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?

"I'm just pointing at things and saying, make it like this and ship this and do this. The altitude is so high. It's wild. Even compared to two months ago." @thorstenball on using agents for the whole lifecycle of a feature: building, testing, shipping, monitoring the rollout, toggling feature flags, talking to users.

What is the computer for now? In Raising an Agent S2E4, @sqs and @thorstenball explore how agents can do more than just ship code, discuss recent outages, and ask what happens to our computers when more work moves into Orbs. 00:00 Intro: What is the computer for now? 01:32 New models & aiming higher with agents 04:42 An agent's job after the code ships 10:19 Mini evals & keeping the team's context in threads 13:38 Performance testing & production canaries 18:36 Amp's recent outages & what went wrong 24:44 Shipping faster & the tradeoffs of redundancy 28:34 How fast can an engineering team go? 32:27 Making production logs & metrics useful to agents 37:15 Beyond local development 41:30 Agents building 3D models & virtual worlds 43:35 Simpler devices & dictating work on the go 47:29 What happens to the tools we love? 49:27 Steering, queuing & changing workflows

Queueing your messages to the agent made sense for easily distractible models. Now? Amp sends your message to the agent without waiting. It's way faster and more natural. Plus, more information makes the agent do better.

"I could just replace that with EC2 and some shell scripts and get the same thing." But could you? Thoughts and question of when to build and when to buy. And @sqs putting @thorstenball fully on the spot by asking him whether people should build their own Amp or not.

Amp now has GPT-6 Astra. It's very good and uses your linked ChatGPT sub. Amp's `high` mode is gpt-6-astra medium (not xhigh, why in thread). In other Amp modes soon. BTW: customize your Amp dial at or pick any GPT-6 Astra effort level in Raw Models

Amp's ultra mode now uses Fable 5.1 Some notable improvements compared to Fable 5: better steerability, better at writing, and 35% cheaper per thread. Read the full post with our observations and recommendations here:

Raising an Agent, S2E3: When Tokens Flow Like Electricity. @sqs and @thorstenball on what happens when intelligence gets cheap: build vs. buy flips, why everyone building software to build software to do X is probably incorrect, and when software starts to feel like owning a house instead of renting one. 00:00 Intro: Cheap intelligence & doing X instead of building software to do X 01:06 Back home after Munich & what's changed in how we build 02:06 Predictions: the price of intelligence drops, tokens flow like electricity 05:54 Everyone is building dev tools & why that's the wrong bet 06:57 Pet theory: engineers can't believe it's this easy 10:33 Orbs vs. EC2, a VPN & tmux 17:18 Build vs. buy when the models keep changing 20:14 If Amp is worth buying, why isn't an issue tracker? 27:34 Killing features & vibe coding the 2% of a SaaS product you need 30:03 Build on a solid core: glue code is the worst part 33:19 The two Amps: jellyware, homeowners & renters 37:01 Exporting the Amp way to more teams 43:42 Opinionated software from people you trust