OpenCode + NEAR AI Cloud: Private Inference for Your Coding Agent
Nothing is loaded from YouTube until you press play, and then without tracking cookies.
About this video
Every AI coding agent sends your code somewhere. This one sends it into a TEE.
OpenCode is an open source coding agent that lets you bring your own model provider. Here I install it with curl, connect NEAR AI Cloud, pick a model that runs inside a trusted execution environment, and hand it a build task. The whole thing takes under three minutes.
One thing I call out in the video: not every model in that list carries the same guarantee. Some options are anonymized rather than TEE-protected, which strips identity but does not give you hardware-enforced isolation. Check which model you've selected before you start prompting.
CHAPTERS 0:00 Why privacy is on my mind this week 0:17 What OpenCode is 0:32 Installing from the command line 0:57 Launching OpenCode 1:12 Connecting NEAR AI Cloud as a provider 1:32 TEE-protected vs anonymized models 1:54 First build request on private inference 2:24 What this protects, and wrap-up
WHAT PRIVACY MEANS HERE Private inference means the model runs inside a trusted execution environment, so the provider cannot read your prompts or outputs during processing. It does not encrypt your local files, and it does not apply to the anonymized-only models. Anything you paste into the agent still leaves your machine.
LINKS OpenCode: opencode.ai NEAR AI Cloud: near.ai How TEEs and attestation work (previous video): • WTF is a TEE? The Room That Locks From the...
Leave a comment if there's something you want covered next.
Transcript
551 wordsCaptions as YouTube publishes them, usually machine-generated, so product and people names can be misheard.
- 0:00
Hello. I figured today was a good day with all the news about privacy and these AI companies and what's being trained on and how different things are being used that we just kind of go back over how you can use private inference from Near in different ways. And one of the ways that we're going to do that is actually using Open Code, which is an open source AI coding agent, uh, similar to like a cloud code, things like that
- 0:25
that allow you to use any model you want. And one of the ones that we can actually plug in here is Near AI's private inference. So let me show you how we do that. There are two ways I can do this. I can install it here from the command line or I can just download a binary and get started. It's up to you of how you want to go about doing that.
- 0:45
So I'm going to get started and I'm going to do it on the command line. I'm going to just use standard curl. Just get it started. Refresh our ZS. Okay. And now I should be able to just run open code. And here we are. So now I can start to pick out uh what I want to do. And I'm just I can obviously start asking questions, things like that. But let's actually pick a different provider. So we're going to run connect. There's tons of ones here. So we're going to just search for near near a cloud.
- 1:24
Now I'll put a key in. Okay. Okay, I have a key which I've obscured. Once I put that key in, then it gives me some different choices of different ones I can use within your AI. So, you can see there's all these different choices here. Uh I can even use different versions of cloud which are anonymized which are not fully uh protected by the TE. So, the private inference is more anonymous inference
- 1:47
than anything else. So, I'm going to actually use one that I know is going to give me that full protection. And you can see there's tons of choices here to choose from, [snorts] but I'm going to jump and get started with the latest Quinn.
- 2:03
And now I'm off and going. And you can see here what it is. And I can now ask it to build whatever I'm looking to build, whether that might be something like car collection. [snorts] And you're off and running. And now I know it's using private inference.
- 2:24
Anything I ask it is going to stay private. It's going to use TEES. I have a whole video that you can I will link back to that you can go see and how this all works. Leave a comment or suggestion if you have anything you'd like to see uh in future videos about this. But it's really easy to get started with Near AI.
- 2:42
I recommend that you check this out and try this on your own today and see how this goes for you. I think he'll be pleasantly surprised about the performance. And again, this privacy and security is a huge piece of all of this.
- 2:53
And now is the time to start really thinking about privacy