AI shopping agents are forecast to drive $100 billion in online sales by next year, according to a projection reported by MediaPost on 10/05/2026.

That is the single hardest number in the story, and it deserves precise handling: $100 billion is a forecast of purchases AI agents will influence or execute, dated to arrive within roughly twelve months of publication. It is not a measured figure from a completed period. No methodology, sample size, or baseline year appears in the source, so the number should be read as a projection whose assumptions — how an "agent-driven sale" is defined, and whether assisted discovery counts — remain undisclosed.

What does a $100B forecast actually change?

For revenue teams, the operative question is not whether the total is right but where those dollars shift. If autonomous agents increasingly complete product selection and checkout, the buyer that sellers and marketers must persuade stops being a human browsing a storefront and becomes a machine evaluating structured product data.

That distinction matters at specific deal stages:

• Top of funnel: agents compress discovery; content optimized for human attention may never reach the decision path.
• Evaluation: pricing, specifications, and availability need to be machine-readable, or the agent skips the offering.
• Checkout and retention: repeat purchasing logic could move wholesale to agent-side routines.

Measured versus asserted

What the source asserts: a $100 billion scale for agent-driven online sales by next year.

What the source does not show: current agent-driven sales volume, growth rate from a stated baseline, category breakdown, or the share of transactions agents complete autonomously versus merely recommend.

Analysts covering e-commerce forecasts will recognize the pattern. Large round-number projections for emerging channels routinely arrive without published methodology, and the burden of verification falls on the reader. Until the forecasting party releases its model, the $100 billion figure functions as a directional signal of vendor and analyst sentiment, not as an auditable market-size estimate.

Who should act, and when

The forecast's practical weight varies by team type. Large e-commerce and retail revenue operations teams face the nearest-term exposure: if even a fraction of the projected volume flows through agent-mediated checkout, product data quality becomes a revenue lever rather than a catalog housekeeping task. Smaller direct-sales teams selling through human-led pipelines face a longer horizon, though agent-assisted discovery could still reshape inbound demand before the headline number materializes.

A twelve-month window is short. Teams that wait for measured transaction data before auditing their machine-readiness may find the channel already allocating spend by the time clean numbers exist.

MediaPost's report signals that agent-mediated commerce has moved from experiment to forecast-scale economics. The next testable milestone will be whether platforms disclose actual agent-driven transaction volumes against this projection as the deadline approaches.