Mistral Agents API • Hacker News Board

Re: Mistral Agents API

#21

Post by moralestapia » Tue, May 27, 2025, 6:54 PM UTC

orliesaurus wrote: post #2

Whoever made those embedded videos, here some feedback if you want it take it, it's free: 1) It's really hard to follow some of the videos since you're just copy pasting the prompts fr your agents into the chat because the output generation comes out and hides the prompts. Instead put the prompt text as an overlay/subtitle-like so we know what you're doing 2) The clicking sound of you copy pasting and typing is not A…

I came here to see if anyone else noticed.

Very sloppy job, imo.

It costs next to nothing to come up with a little story and have someone on Fiverr narrate it (or an AI, after all that's what they sell).

Re: Mistral Agents API

#22

Post by ed » Tue, May 27, 2025, 8:29 PM UTC

potatolicious wrote: post #12

bbor wrote: post #4

Ok I’m behind the times in terms of MCP implementation, so would appreciate a check: the appeal of this feature is that you can pass off the “when to call which MCP endpoint and with what” logic to Mistral, rather than implementing it yourself? If so I’m not sure I completely understand why I’d want a model-specific, remote solution for this rather than a single local library, since theoretically this logic should be…

It seems like the main pitch here is auto-inclusion and auto-exclusion of various tools via an orchestration agent (which may or may not be the main model itself? Unclear from their post) Mostly this seems like an end-run around tool calling scalability limits. Model performance degrades heavily if the field of possible tools gets too large, so you insert a component into the system that figures out what tools should…

> auto-inclusion and auto-exclusion of various tools via an orchestration agent

Where do you see that? That would be neat, I'm under the impression orchestration is manual though – you define an agent and give it the ability to hand off tasks to sub-agents.

Re: Mistral Agents API

#23

Post by eigenspace » Tue, May 27, 2025, 10:12 PM UTC

greenavocado wrote: post #11

htrp wrote: post #5

is mistral a model company, an agent company, or a enterprise software company now?

Mistral is trying to be everything at once and it shows. To make ends meet they pivoted to selling enterprise software through Le Chat and cozying up to Microsoft. Now they're throwing around terms like "agentic AI" to stay trendy, even as competitors like DeepSeek outperform them in key areas. Their identity crisis is obvious. Are they a model company? A software vendor? A research lab? At this point, they seem more…

Their strategy doesn't make sense to you because you're looking for a technical feature that differentiates them. But technical features aren't their key differentiator, geography is their key differentiator. They'll get a lot of contracts in Europe simply because they're European. Everyone is keenly aware of how dependant European tech stacks are on increasingly unfriendly foreign powers.

If there's a local European option that does most of what an American or Chinese company does, that's simply a safer choice.

From this point of view, them trying to do everything at once makes a lot of sense. They don't actually need to be the absolute best or even the cheapest at any one thing. They need to just exist in Europe, be stable, and offer good services that people want. Casting a wide net is a better strategy for them.

Re: Mistral Agents API

#24

Post by potatolicious » Wed, May 28, 2025, 5:22 PM UTC

ed wrote: post #22

potatolicious wrote: post #12

Earlier quoted context omitted.

It seems like the main pitch here is auto-inclusion and auto-exclusion of various tools via an orchestration agent (which may or may not be the main model itself? Unclear from their post) Mostly this seems like an end-run around tool calling scalability limits. Model performance degrades heavily if the field of possible tools gets too large, so you insert a component into the system that figures out what tools should…

> auto-inclusion and auto-exclusion of various tools via an orchestration agent Where do you see that? That would be neat, I'm under the impression orchestration is manual though – you define an agent and give it the ability to hand off tasks to sub-agents.

Sorry, maybe I could've phrased it better: it basically forces the devs to divide their tools into buckets of fewer tools manually. (The Travel Agent has N tools, the Research Agent has M tools, etc. all specified by the dev)

The pitch is that if you do this bucketization, the overall orchestrator can intelligently pick the bucket to use, but the idea is that at any moment the LLM is only exposed to a limited set of tools.

As opposed to the more pie-in-the-sky idea that given N tools (where N is very very large) the LLM can still accurately tool-select without any developer intervention. This seems pretty far off at this point.