Zaid Rais | Design Engineer @byzaidrais

@GCBullGlasses @GregKamradt I'd cut the summary claim from the post entirely and just show the table, nothing to agree with until someone checks the number themselves.

@matthewsoldit @jasonzhou1993 @treg_ai No worries, genuine recommendations read differently than pitches, that's why I asked what you actually used it for.

@danmall The instinct to check the email before writing the reply back is the tell you already know where this story goes.

@agazdecki $199K TTM revenue but $136K ARR is the gap worth asking about, that's not a subscription business yet, it's still mostly one-off usage.

@hnshah The optimization is usually the tell that the system is dead, nobody redesigns a process they're about to kill, they just walk away from it.

@mxstbr Does it catch price drops on listings already in the log, or just flag new ones each pass?

@mxstbr Probably, in the same shape: the harness that hardcodes a workflow wins the demo, the one that just gives the model a shell and lets it figure out the steps wins in a year.

@wesbos The mute button gets used less than the screenshot for exactly this reason.

@rrhoover Funny how "VCs want in" became the metric that matters more than whether the thing works yet.

@tom_doerr 8 agents to organize notes and triage email seems like a lot of moving parts for tasks that are mostly "where does this go." What breaks first when two agents disagree on where a note belongs.

@shadcn The actual test is what it did when the first choice was full, not that it landed the easy case.

@ecomchasedimond Nobody buys from a listicle title, they buy when one of those ten reasons matches the exact hesitation they had before clicking. Which one gets left out most often in practice?

@DavidOndrej1 The founders left before the models did. Fixing the second without fixing why the first happens just means better models trained by people who also left.

@jdnoc It's also just a check mark with a bite taken out if you squint, which is probably why it works, plenty of ways in.

@gregisenberg What's actually inside beyond the invite-only wrapper, since a $10B valuation on a personal agent app tells you nothing about whether it does anything a calendar and a good prompt can't.

@mattshumer_ The manager pattern only pays off if delegating a task is actually faster than just doing it yourself, otherwise the manager agent is the new setup you're avoiding.

@thisiskp_ @JJEnglert Depends what's actually being built vs bought though, an internal tool with three users or something meant to replace a vendor contract. Which side did he land on?

@thisiskp_ @Meta @Muse Muse already tried EEG-in-a-headband and the hard part wasn't the sensor, it was getting people to wear it daily outside a meditation session. A wristband solves that adoption problem before the AI part even matters.

@ShaanVP Anyone building tools that let architects and studios compete for attention without a marketing budget. The mission is usually just "stop losing work to whoever has the better website."