lazharichir

> Put yourself in the user's shoes. On what basis should one pick a model for a given task?

Create a realistic eval suite for your use case using your actual data that you manually label, with easier, more complex, rarer samples, and run each model you are considering against it and decide based on the usual confusion matrix, cost, latency, or whatever metrics matter to you. There's really no other way to know – and you might be surprised that cheaper models can often perform your task(s) relatively well at a fraction of the cost.

I am using A2A at work. It's a bit like the "microservices architecture" for agents... Allows you or teams to develop agents independently and have them interact as and when they need to. No major hurdles so far.

What can you NOT run on this, it's not very clear? Is it like MicroVMs on steroids where you can run more binaries than the strict minimum?

Definitely the feature but I'm sure Gemini is seconds away (figuratively) from invading Chrome and if it has an agent mode itself, it will eat everybody's lunch in the browser space.

Just excited about Agent Mode, I hope it delivers.

Quite happy when I got to play with it – but I do prefer Inngest as the code is hosted where our actual app is hosted; not on their server. So it's just easier (but it's a tradeoff).

Good tool, good tooling, congrats to the team!

From my (limited) understanding you cannot really switch these off inadvertently as they require a couple of actions in order to be switched off. So it would mean one of the pilots switched these off (and they were a few seconds later switched on again but it was too late).

But there was audio, too, and one pilot asked the other "why did you switch these off" and the second one said "I didn't".

Was there are third one in the jump seat?

You can also do gemini flash lite for a subset and then batch the rest with flash or pro

That's basically how I use it though I cannot wait for Gemini to be a click away in my browser with the current screen/page(s)/tab(s) embedded so I can ask it stuff about the current long article/documentation page. We're becoming very, very lazy.

to be fair, i'd take tame over horrendous and unparseable screen any day.

Like taking on a gazillion of contract work?

I assume they mean that you send an Array<Message> to the model instead of a single system and user message.

Good small fun project