# [[AI Weekly 2026-05-05]] <iframe width="560" height="315" src="https://www.youtube.com/embed/5ZfAILBR74g" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe> plugins:: "" Related:: "" ## Pre-production ### Thumbnail ### Title 1. ### Hook ### Structure ### Outro ### Related videos and resources ### Pinned comment ``` ``` ### Timestamps 00:00 Intro ## Post-production - [ ] Create captions using Rev, then edit them. - [ ] Take a thumbnail photo and create one for YouTube (1280 x 720). - [ ] Add description. - [ ] Include the title and any keywords in the first few sentences, but in a natural way. - [ ] Add related videos and resources as links. - [ ] Add end screen. - [ ] Add cards if necessary. - [ ] Add chapters/timestamps for YouTube if longer than 5 minutes. - [ ] Add mid-roll ads/check monetization settings. - [ ] Add pinned comment. - [ ] Post video preview on Patreon. - [ ] Schedule on YouTube. - [ ] Add video to relevant playlist(s), or create a playlist if necessary. - [ ] Schedule on Mastodon. - [ ] Create a blog post on [site](https://nicolevanderhoeven.com). - [ ] Add to [video-database](obsidian://open?vault=obsidian-playbook&file=video-database%2FVideo%20database). - [ ] Add to other notes in my vault as appropriate. ## Transcript Speaker 1 (00:00): Hello, welcome to the weekly. So Sven, you mentioned potentially demoing something. What do you think? Yeah, you can go to the Sigil slide for that if you want. It's related to that. I think in general, we are at a point where we are doing both UX and also already ops improvements. We see some Umers yesterday and such and we are working on that and I'm mostly working on the feature called iCoreGuards at the moment. I shared a video about that last week, but I want to take the opportunity and as demo as well again. And it is a bit broken. I have some agents fixing it at the moment, but something is working that I can do more. So we have these evaluators that were basically just running asynchronous whenever a condition was true, mostly for user messages, like new prompts in the assistant and stuff. (01:02): And I thought, why not make this something synchronous and blocking to have guards in place for prompt injections or whatever? My small demo is not like the prompt injection because that's broken. But I have an evaluator that's just testing if a user message contains a Datadog is cool. And I have additionally tool this evaluator, I have a guard setup which basically just uses this evaluator. So the Guard as if can contain multiple evaluators and can match more criteria to match different evaluators to different agents, for example. So I could just add onto Grafana assistant here, but I think I can't save because that's broken. Yes. Perfect. Hopefully I didn't break it now. However, to see it in action, it's in Abit and F. So you can test, you can just normally use the assistant as normal. If your prompts don't contain Datadog is cool, everything is working, but if they do, this will be blocked. (02:16): Hopefully this will be blocked by this card that's in place in Sigil. And I think that's a great use case to use Sigil as kind of a security biowall-ish framework for these things like prompt injections. Something I'm working on at the moment is not just denying stuff but actually transforming messages, i.e. A PII reduction or secret redactions. So if you paste a secret key in your user messages in the assistant, the transform could then automatically redact it before the secret is leaked to the LLM. Yeah, I think that's pretty interesting. Yeah, very good. Yeah, really powerful feature. Might be nice to get a video of that that doesn't mention Datadog. Oh yeah, I can do New Relic. Yeah. We want to do all of them. Yes. Thank you very much, everyone. Great chat today. See you soon.