The New Enterprise Threshold
A high-growth AI agent vendor just drew a line: customers with fewer than 5,000 employees no longer get dedicated Forward Deployed Engineer support. Self-serve documentation only.
The reason is scaling capacity. When you are growing that fast, FDE hiring and training cannot keep pace with customer acquisition. So you triage: put limited humans on the biggest accounts.
The problem: the customers who most need AI agents to work are not always enterprises. The 200-person company running a three-person sales team needs agents as a structural shift, not a productivity bump. And they just lost deployment support.
Why This Actually Matters
The single biggest variable in whether an AI agent works is not the model or the prompt. It is whether you get real human deployment help from the vendor.
Zendesk CEO shared the data: enterprise customers with proper FDE deployment hit 60-80% automation rates. Self-serve customers land around 20%.
That is not a small gap. That is the difference between an agent that changes operations and one you turn off after 90 days.
Salesforce hit $540M ARR with Agentforce, but only 8% of their customer base adopted. What got results was not the initial discount. It was deployment: FDEs getting in the system, configuring against actual data, making it work.
The Two-Tier Reality
Enterprises get deployment help that makes agents work. Everyone else gets a knowledge base and a Loom video.
The math is honest: you cannot staff a real FDE engagement against a $10K account. But mid-market companies are exactly the ones who need agents to work, not as a nice-to-have but as a way to scale without proportional headcount growth.
Palantir solved this by separating "prove value" from "full deployment." Their AIP Bootcamp compresses multi-month pilots into five days with customer data, carrying a 75% conversion rate. The bootcamp handles initial value proof, then FDEs deploy once customers commit.
For mid-market sales teams evaluating AI agents: ask about deployment support before you sign. Self-serve onboarding might get you started, but getting to production takes 30 to 60 days of real configuration. The vendors restricting FDE support are making a bet that enterprises matter more than volume. Your data integration complexity does not care about your employee count.