Vikram Taneja, who leads AT&T Ventures, makes the case that the old playbook for seed-stage defensibility has broken. In a market where a capable model can replicate a clever feature in a matter of weeks -- and where incumbents like Adobe and Microsoft are folding the same capabilities into products people already use -- a slick demo is no longer a moat.
His reframing points to durable advantages that AI can't trivially copy: proprietary data that compounds with usage, distribution into channels a newcomer can't easily reach, and workflow lock-in that makes switching painful. Those are structural rather than technical, and they're what early investors increasingly underwrite when the product edge has a half-life measured in months.
“Those are structural rather than technical, and they're what early investors increasingly underwrite when the product edge has a half-life measured in months.”
The argument matters because it filters down to which companies get funded. If defensibility now lives in data, distribution and lock-in, then the founders who can articulate a credible path to one of those -- not just a better model wrapper -- are the ones who clear the bar. For emerging managers, it's a useful lens on a market where 'we built it first' stopped being an answer.