality is directly relevant here. Many tools claiming to replace SDR functions are producing templated outreach at scale, not genuinely autonomous sales activity. Teams evaluating this category should distinguish between tools that automate outbound execution efficiently and tools that claim to replicate the judgment-intensive parts of prospecting.
The job function that is changing, not disappearing
The realistic near-term picture from the research is not AI replacing sales jobs but AI changing what the sales job involves. If AI handles the research, first drafts, CRM hygiene, and call preparation that currently consume 60 to 80 percent of a rep's time, the sales job becomes more concentrated on the activities that require human judgment: complex discovery, multi-stakeholder coordination, technical selling, and negotiation. That shift means the floor for effective selling moves up. Junior reps who currently add value mainly through volume of activity will find that AI compresses the advantage volume used to provide.
Forrester's 2026 B2B predictions report is consistent with this framing. The analysis that B2B companies face more than $10 billion in enterprise value risk from ungoverned AI use is partly about sales contexts: AI-generated content that contains inaccurate claims about product capabilities, competitor positioning, or customer results creates legal and brand risk at a scale that manual review cannot catch without deliberate governance design. The risk is not that AI cannot generate the content. It is that generating content without human review at scale will produce errors that a rep reviewing their own draft would catch.
What the 40 percent productivity finding actually means
Gartner's prediction that fewer than 40 percent of sellers will report AI agents improved their productivity by 2028 is sometimes read as evidence that AI in sales does not work. The more accurate interpretation is that most deployments will be miscalibrated. Teams will buy AI tools expecting them to solve the high-judgment parts of selling, find that the tools perform well on the administrative and content generation tasks, and judge the tools against the wrong benchmark. That gap between expectation and deployment is what drives the low satisfaction rate, not a categorical failure of AI in sales contexts.
The teams that will report productivity gains are the ones that deploy AI against the tasks it is actually good at: reducing the administrative overhead that consumes rep time without requiring human judgment, then ensuring reps use that recovered time for the selling activities that require judgment. That is a management and process design problem as much as a technology problem, which is why simply buying better AI tools does not resolve it.
The realistic timeline for more significant displacement
McKinsey's projection that AI could eventually automate activities consuming 80 percent of a rep's time uses "eventually" deliberately. The current production state of AI in sales is capturing efficiency on a subset of administrative tasks, not transforming the selling motion. The pathway to more significant displacement requires advances in AI's ability to handle sustained, contextually complex conversations, read organizational politics, and make judgment calls under uncertainty. None of those capabilities are on a short time horizon based on the current performance of deployed systems.
For sales leaders making near-term decisions, the relevant frame is not whether AI will replace sales jobs in five to ten years but how to deploy AI against the tasks it handles well today, while maintaining the human judgment infrastructure that complex B2B deals require. Teams that treat AI as a replacement for that infrastructure before it is ready will underperform teams that treat it as a tool for freeing up human judgment for where it matters most.
For a practical review of which AI tools are delivering measurable results in sales today, see our analysis of the best AI sales automation tools based on review data.