Paul Morrill

brand @ statsig + amplitude

[Paul Morrill @paulmerol]

@spenserskates Give out toys! Spider rings and vampire teeth have been trick-or-treaters’ favorites (over gmo reeses and snickers) for two years running at our house.

[Paul Morrill @paulmerol]

Be honest: if you use Claude Code, are you using it for tasks you could just as easily do in Cowork or Chat? Probably. Because of the little crab guy. The fun ASCII art. The different typeface. Oh and maybe your task needs a little javascript? Be so for real. It's branding.

[Paul Morrill @paulmerol]

This is the direct result of branding. There's no clear brand difference between Cowork and Chat. They look the same, they help you do the same thing, they're for the same people. Obvious move that will hopefully be a good lesson for other product teams.

[Claude @claudeai]

Claude Cowork and chat are merging into one Claude.

Ask a quick question or hand over a report, and Claude takes it from there, even after you close your laptop. If something's unclear, Claude asks—you keep the final say.

Rolling out to Pro and Max over the next few weeks.

[Paul Morrill @paulmerol]

I don't think the same will happen with Claude Code because the branding is more distinct and fitted to its target audience. The product capabilities don't matter, it's the branding.

[Paul Morrill @paulmerol]

@clairevo Sounds like she went to the Stringer Bell School of Management

[Paul Morrill @paulmerol]

@CarmenDecouto @Amplitude_HQ wait does it count as working if i come play tailgate games too

[Paul Morrill @paulmerol]

Say what you will about Yahoo! but it’s the one news source I know will never have a paywalled article and that’s pretty special

[Paul Morrill @paulmerol]

@chriswtam resigning won't keep the self-improving products from coming for you. the only reasonable action is to fully support the advent of self-improvement. sorry.

[Paul Morrill @paulmerol]

@jaykreps especially for interacting with an llm. if you try to talk to it mechanically, it doesn't work as well. you have to converse naturally to give it all the right language vector cues.

[Paul Morrill @paulmerol]

From the company that helped you solve "wtf are my users doing?", here's "wtf are my agents doing?"

congrats to the agent analytics team!

[Spenser Skates @spenserskates]

Today we're launching Agent Analytics

Every team shipping an AI agent has the same blind spot. Offline evals pass, you ship, and then you have no idea what's happening in production. AI fails silently. Users ask a question and get different answers. They all look 'engaged' in a classic dashboard. You don't know who got a great response and who got a terrible one.

Agent Analytics solves it:
- Every session scored out of the box on task completion, response quality, friction, safety, and negative feedback
- Topic clustering across thousands of conversations, so you know if a failure hits 1 user or 10,000
- Eval agents that watch for regressions, and if you want will file a Linear ticket or the pull request themselves
- Agent quality sits next to product data, so 'payment scheduling fails 31%' becomes 'which renewals did that cost us?'

The Economist got their agent to a 96.9% task success rate and cut weekly failures 84%.

Included on every plan. Free tier included. https://t.co/K5vqrlMLvd

[Paul Morrill @paulmerol]

Customers care if you shift the cost over to them, though, right? Is this predicated on Amplitude not charging more for different levels of inference? Or does Amplitude charge differently based on different models, and customers don’t care about that pricing difference, only the quality?

[Paul Morrill @paulmerol]

@willdjthrill that's super interesting what you note about the html flexibility leading to single-player style. wonder if figma will develop an html reader or something like that!

[Paul Morrill @paulmerol]

@chriswtam @Amplitude_HQ Claude should stand on business with referrals, you'd think they'd want to take credit if they're trying to kill search

[Paul Morrill @paulmerol]

@willdjthrill me: it's a simple web crawl should this really take—
claude:

[Paul Morrill @paulmerol]

man claude come on

[Paul Morrill @paulmerol]

Add to the file of “AI as a costly fix for poor UI”

[Andy Fang @andyfang]

Today we're opening up the DoorDash CLI in limited beta.

`dd-cli` lets you order DoorDash directly from your agent: search stores, find the best deals, check out, and more.

Early access for US/Canadian macOS developers by waitlist. Excited to see what folks build!

[Paul Morrill @paulmerol]

4. It’s a crime against the scientific method, ok?!

Read the blog for more!

What science theory topics do you want to talk about next?

[Paul Morrill @paulmerol]

Getting back to my science teacher roots on the Statsig blog!

It's easy to run an A/B test without a real hypothesis. Here's why you shouldn't.

Read more: https://t.co/jeDJcuISlS

[Paul Morrill @paulmerol]

3. Scientific learning builds one step at a time. Your hypothesis is the firm footing to start. Without it, you can’t support the predictive model you’re experimenting on.