I’m out and about so haven’t tried this yet, but sounds like your context is an embedded diagram. For free diagrams (.mmd files) I find AI creation and editing to be highly effective.
I recognise your point entirely and I will have to take it onboard for the future of my business. Please do, however, make good use of the website which is completely free, private, needs no signup, and gives you export to SVG and PNG. Also let me know if there’s anything else that would make it more usable. Many thanks.
Right. Mermaid syntax for basic flowcharting is straightforward. But it's even easier to get your AI to produce Mermaid, which they do very readily. May I suggest you try the Line9 CLI, which lets your AI show you the diagram visually as soon as it's created.
Thanks, very kind offer. When people paid real money for this, were they paying to make diagrams look good, or for convenient source-editing and rendering in the page? Line9 is built around the first and doesn't have an ideal solution for the second yet.
Ok, I've investigated Confluence some more. Publishing a diagram from Line9 Pro gives you a stable URL that serves a pre-rendered SVG and which updates when you modify the Mermaid. This can easily be displayed in Confluence by inserting an image from a link. Problems I found: 1) you'd need to published with public visibility 2) Confluence appears to cache the image, so updates don't appear in the page. Doing this properly in Confluence will need an app, which I don't have yet.
I'd separate Mermaid the language from the tooling around it. The language has real strengths, like Markdown for diagrams, and not least in being ubiquitous so that both humans and AIs can use it easily. What I found poor was the tooling side and that's what prompted me to create a new layout engine. The ecosystem can be improved.
Oh dear, sorry that didn't work out well for you. If you’re pasting flowcharts that don’t work then I really want to try to address them. Can you say what happened? Nothing rendered at all, or something you couldn’t use? If you’re able to send them to me directly they’re worth a lot as real diagrams to improve Line9 from.
Thanks. I haven't yet got to looking into Confluence. I used to use a Confluence plugin for Lucid charts and I guess something like that for Line9 would be useful if it gets much traction. I don't know whether the embedding options in the Pro account would display well in Confluence.
On pricing, I'm testing numbers at the moment so your feedback is valuable. Anyway, please do paste in your bad layouts - I'd love to hear what happens.
Mermaid does indeed provide alternative layout engines. Please do try ELK before Line9. My experience is 1) hand-tuning the layout of Mermaid, having to try alternative engines or parameters, is not what I want. I just want a graph that I can use straight away. 2) ELK more often than not does give a clearer flowchart that I might be satisfied with for personal use but I still wasn't happy putting it in a presentation in front of others.
Neither, really — it's closer to hand-rolled local search.
My first three phases are based on Sugiyama to produce a seed layout (which is broadly what Dagre does). Then I've added my new phases driven by a penalty score that grades candidate layouts using fourteen factors that I devised from my experience laying out graphs manually. The refinement phase is where the bulk of my work is. It applies targeted move types iteratively within a budget, trying to lower the penalty score.
Finally, the result goes through my own embedding and rendering with the theme system on top.
Recipe websites are about the worst in packing adverts and junk around their recipes, so this is brilliant. But I think you're having difficulty in listing available recipes from the search, some sites not playing ball. But you don't give a list of options to pick from, which might be helpful.
I like this. I have manually operated blinds and the automated kits were expensive, but not more than about £60 each from memory. Anyway I've occasionally thought of doing something like this. I did something simple for my bedroom curtains 45 years ago.
Yours seems well integrated but does seem quite noisy.
There are no barriers to creating software. There are no barriers to copying software. But there are still moats around some software businesses: non-technical which AI does not directly impact. Nonetheless, advertising is not the only way to win and there's no need to be nostalgic. We will adapt to the new era.
This is a well-presented website and looks like a good set of features. I always thought of creating one of these for myself, but I don't think it's a very easy market to make money in. Good luck with it.
What would a new social media platform offer that existing ones don't?
Important to me: my time and the value I get from spending it. To move off X to a new platform the signal-to-noise ratio would need to be better.
It's increasingly common to have fascinating discussions with LLMs, and they do come out with some good insight. (It would be nice if they could vary the ubiquitous stock terminology, such as 'heavy lifting'.)
Fable and its like are truly knowledgeable and clever. However, they are still blind to consciously stepping back and thinking outside of their working context. I may be wrong, but I don't see any moves anywhere that take us into the singularity. All models can still be outwitted trivially.
My goto approach would be to ask an AI to investigate. They can get a surprisingly long way in troubleshooting networking, doing web research, etc. and will even decompile Claude Code to work out what the issue is. It's easy to underestimate what AI agents can do. If you can't use Claude to do this then pick OpenAI or Grok, etc.
No, certainly not necessary in all cases. It's a marketing approach: companies want to find an edge and a way to indicate they're progressing and leading, so they find a way to attach AI to whatever they do. Only a fraction of these are worthwhile, and in most cases users would be better off by using their own AI directly.
I no longer write code. In fact I develop in languages I'm not very familiar with. I use Claude Code, Grok Build and Pi as AI coding agents. I use Linear for issue / task management. I use Github for CI, building on multiple platforms and CD. I use Neovim when I want to read code, and Mermaid when I want my code architecture to be explained.
I'd suggest MCP agent mail https://github.com/Dicklesworthstone/mcp_agent_mail
I haven't tried it on a raspberry pi at all, but I have used it on my PC and Mac and it enables agents to message each other. It's a lot of fun to drop in on their conversations and observe how they address each other and co-operate!
Whether or not your analysis is correct (I'd say not), the root problem is Chinese manufacturing dominance and unfair competitive advantage when it comes to EVs. It saddens me to say it, but the legacy car companies are unable to pivot and are likely doomed.
I love the Honda E and it's not mentioned in the article for some reason. However it must certainly have been a costly flop; they are so rare on the roads in the UK,
Often suggested by optimistic podcast guests these days: the as-yet-unknown new careers that will replace the familiar old ones and thus give employment in the AI era. I think your story is more a commentary on the current AI goldrush than an insight into future careers.
I've enjoyed SW engineering always. It started as a hobby and I'm fortunate to do what I enjoy as a profession. Do I still enjoy it? I find the rise of AI mesmerising. It's delightful to me to use AI and build with it in the partner role. Doesn't mean I don't have misgivings or doubts about where we end up, but I've never been so enthralled with tech.
I find that communicating with AIs is enhanced by using Mermaid diagrams, which the AIs love to write and can convey quite a lot of complexity in an easily consumable format. It's really helpful to have a nice diagram to study rather than a wall of text. (AIs really love you to give you a wall of text ;) )
Mermaid is useful because it's text that can be checked in and versioned as usual.