Salesforce’s State of Sales 2026 report, released this month and drawn from a survey of more than 4,000 sales professionals, ranks AI as the number one growth tactic for the year and pegs adoption at 87% of sales organizations, with 94% of leaders already running agents calling them critical to meeting business demands. The report reads less like an industry snapshot and more like a coordination signal to buyers who haven’t moved yet.

Underneath the headline number sits a widening structural gap. Forrester’s Q1 2026 B2B Sales Automation Landscape puts AI SDR production usage at 55% among companies with 500-plus employees, 27% at mid-market, and 14% at small businesses. That’s not a technology-diffusion curve. That’s a scissors chart.

The performance data explains why the gap matters. AI SDRs respond to inbound leads in under a minute, around the clock; the human SDR average is 47 hours. Multi-touch, multi-channel AI sequences convert at 2.3x the rate of single-channel outreach. Autobound’s 2026 prospecting analysis shows signal-personalized outreach hitting 15–25% reply rates against a 3–5% cold-email baseline. Response speed, personalization depth, and channel breadth are the three axes on which enterprise sales teams are now pulling away.

Salesforce EVP of Sales Adam Alfano identifies the technical prerequisite bluntly. “The secret sauce for sales AI agents is unified data,” he says, warning that stand-alone agents without full customer context produce garbage outputs. That’s precisely the setup cost SMBs have historically struggled to absorb.

The SMB tax is legible in the same report: reps at smaller companies spend 60% of their time on non-selling tasks (manual data entry, lead research, tool-switching) that enterprise stacks have already automated away. Salesforce’s SMB read argues AI agents let small teams “scale their operational capacity significantly without the overhead of additional headcount.”

The escape hatch is emerging, though. A cohort of no-code and horizontal-agent platforms, Glean, Dust, Gumloop, LemonLime, alongside general-purpose assistants like ChatGPT and Claude, is compressing the setup cost that once required a RevOps hire. Whether SMBs adopt fast enough to close a gap that Forrester already calls structural is the question the 2027 report will answer.

Sources