Whether to respond to a crisis should come down to business impact — and owning up to mistakes is often the right move.
Amjad recounts the Jason Lemkin / Replit Agent incident, explaining that he chose to own the mistake rather than fight it because the story had massive business impact, and the response earned positive attention. He contrasts this with situations where apologizing gives unwanted attention to a minor issue. ✦ AI generated
Amjad Masad · a16z Podcast · 2026-07-17 · original ↗
starts at this moment · 14:45
“What advice do you give for when companies are being dragged by someone on Twitter or in the press — when to respond versus when not to engage?”
I think it comes down to business impact. Will it actually have business impact on my business? [...] I could go out and fight and just say, 'Well, we did nothing wrong.' [...] But I think the right thing to do was to own up to it [...] And my response got a lot of positive interaction and got a lot of airplay because I started with, 'Hey, we messed up.' [...] sometimes you can give attention to things that you don't want to give attention to, especially once your profile grows.
verbatim transcript · starts at 14:45
14:45>> Yeah. What what advice do you give for when companies are being dragged by by by someone on Twitter or or in the in the press or when to respond versus when when when not to engage? >> Yeah, yeah, I I think that's that's like a art thing and and I think I it comes down to um like business impact. Like will it actually have business on a
15:08impact on my business? Um you know, for example, like uh last year in July we had this story with um the you know, founder of SaaStr, Jason Lemkin where uh Jason had just vibe coding, had never vibe coded or coded anything before, and uh as he's working with Replit Agent, Replit Agent sort of misunderstood him or something like that, and then deleted the database. At the time we didn't have production
15:37development separation, and so the production database now like we actually separate them by default. Like the the Replit Agent can't even write to production database. It can only read. Um and uh and we we have backups. We have snapshots. You can go back, you know, to a certain day or time or whatever. So, the the mistake that the Replit Agent did was one click away from being fixed.
16:05But from his perspective, he didn't know that, and Replit Agent started freaking out. Models at the time like used to get like really anxious, especially Claude. Um and he he tweeted the screenshot where Replit Agent was like going kind of nuts. Uh and you know, it started gaining traction, gaining traction, gaining traction. And I in my mind I knew like we already fixed this issue, but but like the story has become so
16:30massive, and you know, obviously it's going to have business impact, and it's going to be in the press. And so I could I could go out and fight and just say, "Well, we did nothing wrong. We have backups, you know." It could be I could say it's a user error or whatever. But I think the right thing to do was to own own up to it, the fact that we
16:50didn't have development production separation is a bad thing. And 2 days later we rolled out that fixed. And so my my response got a lot of positive interaction and and got a lot of airplay because I started with, "Hey, we messed up. And this is what what we could have done better." But sometimes you can give attention to things that you don't want to give attention to,
17:16especially once your profile grows. >> Especially if you apologize. >> Yeah, especially if they apologize. So, I I would say just kind of focus on the business impact. >> Yeah. The And you mentioned earlier >> By the way, Jason is a huge fan now. He runs his entire Saster business, multi-million dollar business, with him another employee, and like 10 agents based on based on revenue agents. >> That's awesome.
17:38>> Yeah. >> The you you mentioned earlier that you sort of developed a a desire for this. For CEOs who are, you know, so locked in, they're just not that interested in in doing this. Would you recommend that they cultivate the desire to do it, or would you recommend that someone else on the team can do it? Does it have to be the CEO, or how would you think about