You can set this up yourself with API keys to the corresponding providers and creating an Agent Group in https://github.com/lobehub/lobehub. Agent groups allow you to easily create a room of agents and have them discuss any of your topics. Easily make agents with types and skills, it even assists in drafting starting prompts and even team members depending what your query (and selected model) is.

You can self-host as well, but not via desktop app. Sever setup required.

Be careful of your token context, you can easily rack up costs if you leave Opus selected as the model and get lost in some rabbit hole of results.

Enjoy enjoy!

Nothing stops that from happening. Just needs to be trained in that DSL. Though at that point it returns to it's original form as a better autocomplete/IntelliSense :).

That will likely happen in the specialized fields. We can already see tools like Figma, Mira, and others that generate functional-ish frontend components in full typescript and corresponding styles (that are also selectable and configurable in the interface). Though, not quite as free, since they do load their base framework and components to ensure consistency and sanity / error-checking, etc., but even then it is in fact generating you useable, modifiable components that you can engage with in precision in your normal DSL.

For video, this likely exists, or is being worked on as we speak. All specialized domain tools will go towards this model to allow those domain experts to use the tools with the precision they expect AND the agentic gains we already take for granted.

I know the developer who worked on it took pride in the outcome. Hopefully they added some additional characters to keep it fresh.

To be fair, it was really cool. It was also a tech demo with no real practical application.

It was really cool, unlike my phone after doing it for 5 minutes!

There were social games that used it as a feature, and it was fun when it worked, but it had to be disabled soon as it drained the battery so fast.

Despite their confusion as to whether they want to really support their free and open source versions (without some absurd user count cliff), Mattermost (https://mattermost.com) is quite excellent and IMO is better than Slack. For example, editor leans towards native Markdown so things like syntax highlighting with backticks work as you expect.

Their recent update removed the paywall from SSO (and unfortunately the Gitlab SSO workaround) for social logins up to 100 seats, afterwards there's an absurd per seat cost similar to its non-open source brethren. One day if needed, I plan to drop-in an SSO middleman allowing anyone to leverage their own SSO layer (which will map to the login form with username/password) to avoid the SSO limits altogether. Though good enough for my needs, and likely yours too. Especially if you're open to paying for their seats.

Unfortunately there are many companies that actually rely on SMS confirmation codes in real-time, which include reading it back to them.

A legitimate and generally well liked company, and its real helpful service representative used this method to verify my identify before they could finish their support effort.

I got this interesting pair of messages from Schwab recently - not sure if any other companies do this

On login:

Schwab Watch out for scams. DON'T share this security code with anyone, EVEN IF THEY CLAIM to be from Schwab. Your code for online login is XXXXXX

And then on a later phone call with an agent:

Schwab: XXXXXX is your Schwab security code to confirm your identity with the agent.

This is a nice touch, though I'm not sure how much it would help in a real scam situation for say, my grandma.

yeah someone that gets paid a lot needs to talk to someone whos pay depends on implementing that IT consultants directives.

relaying security codes by voice is how the bad guys do it, dont train your users to think its normal.

its probably not a bright idea to have your phones camera pointed at your screen while 2FA-ing or password resetting, or else someone will watch you login, and will see your codes, and use automation to authenticate with your digits faster than you can move a cursor and click.

Probably safe if you call them at a well-published number. If they call you, absolutely not.

How does this compare to the already established open source solutions such as Chatbox (https://github.com/chatboxai/chatbox), or Lobechat (https://github.com/lobehub/lobe-chat)?

Been using both, like Chatbox for how snappy it is, but is local only, vs Lobechat which allows you to setup centralized host to have shared host across clients but feels a bit clunkier.

One of the biggest differences I noticed off the bat is llms includes prompt caching which I'm not sure I've seen in any other self hosted UI options

I see Lobe and Chatbox both have prompt caching toggles, are you referring to something else?

I've mistakenly given Chatbox a new feature, sorry :). In LobeChat, after you select a particular model, it enables a mini-settings menu next to the model that lets you set caching, deep thinking, and thinking token consumption.

Ah that must be new since the last time I tried lobechat

Where do you see that? I can't seem to find it in the web or desktop apps for lobechat.

EDIT: I also don't see it in Chatbox

#freenode was generally the main IRC node I used with all the good dev rooms.

Seems to still be chugging along. You can even join directly via their web-client: https://freenode.net.

Personally I still use pidgin.im to connect to all the relevant #freenode goodness. Seems people forget it still works and is pretty great even all these years later :).

I forgot Freenode was still running, most projects left for https://libera.chat/

Due to https://en.wikipedia.org/wiki/Freenode#Ownership_change_and_...

Thanks so much for the update. Had no idea. Seems all the main channels moved over. Thanks!

Most went to libera, some went to OFTC[1], and a few moved to matrix as well.

[1] https://www.oftc.net/

Gift article: https://wapo.st/4od1but

Seems to be missing quite a bit of history. As many here mention, there was an entire ecosystem of tools to convert PSDs to HTML such as CSSHat, Engima64, etc., and its evolution into Avocode, Sketch, Zeplin, Invision Craft & Inspect, and other preview/prototyping/inspect/export tools.

Eventually all roads led to Figma somehow, which honestly I would've never expected. Still surprised Figma became Sketch before Sketch could become Figma.

Yes I left out a lot to keep it short and to the point. There has been many attempts at solving the handoff and none of them has been effective.

The solution is to get rid of the design handoff all together. Instead of designing user interfaces in vector drawing tools we need web design tools.

There's also open-source solutions such as rocket.chat, Mattermost, and likely a few more I haven't played with.

I wish Mattermost wasn't always trying to nudge you out of community version, but otherwise pretty solid, better than Slack IMO. Is unfortunate they require weird gitlab spoof bypass to use SSO in community version. Shameful it's not out of the box.

Many years ago Pidgin with multi-channel IRC was all I needed, but seems Slack killed that whole party, which brings us to the current situation :(.