Sabitlenmiş Tweet
Shavin Peiries
8.7K posts

Shavin Peiries
@shavin47
10+ yrs building startups. Ex-PM @midfunnel. Writing about validation, design & growth. Check out my blog!
Sri Lanka → 🇱🇰 Katılım Mart 2012
563 Takip Edilen755 Takipçiler

@adityaag you cant even continue a conversation when you share a chat on claude
English

Shavin Peiries retweetledi

@quejicus91 It's the exact same problem as with humans.
Parallel work assumes perfect knowledge of the task.
When you have perfect knowledge, then you can predict what will happen in the parallel streams and — crucially — how they will integrate afterward.
English

Yes, agreed! Just let me go straight to Google Docs, not have to import the .docx then convert every time I want to make a change.
I also feel like when I try to find a workaround, Claude Cowork keeps trying to push me to view docs as .md files in Windsurf (which I don't use) which is pretty annoying as well. Hopefully this experience improves soon.
English

Claude in PowerPoint is now available on the Pro plan.
It also now supports connectors, bringing context from your daily tools directly into your slides.
Try it here: claude.com/claude-in-powe…

English
Shavin Peiries retweetledi

im curious to know why they aren't using an AI CRM?
Cam Wright@thecamjwright
Anthropic going with Salesforce while xAI uses Hubspot… interesting.
English
Shavin Peiries retweetledi

@0xSMW I've tried it, but it's never been good. I thought you had some special prompting technique lol
English

codex spark is back on and fast 1k/s tokens will change everything — added a test blog post to my site in 30s, it read other articles and wrote this somewhat like me


Stephen@0xSMW
codex spark completely unavailable
English

@SanjanaLaddha i wrote this when chatgpt first came out
shavinpeiries.com/chat-gpt-custo…
There's a bunch more related articles on using AI for research
English


Shavin Peiries retweetledi

Just shipped /last30days. A Claude Code skill for @claudeai that scans the last 30 days on Reddit, X, and the web for any topic and returns prompt patterns + new releases + workflows that work right now.
Last 30 days of research. 30 seconds of work.
👉 github.com/mvanhorn/last3…
English

@KusumW It's because most of them are using white-labeled products from Scienter or Idea Hub. Both are terrible. The first bank/finance company to invest into a skilled in-house product and development team will win just by differentiation and good UX.
English

@hobdaydesign @mschoening I definitely agree that it's useless if not acted upon.
English

@shavin47 @mschoening Understood. I'm just curious if anyone actually does that. I've been in situations where that was the plan, but it never happened, and I'm interested to know if we're all adding onboarding steps and then not doing anything with the data.
English

@hobdaydesign @mschoening From my understanding, it helps with attribution and figuring out where to place bets at the top of the funnel.
Let's say they've identified a group of people where the product is a perfect fit and retention is high. They can use the onboarding data to target similar audiences.
English

@shavin47 @mschoening But any proof that it helps for sales and marketing in a significant way?
English

@hobdaydesign @mschoening People frame it as “customizing the product for the user” but it’s a myth. It’s mainly for sales and marketing.
English

@mschoening Does anyone have any proof that these questions make a noticeable difference to anything? We added them to the sign-up flow at a previous company, but that wasn't based on any data and they didn't help.
English










