AI-Native digital product & venture studio 100+ products since 2013. Toronto 🇨🇦 + Barcelona 🇪🇸 Agentic AI · product design & engineering · fractional CTOs
Our Ventures →
Joined April 2012
- Tweets3K
- Following33
- Followers1K
- Likes4.7K
Pinned Tweet
iPhone Duo ships Oct 23.
Introducing — the iPhone Duo Live Browser Simulator.
designmd.me/iphone-duo-test
Every website looks fine — right up until it folds.
See your site render on the iPhone Duo before your users do across all the new aspect ratios and viewports (folded / unfolded) — no install needed!
Brought to you by the team that built the most popular DESIGN.md generator (~78K DESIGN.md markdowns generated)
Test away!
Crowdlinker retweeted
apps get duo updates one by one. websites don't get an update button.
unfolded, the duo is 890 × 626 css px, wider than it is tall. most sites were never laid out for that.
we built a free tester at @crowdlinker: load any url at both duo sizes and see what breaks.
We need 10 Slack workspaces to install a free app.
Slack won't review Kudos for the Marketplace until it's live on 10 active teams. We're short.
Type /kudos, tag a teammate, say what they did, pick a GIF. Free. No catch.
Link below. Reply "in" when you've added it.
Add Kudos to Slack: slack-app.kudos.crowdlinker.…
Demo first: kudos.crowdlinker.com
48 seconds on why AI labs are buying company data, and the one question to ask every software vendor.
Google bid $10M on a bankrupt airline.
Not for the planes. For the inbox: 100M employee emails and 500M Teams messages, to train AI.
Here's where AI training data is coming from next, and why it's your problem too.
ALT Headline "Google bid $10M on a bankrupt airline. Not for the planes. For the inbox." An illustrative inbox with redacted senders and a "Sold $10,000,000" tag.
Ask every vendor in writing: do you train on our data, or sell it? Then make the answer survive a shutdown or acquisition.
Which tool in your stack worries you most?
ALT Five moves for your next vendor renewal, from mapping where work data lives to deciding what your data is for.
We are doing this for enterprises.
Start owning your AI infrastructure. Stop renting it. Start off with a free LLM bill teardown.
crowdlinker.com/contact/own-…
five guys at an American startup are trying to pull off what DeepSeek did to Silicon Valley
None of them have an AI background, a PhD, or worked at OpenAI.
that’s what happens when investors realize American AI labs are spending way too much money… and something like that is about to happen again
instead of building from scratch, they took an existing Chinese model and spent $300k training it to handle white collar work
and they’re releasing the weights so companies can run it on their own servers instead of paying a provider every time they use it
ironically, China’s headstart in cheap models is giving tiny American startups a way to compete with companies worth hundreds of billions
Spectr is officially out of beta!
The team redesigned the website to celebrate the launch and made this wild motion video to go with it.
Check it out at spectr.pm
Meet the brand new Spectr. 🟣
Redesigned, rebranded, and built to run your product ops for you — every meeting turned into requirements, specs and ready-to-ship stories.
Take a look → spectr.pm
49% of US adults use AI chatbots.
~2% of households pay for it.
Smartphones took 6.5 years to get from 2% of America to 50%. The internet took 7.5. PCs took 19.
Paid AI is at year zero.
How many years will it take for 50% of US households to start paying for it?
22%Faster than anything yet
33%Beats the smartphone
33%Smartphone/internet pace
11%Car and PC pace
9 votes • Final results Crowdlinker retweeted
That Granola → Paste into Grok → Update a file loop works until the third meeting of the day.
We got annoyed enough to make the last step write the stories into Linear or JIRA after you glance at them.
spectr.pm is the pain the summary, or the "now put it in the tracker" bit?
Hey @bot make me proud.
1T dollars.
No mistakes.
New Feature Release for @tryspectr
Spectr now scans your Figma designs and attaches relevant pages/screens to your stories. You can then easily import these stories to your product board in Linear, JIRA or Shortcut.
Your whole product delivery pipeline can be run through spectr.pm
Model upgrade on @tryspectr.
Using Sonnet 5.5 across core generations:
- spec extraction
- story generation
- context gathering and the AI Assistant’s default model now run on Anthropic’s Sonnet 5.5 as well.
Result: noticeably sharper spec sheets and stories from the same meetings and faster.
Sonnet 5.5 also joins the AI Assistant model picker (replacing Sonnet 5).
Requirements stay on Opus 5.5 for versioned BRDs and PRDs.
