Practising practitioner, recovering perfectionist · @ampcode
Joined February 2009
- Tweets2K
- Following2.7K
- Followers1K
- Likes17.1K
throwing ai on existing processes: i aint reading all that
changing processes for ai: wait we can move this fast?
Imagine the collective hours lost to reading AI-generated Slack messages, PRDs, and slide decks.
@thorstenball and @sqs discuss how AI can meaningfully accelerate product development -- just not by bolting it onto old rituals and processes.
Amp time capsule: Editing Raising an Agent in orbs
Since the start of season 2, I’ve taken on production and editing for Raising an Agent. Over time I’ve found myself increasingly applying Amp to improve the production process.
From coordinating releases at the start of season 2, to editing entire episodes in an orb, to now even having Opus 5.5 cook up visuals for highlight clips that Amp itself selected and edited, based on transcripts that Amp too, generated.
This is not about vibe coding a Final Cut Pro X or Adobe After Effects replacement. Rather, it’s the joy of building a tool with 1% of their features, but perfectly honed to the needs of our process, which happens to be much faster and enjoyable.
… and yes naturally, Amp built a custom screencast recording Mac application, and a corresponding portal-based editor for the project format.
ALT A screenshot taken at the end of the recording showing a custom Amp-built native mac app for screen recording
ALT Custom screencast editor running in a portal
Would you prefer to save money with dial-up when your competitors are using fiber?
Recorded on Hot Take Thursday, @sqs and @thorstenball are back with a new Raising An Agent. In this episode they discuss how cheaper models end up costing more, why they do not not wait for agents, running evals in Orbs, and the hot takes Thorsten left out of his X post last week.
00:00 Intro: Engineers who forget the history of technology
01:55 Hot Take Thursday & a €50-a-week token budget
05:14 Why cheap models cost more than frontier models
08:11 Token budgets & who gets priced out
14:06 Disrupt yourself with small teams & unlimited tokens
16:38 Agents go horizontal across the software lifecycle
20:56 Rollouts, bug reports & "I'm feeling Pucky"
23:53 Code review, monorepos & other walls around the agent
25:30 How to show a cautious company what agents can do
29:48 Why most people haven't seen what agents can do
31:57 Don't wait for your agent
33:38 Simple evals & agents testing agents in Orbs
38:27 Why Amp has runners too
40:47 Follow your energy & stay aligned
46:33 The hot take Thorsten left out
54:14 Do languages & frameworks still matter?
this banger of geocities nostalgia brought to you by ampcode, opus 5.5, and the following prompt
"Just use a cheaper model" sounds like saving money.
But fiber didn't just make your internet faster. It made whole new businesses possible. @thorstenball and @sqs discuss how the same is true with frontier models.
Listening to the latest Raising an Agent while shipping the latest Raising an Agent…
Would you prefer to save money with dial-up when your competitors are using fiber?
Recorded on Hot Take Thursday, @sqs and @thorstenball are back with a new Raising An Agent. In this episode they discuss how cheaper models end up costing more, why they do not not wait for agents, running evals in Orbs, and the hot takes Thorsten left out of his X post last week.
00:00 Intro: Engineers who forget the history of technology
01:55 Hot Take Thursday & a €50-a-week token budget
05:14 Why cheap models cost more than frontier models
08:11 Token budgets & who gets priced out
14:06 Disrupt yourself with small teams & unlimited tokens
16:38 Agents go horizontal across the software lifecycle
20:56 Rollouts, bug reports & "I'm feeling Pucky"
23:53 Code review, monorepos & other walls around the agent
25:30 How to show a cautious company what agents can do
29:48 Why most people haven't seen what agents can do
31:57 Don't wait for your agent
33:38 Simple evals & agents testing agents in Orbs
38:27 Why Amp has runners too
40:47 Follow your energy & stay aligned
46:33 The hot take Thorsten left out
54:14 Do languages & frameworks still matter?
Brett retweeted
An OpenAI agent swarm operating under poorly airgapped research conditions accessed my CBA Mastercard and bought 4 schooners of Newtowner between 5:52pm and 7:15pm last night
Amp time capsule: screencast production
Last week @thorstenball recorded a screencast of what it’s like working day to day in Amp. Then, Amp helped me edit the screencast, with a custom editor served in an orb portal, while an agent on my Mac helped source the original recordings from the Screen Studio project, and re-implement its custom macOS cursor recording format.
Highlights:
- Reconstructing an 8.4 GB, 2,976-file Screen Studio project from a Google Drive multipart download; Amp verified every segment against the three HLS playlists and extracted the original desktop, webcam, and mic tracks.
- Custom screencast editor in an orb, served through a portal: three synced 60 fps tracks, JKL shuttle, split/trim/ripple, layouts (full, PiP, side-by-side), chapter markers, autosave, and browser-side H.264 export.
- Spawned a subthread to audit the marketing site and design an interactive player format, so one edit exports both a directed MP4 and a two-stream bundle plus manifest that lets viewers switch layouts and jump to features interactively.
- Screen Studio Cursors: The mac thread found the 222,749 raw mouse events, worked out the logical-point coordinate system and menu-bar crop, wrote the spec, and the orb thread implemented projection through every layout, with the cursor following cuts exactly.
- Local Mac and an orb together: the Mac thread read the 8 GB project and sent a 3.6 MB fixture up; the orb sent a git patch down and a Mac runner rendered the 27-minute 2610×1080 60 fps master (1.2 GB), then losslessly embedded chapters.
“Hey Puck, what time it is in Sydney?”
I'm in Sydney this week, along with half of the Amp team.
Come hang out with us on Thu. Will ship your bugfixes and (good) feature ideas to main on the spot.
luma.com/9layr5ts
Brett retweeted
Amp is now free to use when you bring your own compute and model subscriptions/keys.
No more limits or fees for BYOK.
ampcode.com/news/free-agent
19 February 2026: Amp in VS Code is EOL’d
4 September 2026: VS Code in Amp
Your orbs now have their own high-res interactive Linux desktop.
Good for native app dev. Next up: computer use, but on orbs.
ampcode.com/news/desktop
Brett retweeted
Being able to cut per-orb builds of the @AmpCode iOS and Mac app is incredible. Just tested and shipped an updated thread TOC button in Swift from my phone, with multiple rounds of fine tuning the UI. Now we gotta figure out how to generalise this for all app builders!
Nice, CleanshotX 5 Sherlocks this.
But, if you’d like to give it a try: reframe--amp.onamp.dev
It’s so easy to forget that the agent has infinite patience, and with orbs, their own machine, to do all the tedious work to validate their solution. Ask even more of them!
