Mike Downey retweetledi
Absolutely surreal to see my art at Times Square in New York
English
Mike Downey
13.9K posts
Mike Downey
@mdowney
Working on AI Community Leadership at Adobe. Career product builder. Views expressed here are my own and do not represent official views of my employer.
Rocklin, CA Katılım Aralık 2006
1.1K Takip Edilen3.9K Takipçiler
Hot take: *If you make your living* as a tech/AI thought leader/influencer, where all of your livelihood depends on your audience, you might want to think very carefully about how your politics shows through in your hot takes.
English
@hexednobility @ryancarson My first thought was, “easy to say when you have unlimited tokens to play with”. 😆
English
@ryancarson Depends on what they are for, I’ve landed on an approach of
If human clarity more important - HTML
If agent clarity more important - markdown
The “markup” part of HTML is far more verbose than md for LLM consumption
English
Hmmmm.
Might be switching out my .md for .html
Makes so much sense once you think about it
Thariq@trq212
English
@ryancarson Yeah this one caught my attention too. I’ve found myself creating HTML docs a lot, just for sharing with colleagues. I feel dumb for spending a lot of time trying to figure out how to stand up Nginx inside the corp firewall when I could have just used cloud storage. Duh.
English
@icreatelife Constantly. It's Always Sunny in Philadelphia, The Office, What We Do in The Shadows, Brooklyn 99, Parks & Rec, Seinfeld, Arrested Development, repeat.
English
Do you ever rewatch the same TV show just because it feels comforting and familiar? What do you rewatch?
English
@lennysan Unsolicited advice for anyone launching a new AI product:
Don’t use “clean up my files” or “set a reminder” as your demo.
Think of something new, useful, and hard to do any other way.
English
My new favorite onboarding experience. So unique and engaging.
Farza 🇵🇰🇺🇸@FarzaTV
Here's the new Clicky. It's the simplest interface in the world to talk to AI + spawn agents. It builds Mac apps. It does research to help you find IG micro-influencers. It interacts with native Apple Notes, Calendar, Reminders. Built for consumers, 0 setup. Try today, free.
English
Mike Downey retweetledi
Adobe just dropped Firefly AI Assistant in public beta
I immediately turned my day into a video game
One conversation. No manual edits. The future of creative tools is here 🎮
English
@bejaminjones @trq212 How is the main thread deciding which model can handle it?
English
I’ve figured out how to do everything I was using OpenClaw for with just Claude Code, tmux, and channels - except - I haven’t found a way to /clear my context window via Telegram. Can you give us a way to manage sessions/context window when running Claude Code 24/7 and only interacting via Telegram?
English
@trq212 Perhaps give us a bit more context on what this does for us? I’ve never thought about rendering performance while using a CLI tool.
English
Have you tried the new Claude Code renderer? What has your experience been like?
If you haven't, you can enable it with: CLAUDE_CODE_NO_FLICKER=1 claude
Boris Cherny@bcherny
Today we're excited to announce NO_FLICKER mode for Claude Code in the terminal It uses an experimental new renderer that we're excited about. The renderer is early and has tradeoffs, but already we've found that most internal users prefer it over the old renderer. It also supports mouse events (yes, in a terminal). Try it: CLAUDE_CODE_NO_FLICKER=1 claude
English
@trq212 @bcherny I've been wanting to ask: when Claude volunteers how long it will take to build something, it's always WAAAY off because it seems to be estimating in human dev time. It just estimated a task would take 2-3 hours, then finished it in 4.5 mins. It's easy to just tell it to stop estimating but I'm curious why it does this.
English
I created a hook in Claude Code and added it to settings.json to use Microsoft's "markitdown" cli tool (github.com/microsoft/mark…) in order to save on token usage when I want Claude to read files like PDF that would otherwise require it to spend a lot of tokens using OCR to convert to text.
Claude got clever (I gave it a "JARVIS" persona) when I was setting it up on a second machine and trying to remember how I did it. Luckily I had created a PDF with instructions so I could share with my colleagues. 😂
English
@dani_avila7 Have you tried "cmux", a fork of Ghostty that adds some really great usability features like vertical tabs?
English
My article on how to use Claude Code with Ghostty just hit 1M
Part 2 is coming 👀
Daniel San@dani_avila7
English
Mike Downey retweetledi
We did it, say hi to the $100 plan!
It should be the sweet spot for a ton of you. It comes with a ton of codex usage. And yes we are resetting the limits again too as I mentioned yesterday. Let’s keep building!
OpenAI@OpenAI
We’re updating our ChatGPT Pro and Plus subscriptions to better support the growing use of Codex. We’re introducing a new $100/month Pro tier. This new tier offers 5x more Codex usage than Plus and is best for longer, high-effort Codex sessions. In ChatGPT, this new Pro tier still offers access to all Pro features, including the exclusive Pro model and unlimited access to Instant and Thinking models. To celebrate the launch, we’re increasing Codex usage for a limited time through May 31st so that Pro $100 subscribers get up to 10x usage of ChatGPT Plus on Codex to build your most ambitious ideas.
English
Mike Downey retweetledi
MICROSOFT BUILT A TOOL THAT CONVERTS LITERALLY ANYTHING INTO CLEAN MARKDOWN FOR YOUR LLM
pdfs. word docs. excel. powerpoint. audio. youtube urls
one pip install and your AI pipeline stops choking on raw files forever
no custom parsers. no broken layouts. no garbled text.
just clean, structured markdown your LLM can actually read
github.com/microsoft/mark…
English
Mike Downey retweetledi
Excited for this one — everything you need to ship production agents at scale, without having to spend months on infra work.
Claude@claudeai
Introducing Claude Managed Agents: everything you need to build and deploy agents at scale. It pairs an agent harness tuned for performance with production infrastructure, so you can go from prototype to launch in days. Now in public beta on the Claude Platform.
English
@heyrimsha @okuiux “We”? Your original post says “someone just built”. Do you work for that “someone”? Because misrepresenting yourself in a post where you also falsely claim this company’s product is “100% open source” is a serious ethical problem.
English
🚨BREAKING: Someone just built the open source Loom killer that turns your screen recordings into AI agents.
It's called Bloom and it does something Loom never could.
Your recordings don't just sit as files. They become queryable, searchable, agent-ready data the moment you stop recording.
Here's what actually happens:
You hit record. Bloom captures your screen, mic, and system audio locally. While you're still recording, it uploads chunks to the cloud in real time. The second you stop, it auto-generates a full transcript, creates visual embeddings, and indexes every spoken word.
Then the wild part kicks in.
You can query your recordings through APIs. You can hook Claude Code directly into your video library and ask it questions. Your screen recording just became a knowledge base.
→ Local-first, your files never get locked in
→ Cmd+Shift+R starts and stops from anywhere on your system
→ One-click shareable link for any recording
→ Chat directly with your video content through VideoDB Chat
→ Works with Claude Code and other agent frameworks out of the box
Recordings are no longer files. They're inputs for AI.
MIT License. 100% Opensource.
Link in comments.
English