CEO & Member of Technical Staff @ampcode, founded @sourcegraph
Joined February 2007
- Tweets 12,640
- Following 3,902
- Followers 16,745
- Likes 31,210
1,016 Photos and videos
6h
You can now proactively verify your identity (with a passport or government ID) in case it’s needed for future frontier model access in Amp.
We think it probably will be, and we want Amp to keep giving you access to the best models available to you.
We can’t guarantee access criteria or timelines. Those depend on (highly uncertain) government and model lab policy. We don’t plan to impose any additional restrictions beyond what is required by law and the model labs.
We are covering the cost for identity verification for all users, and we’re using Stripe for identity verification, so Amp stores nothing and sees only the outcome.
ampcode.com/settings/identit…
16
10
101
21,366
5h
By the way: I don’t think that this ID verification KYC, or really anything that exists today, would work to enforce export controls. That isn’t what we built this for. I believe the export controls applying to non-US nationals (or whatever the precise details are) imposed on Friday are an emergency measure, and there will be a negotiated compromise between the parties or in law that creates a much more efficient controls regime that’s easier to enforce. I have to imagine ID verification will be a part of that.
3
9
1,800
Quinn Slack retweeted
Jun 12
Surprising take: @AmpCode Web UI is the first AI interface that’s actually pulling me away from the CLI.
Not because it tries to do everything. Because it gets out of the way. Flexible, controlled, no useless bloat.
It’s not complete yet, but the direction is right.
My only asks:
• start threads directly inside server/local workspaces
• choose models per thread
• create isolated sandboxes per project
• keep keyboard ergonomics at CLI level or better
Right now I start a thread from the server, send “hi” just to make it appear in the Web UI.
When that flow lands, this thing is going to be hard to beat.
Knowing what the AmpCode team has already shipped, maybe this take shouldn’t be surprising at all.
2
1
36
5,664
Jun 13
OH: "Fable is so expensive, but now it makes GPT-5.x seem cheap in comparison, so I'm using more of that"
10
1
100
7,970
Quinn Slack retweeted
Jun 12
Day 3 with Fable.
Gave a huge prompt to implement a feature across CLI, web server, and another server to both Fable and deep^2 in Amp.
deep^2 was done before I went to the gym. It stopped short. Sent another prompt. $20.
Fable ran for 1hr40min and cost $350.
Results:
They both understood the assignment and built the same thing. Maybe that's due to my prompt.
Fable's worked on first try. Well done.
Deep's looks correct but didn't work on first try.
$20 vs. $350.
I'm sure I could get deep^2 to make it work and we'd end up at, what, $40? While Fable is now at $457 after I asked some follow-up questions.
46
16
511
360,245
Quinn Slack retweeted
Jun 11
Day 2 of using Fable as my main model in Amp.
Some impressions.
It's smart, yes. The codebase for the new Amp contains a lot of intricate dist-sys business logic: this resource has this lifetime, this depends on that, this runtime has this behavior and thus, ... Fable seems to grasp this easily. I rarely had to explain business logic.
But it also feels dumb somtimes? I asked it to fix a bug where we wouldn't retry the creation of a resource after a timeout. It fixed it. I looked at the code and was skeptical. "Are you sure?" And it went "You're right, this is not correct." Then it took another turn and got it. Felt strange, like 2025.
It's slow. Oh my god is it slow. Now, I don't mind slow models that much if I can rely on them doing the right thing, which is what started with 5.3-Codex, but well, see previous oint.
It's expensive. Oh my god is it expensive. I have 3 Amp threads open right now that are at $704.27, $451.65, and $271.12. These are *not* outrageously large or crazy threads. I wouldn't have thought about their cost with any other model.
Those costs confused me, until @nicolaygerold pointed out that I'm constantly busting the token cache. We're using the cheaper Anthropic cache writes with a 5min TTL. That means if you don't use a thread for 5min, then the next inference call busts the cache and costs *a lot*.
Now combine point 2 (sometimes it's not that smart?) and point 3 (it's slow) and I end up with multiple checkouts where I'm constantly waiting on a model but might miss when it's done and then once it's done I have to do a follow-up: cache bust russian roulette.
Since it's only Day 2, I'm sure I'll have more thoughts soon, but so far, my verdict:
Its smarts are overshadowed by its ergonomics and economics.
20
14
303
39,048
Jun 10
New Amp plugin API fields for those of you with personal & work Amp account - you can make a plugin to show your current signed-in username in the Amp CLI (or other stuff)
Jun 10
Just pushed a new API for Amp plugins that lets you read the current user, so you can show it in the status bar at the bottom. I did it live here, or just ask something like "Make me an Amp plugin to show my current signed-in username or email in the Amp status bar", and it should work.
As for switching: stuff like alias ampp='AMP_API_KEY=$(cat ~/.my-amp-api-key) amp' (and same for ampw for work) works pretty well. We don't have a builtin account switcher. Let me know if you think we should.
1
32
3,179
Jun 9
Some more impressions of Fable
Jun 9
amp update
amp plugins add --auto-update @amp/fable-mode
amp --mode claude-fable-5
4
64
4,680
Jun 9
If you keep seeing model N 1 fix code from model N (as I do): you should care less about today's code, and more about knowing what you want.
From our team chat:
Jun 9
amp update
amp plugins add --auto-update @amp/fable-mode
amp --mode claude-fable-5
5
2
51
6,425
Jun 9
amp update
amp plugins add --auto-update @amp/fable-mode
amp --mode claude-fable-5
9
14
130
42,964
Jun 9
- Will be built in soon (spoiler alert: this model meets our bar, it's very good!)
- Pass `--target workspace` to the plugin add command to add the plugin in your repository instead of ~/.config/amp/plugins, so other people on your team get it too
15
1,688
Jun 9
The models change. Make sure you’re using an agent that changes with them.
Jun 8
2
76
7,705
Jun 8
I used an iPad and MacBook Neo to save my table at a restaurant while I took the kids to the bathroom
In the USA, in California, in a public place
Narrative violation: nothing got stolen
12
60
10,721
Jun 7
Problem I'm hearing: many people are juggling a lot more simultaneous-ish @AmpCode threads than before, with remote control and the new app
More help on the way, but quick tip: make a macro to send the msg "Commit the changes then archive this thread" to your thread.
Here is the prompt I gave Amp to make an Amp plugin keymap to do this (took ~30 sec, see the video for a demo):
> Make an Amp plugin in .amp/plugins/ called commit-and-archive.ts that registers a command that sends a user message to the current thread "Commit the changes made in this thread, then run `amp threads archive ${ctx.thread.id}`, and keymap that in .amp/settings.json to "<leader> c"
Then Ctrl O `plugins: reload`, and it works. This uses Amp plugins (ampcode.com/manual#plugins) and the keymap support we shipped today (x.com/sqs/status/20633506015…).
Jun 6
Just shipped some "how did we not have these yet"s in Amp:
• Customizable keybindings (see `amp config keymap` and ampcode.com/manual#cli-keyma…)
• Mode switcher that tells you about each mode (Ctrl S in the Amp CLI to use it)
Watch me use these in Amp to fix a bug:
6
3
83
6,588
Jun 7
Here is the plugin code it wrote for me, but you should just have Amp write one for you, it's good to get into the habit of customizing your own tools on a whim:
1
6
1,182
Jun 7
(spoiler alert: you can see the Amp web UI in the video has a few new things that aren't shipped yet)
8
432
Quinn Slack retweeted
I don't believe in "tokenmaxxing" as an end in itself, but certainly it is fine as a means to a valuable end
still going strong with @AmpCode, Dec 1 to date
2
1
25
2,311
Jun 6
Thank you @goinggodotnet for the kind words about Amp in youtube.com/watch?v=2RnSh3fM…! Keep the feedback coming.
3
1
35
3,132
Jun 6
Just shipped some "how did we not have these yet"s in Amp:
• Customizable keybindings (see `amp config keymap` and ampcode.com/manual#cli-keyma…)
• Mode switcher that tells you about each mode (Ctrl S in the Amp CLI to use it)
Watch me use these in Amp to fix a bug:
8
10
122
10,921
