Mistral Patent for “Code implemented tool calls” • Hacker News Board

Re: Mistral Patent for “Code implemented tool calls”

#171

Post by pbhjpbhj » Mon, Aug 10, 2026, 10:03 PM UTC

speedgoose wrote: post #155

CodesInChaos wrote: post #123

Earlier quoted context omitted.

The patent office applies a much lower "non-obviousness" threshold than how a technical person would interpret that word.

It’s old memories from the engineering school in Europe but I remember that a patent office is about collecting fees and documents, and won’t do any actual checks on the patents validities. Whether a patent is valid is tested in court.

Some countries don't examine patents for validity, some don't search prior art but leave it to the applicant. But plenty of countries do check for validity - including USA.

Re: Mistral Patent for “Code implemented tool calls”

#172

Post by amelius » Mon, Aug 10, 2026, 10:04 PM UTC

Let's find an LLM that was trained before the patent was filed. If it can come up with the idea, then it is trivial and the patent can be discarded.

Re: Mistral Patent for “Code implemented tool calls”

#173

Post by dalemhurley » Mon, Aug 10, 2026, 10:07 PM UTC

I don't see anything unique nor novel to what everyone has been doing since GPT-3.5, before OpenAI introduced function calling, I was building my own JSON and tool calling.

Re: Mistral Patent for “Code implemented tool calls”

#174

Post by pbhjpbhj » Mon, Aug 10, 2026, 10:09 PM UTC

c7b wrote: post #12

dummydummy1234 wrote: post #4

So anyone have prior art we can point to? Because no way is a rpc call novel.

Seems that this is a patent application from March, so a challenge should still be possible. But it would have to come from a named entity afaik (not a lawyer).

In applications before the USPTO anyone can submit preissuance observations, afair, https://www.uspto.gov/web/offices/pac/mpep/s1134.html

The submitter needs to be identified, so get a friend/representative to submit if you want to maintain anonymity.

Re: Mistral Patent for “Code implemented tool calls”

#175

Post by rippeltippel » Mon, Aug 10, 2026, 10:11 PM UTC

dummydummy1234 wrote: post #4

So anyone have prior art we can point to? Because no way is a rpc call novel.

Isn't that patent similar to CodeAct [1], originally described in this paper [2]?

[1] https://learn.microsoft.com/en-us/agent-framework/agents/cod...

[2] https://arxiv.org/abs/2402.01030

Re: Mistral Patent for “Code implemented tool calls”

#176

Post by constantlm » Mon, Aug 10, 2026, 10:11 PM UTC

What? How? With no mentions of priors. This idea has been around for a long time - at least since 2024 (Xingyao Wang et al.)

Re: Mistral Patent for “Code implemented tool calls”

#177

Post by pbhjpbhj » Mon, Aug 10, 2026, 10:19 PM UTC

Dotnaught wrote: post #160

What are the chances Mistral used AI for its patented technology, thereby potentially invaliding the claim?

You mean to create the invention? I think most jurisdictions have settled on 'use of ML tools is fine, you just need to name the user of the tool as inventor'. Invalidating a patent on the basis that the named inventor used ML tools doesn't seem like it will work.

Do you know of a jurisdiction that disallows inventions generated with "AI"?

Re: Mistral Patent for “Code implemented tool calls”

#178

Post by pbhjpbhj » Mon, Aug 10, 2026, 10:20 PM UTC

maeln wrote: post #9

How is the US patent registry easy to search ? I would be curious to see the patent registration of OpenAI, Anthropic, Google, etc. One way or another, I do fully expect every of those companies to try to fill these kinds of "almost for sure invalid and overturnable" patent as a way to defend themselves from every other actor (the classic "Oh I infringed on your patent ? Well here are some of MY patent you are infrin…

EPO has Espacenet, which hasn't been mentioned yet. I find it the best of the free patent search tools, ymmv.

Re: Mistral Patent for “Code implemented tool calls”

#179

Post by throw-the-towel » Mon, Aug 10, 2026, 10:22 PM UTC

So the European poster child AI company is now doing something that smells of patent trolling? Gee, Europe really isn't helping its image of a society obsessed with lawfare.

Re: Mistral Patent for “Code implemented tool calls”

#180

Post by pbhjpbhj » Mon, Aug 10, 2026, 10:27 PM UTC

OkayPhysicist wrote: post #106

Topfi wrote: post #33

Still pending, or am I mistaken? Can anyone with knowledge on the US process make an estimate how likely this is to go through? I know of Uniloc and their "efforts", is there a possibility that this could take a similarly damaging path?

In the US, the patent office almost always grants patents as long as the correct process is followed. They don't really concern themselves with whether a patent could be successfully challenged, they just care about whether the the application is shaped like a patent. They leave it to the courts to invalidate all the bad patents they approve. It's pretty terrible: the very least they could do is ask a random dude in…

This is not at all true, [non-final] rejections are very common. Not only for reasons relating to the the form of an application but also due to lack of novelty or inventiveness (35 USC 102, 103).