Jude Edwards

Building with AI at @Blocks

[Jude Edwards @jedwards_27]

AI can build an app in an afternoon. But getting it safely into other people's hands is a whole other challenge!

This is the problem that I've been working on these past few months. I'm proud to finally share how we solved it with Block App Kit!

https://t.co/hXm6NdcMUW

[Boris Cherny @bcherny]

As engineering, product, design, DS, etc. melt into a new kind of role, I was reflecting on what roles might look like in the future. For example, when I look at the Claude Code team I see what I think is five archetypes:

1. Prototyper: comes up with brand new ideas; churns out many ideas, most of which don't ship
2. Builder: quickly turns a prototype/idea into production-grade product/infra
3. Sweeper: cleans up the UI, simplifies the code and system, unships, optimizes performance
4. Grower: takes a product that has been built and iterates on it to improve Product-Market Fit
5. Maintainer: owns a mature system to make it secure, reliable, fast, and efficient as it scales

Many people span across 2 roles, and sometimes 3 roles. I also notice that these roles are not really tied to job function -- eg. across Anthropic, some designers match category 1, some 2, some 3; same for engineers, PM, DS.

A healthy team needs a mix of these, depending on the product:

- A product that is new and pre-PMF needs people that are strong at 1+2+3
- A product that is growing and has found PMF needs 2+3+4 and some 5
- A product that has strong PMF needs 3+4+5 and some 2

Maybe product roles of the future will look more like this, and less like the domain-specific roles of today?

[Jude Edwards @jedwards_27]

@0xblackbirdoff Yes! The team is working on this now. There are quite a few pieces involved so (as you might guess) it’s not a straightforward path to open source. If you’re interested in trying it out in the meantime, please feel free to email me. My email is provided at the end of the article.

[Jude Edwards @jedwards_27]

AI can build an app in an afternoon. But getting it safely into other people's hands is a whole other challenge!

This is the problem that I've been working on these past few months. I'm proud to finally share how we solved it with Block App Kit!

https://t.co/hXm6NdcMUW

[Jude Edwards @jedwards_27]

Does anyone have a good prompt/skill recommendation to generate nice-looking mermaid diagrams from code or design docs?

This prompt would sit on top of something like https://t.co/co2R8Oz9Oh or https://t.co/oaQTe723cv to improve the quality of the agent's output.

[Jude Edwards @jedwards_27]

@bakshim Re: sandboxing, we leaned heavily on our security partners to help us ID all the data ingress and egress paths within the building and viewing flows. We then locked each path down using the available controls for each piece of the stack (e.g. CSP, JWTs, etc). We deny by default.

[Jude Edwards @jedwards_27]

@bakshim Great question! We baked observability into the platform to emit logs and events for each app session. We use a combination of deterministic tools (e.g. Datadog monitors and alerts) and non-deterministic tools (e.g. Builderbot automations, ad-hoc agent sessions) on top of that.

[Jude Edwards @jedwards_27]

@behradkhodayar What you're describing is a combo of Builderbot (https://t.co/HSgVqxXqV6) and Block App Kit. But yes, that's an accurate representation of how many engineers at Block are working these days & one way that BAK fits into the workflow for development 👍

[Block @blocks]

We built an internal AI system called Builderbot. It coordinates agents across our entire codebase. Engineers tag it in Slack, and it researches, plans, and ships. The story so far:

- 200,000 operations per day.
- 1,500 pull requests merged per week.
- 15% of all production code changes across Block.

What used to take months now takes days.

How we built it: https://t.co/ixvmNK87Du

[Jude Edwards @jedwards_27]

@Kevinduan2014 All of the above! The important thing is to make it close to impossible for the agent to deploy unsafe apps. Whether that's sensitive data being baked into the app code, secrets being passed around in the open or a weak auth model (e.g. over-reliance on security-by-obscurity).

[Block @blocks]

We built an internal AI system called Builderbot. It coordinates agents across our entire codebase. Engineers tag it in Slack, and it researches, plans, and ships. The story so far:

- 200,000 operations per day.
- 1,500 pull requests merged per week.
- 15% of all production code changes across Block.

What used to take months now takes days.

How we built it: https://t.co/ixvmNK87Du

[Jude Edwards @jedwards_27]

My son and I vibe-coded a fish tank together this morning! Legitimately spent 45 mins just watching the fish and adding in different features. So cute and fun :)

[Jude Edwards @jedwards_27]

:bufo_waves_hello_from_the_void: