Our practices for covering AI accelerator programs have shifted meaningfully, and naming the specific changes matters.
We stopped accepting general success claims at face value
Early coverage sometimes repeated a program's own claims of success without enough independent scrutiny. We now try to focus on specific, checkable outcomes tied to particular cohorts.
We got more comfortable acknowledging genuine variation
We're now more willing to report that outcomes vary significantly by program and cohort, rather than generalizing toward a single verdict about accelerator value.
We started distinguishing brand from demonstrated outcomes
We try harder now to separate a program's selectivity and brand reputation from its actual, demonstrable track record of producing successful companies.
Why this evolution matters
Founders evaluating accelerator programs deserve coverage that reflects genuine variation in outcomes rather than a simplified verdict, and we think being more disciplined about this distinction is worth stating explicitly.