Pinned Tweet
I’ve learned a huge amount about AI from reading and exploring, but nothing beats getting hands on and learning by doing.
So, I set myself the challenge of building a useful and fully functioning AI-powered product.
Here’s what I learned.
georgecockerill.com/blog/how…
1
91
George Cockerill retweeted
Our music model @elevenlabs is coming together! Here’s a very early preview. 🎶
Have your own song ideas? Reply with a prompt and some lyrics and I’ll generate some for you!
George Cockerill retweeted
A Collection of Design Engineers
maggieappleton.com/design-en…
George Cockerill retweeted
my secret superpower is that I won't get offended if you send me a link to your Calendly
George Cockerill retweeted
I had a blast at our last Future Friday in Fitzroy, thanks to everyone who attended! Join us for another one this week!
It's a weekly GenAI hackathon and community event 🎉
Meeting Friday 1PM at Pony White.
Limited space - RSVP on meetup - Link in comments.
George Cockerill retweeted
Okay this got me thinking… when I’m listening to someone talk, I can start forming my response before they are done talking. Applying it to this…
- somehow stream the speech into model
- have model predict rest of user speech
- form response based on expected user speech
- as user speech continues, compare against predicted speech, and if different, update formed response
- when user speech completion noticed (long pause + reasoning), compare against predicted speech, and only if different (maybe embeddings?), generate new response, otherwise use cached response (which could be immediate)
You could probably inefficiently mock this up with orchestration (code) but ultimately feels like some clever cache mechanism within the model is probably where this goes
No idea, just thinking out loud
working on a demo of a bot you talk to
the flow:
1. speech to text (@DeepgramAI)
2. LLM (@GroqInc)
3. text to speech (early access w/ @DeepgramAI)
Groq speed is a bit variable, but overall really quick
Deepgram is super fast when you stream results (time to first data <280ms)
If you're exploring the positive impacts of Generative AI on digital products, come and join me and @agileben at this brand new event.
George Cockerill retweeted
If you have an idea that you haven't taken action on yet, and it's been more than a few days, find a smaller version of that idea.
Break it down and try the smallest possible thing.
George Cockerill retweeted
There’s subtle little-but-big paradigm shifts in the user interface for Vision Pro that make me pause and smile.
Take password entry. The password never turns into ••• dots — you can see every letter or number of a passcode you enter, because… only you can see it.
George Cockerill retweeted
BIG DAY — “Duly Noted” is finally here! If you've ever wanted to get more from your digital note-taking, this book is for you.
P.S. while you can buy it in the usual places, consider supporting independent publishers by getting it directly from @RosenfeldMedia. As a a bonus, their ebooks are DRM-free!
rosenfeldmedia.com/books/dul…
#ToolsForThought #PKM #PersonalKnowledgeManagement
Creating UIs as needed to augment the conversational experience: youtu.be/v5tRc_5-8G4?si=G-Q-…
1
7
211
George Cockerill retweeted
You can understand your audience through the patterns of how their minds worked, how they reacted and handled their reactions, and what guiding principles drove their decisions.
🧵 1/3
George Cockerill retweeted
Well that's it folks, this meme can now be retired.
I can now draw a great owl by following these 2 simple steps. 😂

ALT The screenshot displays various illustrations of owls. On the left, there's a basic outline drawing of an owl made with two circles for the body and head, next to a detailed grayscale pencil sketch of an owl perched on a branch. In the center, there's a digital drawing prompt instructing users to draw a realistic owl in a scientific style using only gray lead pencil, with an incomplete sketch inside. On the far right, there's a realistic digital illustration of a white owl with yellow eyes and intricate feather details. The screen also features tabs and icons from an online drawing tool on the right, suggesting the user was in the process of drawing or editing the images.
George Cockerill retweeted
I made a quick proof of concept that generates pages using GOV.UK Design System based on a prompt. You can ask it to make a specific form page, a start page, a page with a two-third one-third grid, and all sorts of other things. It was so easy 🫢
Ask @LukeW digital product design questions and get answers based on his 2,000+ articles, 375+ presentations and 100+ videos from the past 28 years.
ask.lukew.com/chat
96
George Cockerill retweeted
Inter version 4.0 → rsms.me/inter/
After two years of work on a new version of Inter, I'm so proud that it's finally released. (It's been in beta for the past year)
The changes are vast. Most notably there are six additional "Display" designs for large delicious type
Photoshop for text by @kepano - "In the near future, transforming text will become as commonplace as filtering images. A new set of tools is emerging, like Photoshop for text."
stephango.com/photoshop-for-…
64






















