The proof gap: agentic spend commits faster than evidence matures

Ex-Google and Verizon media executive Evgeny Popov makes the sell-side case that an AI agent can stay inside its budget and still spend too much. His worked example: a brand hands an agent $200,000 and the agent moves $1,000 an hour into a new mix of CTV and retail media, but the first sales readout fit to judge that change takes 48 hours to mature. By then $48,000 is committed to a decision nobody has assessed, and the readout itself arrives as an estimate with error bars about a mix that is already two days stale. He calls it the proof gap: spend committed while the evidence needed to assess a change is still pending. It is the third in a series in which he argues agents need richer signals than segments, feedback fast enough to reprice the next impression, and standards that establish authority rather than parse data.

The crux

Agentic buying does not remove the measurement problem, it accelerates the rate at which unverified spend accumulates, so the gap between commitment and proof becomes a direct cost line for whoever underwrites the budget.

Sell-side read

Publishers selling into agentic buying need to shorten their own feedback loops, from campaign-read to same-day signals, because the agent's repricing decision happens before your report lands. Inventory that can feed fast, priced signals holds its value; inventory that only reports in arrears gets repriced by default.

Watch

whether any of the agent standards in flight adds a provable-latency requirement, which would make "fast evidence" a sellable attribute rather than an internal cost.