FailMore

I think alignment is the missing part of this post.

Creating something visually interesting Invisible-Cities-ish is basically a task with a billion or more satisfactory outcomes. You don't have a precise view on what makes an invisible city a good one as long as it has all the fancy looking bells and whistles. This means the AI can generate a bunch of tokens which as long as they technically work (which it seems the max-thinking did not fully) produce a "good" result. This makes you think - wow AI is impressive. However if many things in the system really mattered, it might have been much less successful as the AI may have made poor assumptions. This was basically a task that given a large volume of tokens and technical competence it's fairly easy for the AI to do a good job at because we don't evaluate it with any depth/precision. I think the same is true for music, where there are basically an infinite number of combinations of sounds, beats, melodies, etc. which are pleasant to hear - and songs aren't mission critical - so alignment is not a big issue. Generative AI is good at these tasks.

I built and use https://smalldocs.org, which is a rich document format (mermaid diagrams, nice formatting, etc.) for agents. So I’ll be like “sdoc me the architectural ideas for this feature” and it helps.

I think this is an interesting way to explore a possible future. Cool to see. Ty

API pricing?

I have a 30 min cycle of polling yahoo finance every 2s, one per share/index every 2 sec if that makes sense

Thanks for the feedback. Yeah agree re that. Will update soon.

Back online

Thanks, resolving!

What does that mean?

Thanks

> it takes months, years even, to notice the effects of bad architecture or of unmaintainable code.

I think this is a bad take... if you are going that long without noticing, you are (hopefully) delivering end user value all that time. That is the main driver of code. You can normally dig/hack/rearchitect your way out of an ugly code situation. If you've been building value for the last year based on the hacky code, that's a win.

I have leaned heavily into it… and became good at “agentic thinking”, but now I’m going to reverse. I have become dumber, sloppier, lazier, more distracted. I’m less happy than before. It’s hard to escape the productivity addiction and expectations though.

I felt the same when reading

I’m very interested in building tools for coding agents (where I spend a lot of my time).

I’ve been building https://smalldocs.org, it’s a simply styled, but functionally deep, artifact layer for agents. It’s open source and free to use.

Because it can incorporate charts, diagrams, spreadsheets, etc. in one artifact, it’s good for creating (and optionally sharing) agent created analysis.

It has a local library for all your Markdown files, and a cloud library (my one paid feature) too. Both support tagging and rich search/filtering. I find it turns SmallDoc artifacts into little checkpoints in work I can easily circle back to.

Thanks for reading!

As in you change one parameter and the whole equation changes or something else? (I'm bad at math!)

I think this is a very interesting article because it raises an idea I had not considered: these companies found PMF and huge growth through satisfy the demands of coders, it is interesting if they are in a bind where improving the model in one direction worsens it in others

For a personal project a while ago I peeked at Gemini's reasoning tokens in their coding CLI. I was pretty shocked. It was a bit like discovering Marvin (from Hitchhiker's Guide) was hiding in there all along. There was a lot of concern about my needs, "The user want's us to respond in a simple way...", "The user wants a clean front end...". Under the hood the poor model seemed very anxious to please with a hint of depression. It was a bit sad to see!

I’m still stuck on my iPhone 12 mini because I like the size

Because it's hard to find the time to read an academic paper I had an agent summarise it in a few slides:

https://smalldocs.org/s/6kEgfy54oclH4KR9HX847w#k=ywVL86PcTCo...

It's an interesting result (agents develop biases in their context) which reflects a lot of my experience working with agent, where I observe a lot of, what I kind of call, "context nudging" - where a droplet of an idea in an agent's context pushes its direction/output significantly. When it happens to me it always makes me question the type of intelligence LLMs provide.

[I am the developer behind SmallDocs. Source: https://github.com/espressoplease/smalldocs]

Surprised it has not been mentioned, but I think relying on https://github.com/vercel-labs/skills is sound. It handles global installs for a wide range of coding agents. If I was working on an internal only skill I'd probably still use the same foundation.

Hmm maybe it always feels "different" to every generation going through automation of something previously un-automated, but don't you feel that to some extent this is different.

Humans have been the dominant species because of our ability to think, not our ability to get from A to B in any particular way/cook food. If we abstract ~the bulk of our thinking to LLMs, don't we lose something as a species which makes us more vulnerable... Or is it always the case during an automation?

Because I am LLM Dependent I got distracted reading the article and made my LLM summarise it to me in a few slides. For those in a rush/interested in LLM summary slop:

https://smalldocs.org/s/scB6O8KiZEmLWgK855A1vt#k=yiw8FB_G5Ld...

I do agree that we are on a pathway to losing much. If you focusing on maximising LLM-powered productivity you basically lose the learning that used to naturally happen when doing actual work. At the company I work for the focus is on productivity, and it's easy in many ways (less mental workload and I receive more praise) to go with it instead of fight it.

I enjoyed this article... random cat pictures (esp. thumbs up cat picture) kept me going.

I do think on device AI will be a BIG thing... and basically change how we interact with our computers entirely - you will just ask to get things done. Low latency, free to use as much as you want. I think this future is hard to avoid/this will be a chunk of the future.

(I made a similar comment in a different post by Simon Willison yesterday about GPT Work)

I am unsure the direction that AI assisted document creation work will go in. I am opinionated and building something in the space. In this reply I discuss my view + what I have been building.

My opinion is: We now have AI, this new amazing flexible tool. We can throw it at any problem. But in terms of document creation we have a choice between old fashioned siloed document formats (.pptx, .pdf, .xlsx, .docx, etc.) and a freshly rendered - slightly randomly styled - html site with some basic javascript within it. I think these combinations are odd and do not maximise the powers of AI.

I've been building something called SmallDocs [0][1] which is a new take on work documents. It's all Markdown, but it always renders (100% privately) on the SmallDocs site. This means the SmallDocs renderer can convert recognised Markdown combinations into powerful traditional document formats [2]. For example, we convert a ```cells block into an in-line spreadsheet you can export to Excel.

This gives an agent the ability to express itself with a great deal of depth and flexibility. For example, your agent can produce a single data analysis "SmallDoc" with text, charts, spreadsheets, etc. [3] I find that combining formats quickens the speed I can understand a topic.

SmallDocs also have standardised and pretty boring styling. This means you spend less time understanding the layout of a flashy HTML page and more time on the content.

I also believe a pre-determined rendering engine has a lot of advantages over a freslhy built HTML artifact. It allows you to build in lot's of useful defaults that the agent doesn't have to code from scratch. These can be small things, for example, every time your agent renders a Markdown table in SmallDocs you can instantly copy it as a CSV or a PNG. And these can be bigger things, such as a rich slides DSL with solid styling defaults, which saves your agents a lot of tokens and results in consistently attractive HTML slides [4]. (Slides can also be automatically exported to .pdf and .pptx.)

I've been at it for a few months, and am still working on the project. I am looking for feedback and open source collaborators. Thanks for reading!

[0] https://smalldocs.org

[1] https://github.com/espressoplease/smalldocs

[2] https://smalldocs.org/blogs/what-is-a-smalldoc

[3] https://smalldocs.org/s/46fWiAyF77T9c7tqYwAQqp#k=j78ZEOOrgo8...

[4] https://smalldocs.org/s/QrmeWMsULGS871nXQYOacB#k=ubOHH9QtTWp...

(And something totally different, your agent can also use SmallDocs to walk you through code files: https://smalldocs.org/s/JsI_-tWSH0UpSd_YXxVLPm#k=G5jLCYA4JUX...)

I am unsure the direction that AI assisted work will go in. I am opinionated and building something in the space. In this reply I discuss my view + what I have been building.

My opinion is: We now have AI, this new amazing flexible tool. We can throw it at any problem. But in terms of document creation we have a choice between old fashioned siloed document formats (.pptx, .pdf, .xlsx, .docx, etc.) and a freshly rendered - slightly randomly styled - html site with some basic javascript within it. I think these combinations are odd and do not maximise the powers of AI.

I've been building something called SmallDocs [0][1] which is a new take on work documents. It's all Markdown, but it always renders (100% privately) on the SmallDocs site. This means the SmallDocs renderer can convert recognised Markdown combinations into powerful traditional document formats [2]. For example, we convert a ```cells block into an in-line spreadsheet you can export to Excel.

This gives an agent the ability to express itself with a great deal of depth and flexibility. For example, your agent can produce a single data analysis "SmallDoc" with text, charts, spreadsheets, etc. [3] I find that combining formats quickens the speed I can understand a topic.

I also believe a pre-determined rendering engine has a lot of advantages over a freslhy built HTML artifact. It allows you to build in lot's of useful defaults that the agent doesn't have to code from scratch. These can be small things, for example, every time your agent renders a Markdown table in SmallDocs you can instantly copy it as a CSV or a PNG. And these can be bigger things, such as a rich slides DSL with solid styling defaults, which saves your agents a lot of tokens and results in consistently attractive HTML slides [4]. (Slides can also be automatically exported to .pdf and .pptx.)

I've been at it for a few months, and am still working on the project. I am looking for feedback and open source collaborators. Thanks for reading!

[0] https://smalldocs.org

[1] https://github.com/espressoplease/smalldocs

[2] https://smalldocs.org/blogs/what-is-a-smalldoc

[3] https://smalldocs.org/s/46fWiAyF77T9c7tqYwAQqp#k=j78ZEOOrgo8...

[4] https://smalldocs.org/s/QrmeWMsULGS871nXQYOacB#k=ubOHH9QtTWp...

(And something totally different, your agent can also use SmallDocs to walk you through code files: https://smalldocs.org/s/JsI_-tWSH0UpSd_YXxVLPm#k=G5jLCYA4JUX...)

Agreed

You might find my open source project SmallDocs useful. It lets Claude make presentations in Markdown which can be exported to PPT. You can define styles and make them defaults. (You can also share the native SmallDoc presentations by just sending a url.)

Homepage: https://smalldocs.org/

Example presentations:

https://smalldocs.org/s/mTBw58ut_eTAIlGkYePw6-#k=4MotlKf-sLc...

https://smalldocs.org/s/cGRzzCv7LA9Kd8BxZ4bBAP#k=B7_NNXPbpLw...

Code: https://github.com/espressoplease/smalldocs

How has traction been so far? And are there any valuable cross team/agent use cases you actually have experienced or are they theoretical?

I run https://smalldocs.org, which focuses on making Markdown a highly expressive space for agents (think slides, spreadsheets, charts, mermaid diagrams, etc.). I have built a local library which lives in the browser but reflects your local file system - explainer here: https://smalldocs.org/connect?return=%2Flibrary, demo here: https://smalldocs.org/library?demo=1. Which is pretty useful when you want to find local Markdown files.

However I am just about to introduce very similar functionality to yours - a cloud + agent access. Right now I don't know if this will actually be useful, but I think it will likely be something I can charge for. I have a few hundred developers that use SmallDocs on a weekly basis (https://smalldocs.org/analytics), but right now it's just a free local-first tool.

Code available here: https://github.com/espressoplease/smalldocs

I think it's pretty interesting - I can see why companies would want to go for this instead of build everything themselves...

Curious about how your customers are responding to pricing. 20 agents for $499/month without API costs included feels steep... but perhaps within the range of "worth it if we don't have to think about this".

For traction...