February 2026

Speaker 0: Hello everyone, welcome to the changelog from Directus for February. I'm Beth and we have a really great show for you coming up. We've got a product update from James, an AI update from Brian, I'm around with a brand new community program to get involved with and we have a fresh one app ten minute episode with some brand new directors faces, so do stick around if you can. Whether you are joining us for the first time or you are a regular, hi hello, thank you for spending some time with us, and without further ado, let's kick it off with James and a product update for you.

Speaker 1: Hey. This is James from Directus, and I'm gonna take you through some of the highlights in the 11 dot 15 release. Now first up, our AI assistant is now in GA, and it is coming with some very tasty updates. We've added multi provider support with Google and OpenAI compatible providers. So now you can use our AI assistant with Olauma Mistral AI, extending on prior support for exclusively anthropic and OpenAI.

We've also made the AI assistant native across all of the interfaces in the director studio, meaning you can even use the AI assistant in the visual editor now. Now with new power comes new responsibility. And to use this feature, you will need to update the Director's visual editing library to v 1.2.0 plus on your website. We've also added a new deployments module inside Director's. This allows you to connect your Director's instance with Vercel to centrally manage deployments, monitor build status and control your front end projects all without leaving Directus.

We've added support for Vercel first, but Netlify and others are sure to come soon. Let's have a look at how it works. You'll find the Deployments module inside the settings and you'll need to enable this first. Once you've enabled that you're going to get the Deployments module in the sidebar of Directus. Let's take a look.

Let's have a go at configuring Vercel. What you'll need is your personal access token from Vercel, and here's one I have from earlier. Once you add that, you'll see the projects listed from your Vercel account. You can choose to bring one or more of these into directors. So let's bring in a couple.

Now you see the projects listed in the project listing. And if we click into one of these, we're able to hit deploy and start building our site from inside directors. So let's assume that we've made some content changes. Patch we've updated, you know, the price of an item on our website. And as a result, we need our site on Vercel to be rebuilt.

So I'd come into the deployments module after making that content change and I'd hit deploy. Now the other great thing here is we can monitor the deployment status as that is building. So in case that fails, I'm gonna be able to see the reasons for the site failing. And when it's successful, I'm actually gonna have the link up at the top right to be able to visit the end result. So we'll just give that a second while it's building.

Awesome. Now I can see the status is ready. And if I hit refresh, I'm gonna see this link up here which allows me to visit the end result. Now if I come back into here, I can see I can go back and I can see the deployment listing. Now one thing to call out is you're only gonna see deployments triggered from Directus inside the listing today.

So all of your deployments made from Vercel will not appear here at the moment. One last call out is at the moment, the deployments module is only accessible for admins. However, we do plan to add, RBAC support so you can open this up to more users in the next release. We've also brought collaborative editing into core. Now this was previously built as an extension, but we wanted to bring it into core to make some performance improvements, reduce the amount of setup, and make sure that this is a native capability.

Now under the hood, this runs on WebSocket connections for real time sync, so you do need to have this enabled on your project. It also plugs into the existing Director's permissions so users can only collaborate on records they have access to. Let's have a look at how to enable this in 11.15. You'll find this new setting in the project settings in your Directus instance. And once enabled, this will enable the real time sync.

So let's take a look at a record in our content space. We've got a collection of products, and let's assume that two people are working on the denim jacket. And I will just there we go. We can see that both myself, James, and Michael Matthews are now working inside, the Product Datastem. Now let's assume that somebody is working inside a specific field.

You'll see that that field lock comes into play, and that stops people overwriting each other's changes. So that's collaborative editing, and that's now available in the core. Now we've also made some improvements to how you can review view revisions inside the studio. So let's assume that we're updating the price of our denim jacket. And let's come back in to look at the revisions.

Now previously when you viewed a revision, we were always comparing the revision you open to the latest revision of that item. But we've made some changes to make this a little bit more intuitive. So if I update my latest revision, you'll see I have that revision on the right hand side and I'm always comparing it to the previous revision now so that we can see the granular change from a 100 dot 99 to 50.99 in this case. Now we've also maintained some flexibility for you to compare a previous revision. Let's choose a much older one and see how that currently compares to the latest revision.

Now this is advantageous in the case for restoration, in case you wanted to restore an older version but you want to understand exactly what it's going to update on the latest version of that item. So you just toggle this pill and you can switch between what you're comparing inside the revisions comparison model. So we've been through all of the main items inside the 11 dot 15 release. But as usual, you can go to the release notes on GitHub if you want to view every granular change, that got made inside 11 dot 15.

Speaker 2: Hey, guys. Brian here. And I'm gonna showcase some of the exciting new features we shipped to the AI assistant in v 11 dot 15. Alright. First and foremost, it's an absolute banger.

Not only can you now use the visual editor right alongside the form inside the live preview pane, but I can use the AI assistant right alongside the visual editor. I just click the magic AI button here. And now the AI assistant has this visual editor element into our context. And we'll just ask it to, let's punch up the copy a bit for this headline. Cool.

So it understands where that is at on the page, what's going on. And now we can see once we approve that tool call, that gets updated in line. All right, that's just the start, right? That is a heavy hitter in this case. You can also add context to the AI assistant now.

So, I can update pages that I'm not currently on. Let's say I've got this test page. Please update the slug and title for the test page to slash about. Right? And because we are injecting that into the context, it knows what page to update.

And if we just go back to our pages list, we could see now that is updated. Last but not least, you can also reuse the AI prompts, those templated prompts from the MCP. So if you go to your AI settings, make sure that the MCP is enabled and that you've got the AI prompts collection. And then you can reuse these prompts over and over again. So if there are dynamic variables, Directus will ask you to fill those out.

Or you know, if you don't have any variables, you could just

Speaker 3: do this and say, hey,

Speaker 2: tell me a nice dad joke. Alright. We'll see what it comes up with. Guy walks into a library, books about paranoia. She whispered, they're right behind you.

Perfect. Alright. Now, onto some of the other items that you guys asked for, and I wanted to make sure that we delivered here. So now you can also control which models for the three major providers, OpenAI, Anthropic, and Google, so you can lock those down. The other big rock out of this release is going to be the OpenAI compatible provider.

So as long as you have an endpoint that is OpenAI compatible, you can now go in here and set your base URL, add your API keys, you could set up your different models. Make sure you include the context limit, the output limit. You can also pass custom provider options if needed. And then you can use Ollama or, any other self hosted models, any other, OpenAI compatible models. Let's say, hey from Ollama.

And this might be just a little bit slow because my MacBook is absolutely screaming at me right now. We'll fast forward. Alright. So now you can see that we've got the text back. Your mileage is gonna vary with the self hosted models, but, you know, if you've got Azure OpenAI or some other open open AI compatible endpoint that you're using, this is a great solution for you.

And as always, keep the feedback coming. We love to iterate on these features, and we want to deliver real value instead of just the usual AI hype. That's it for me. Back to you, Beth.

Speaker 0: I'm here to talk through a brand new community program we're launching called Directus Builders. Builders is a community champion program for people who use Directus, want to share what they're building, and contribute to the community. Whether you're interested in sharing technical insights and receiving amplification from our social channels, joining a network of other directors users, or getting our support for your own community initiatives, this program is for you if you are using directors to build. By joining, you'll enter a private community with other experienced builders and our team. It's open to contributors, customers, partners, users, really anyone who uses Directus to build something useful.

You don't need to be building something huge, you just need to be building something real. If you're the person who likes helping others figure things out, sharing what you've been learning, or creating something cool, we want to hear from you. Applications to join the first cohort are now open. If you've got any questions or you have an idea that you think might work as part of this programme, we're all ears, we want to hear it. There's a couple of ways you can get in touch with us: submit via the application form, send an email to devreldirectus.

Io, or post on the community forum. All of those work and we can get talking from there. We're really excited about launching this program, we hope you will also share the excitement and want to join and we're really looking to shape the future of the program collaboratively with the builders into something that works for everyone. So if you do have ideas, thoughts, questions, please do let us

Speaker 3: know. Alright, viewers. Welcome to, yet another episode of 100 app, 100 oh, no. No. No.

No. One app in ten minutes. Right? We are doing the remix version today where we have ten minutes to build and plan plan and build an amazing app clone, crazy suggestion, and I have no idea what we're gonna do. So the rules.

Right? Ten minutes to plan and build. No more, no less. How we're gonna do that? We are going to use some, amazing tools that we have built into Directus.

And then, rule number two, the anti rule. Use whatever you've got at your disposal. Today, I've got two awesome dudes at my disposal, mister Alvaro and Mark from our team here at Directus. No strangers to the Vue community. Welcome to the show, gents.

Speaker 4: Thanks for having us, Bryant.

Speaker 5: Thank you very much for the nice intro. Happy, to be here.

Speaker 3: Yeah. Yeah. No. I'm super excited. Have you guys given any thought to what we're what we're gonna build?

Speaker 5: I think, Mark, you have some idea though.

Speaker 4: Yeah. So yesterday, we talked a little bit. I talked with Ava what we could build and, I don't know if if I showed it to you, Brian, but on my website I have a, instead of new year's solutions, I have new year's bingo cards. So you have five by five grid of stuff I want to do in the year. And if I get at least one in a row, so diagonal or horizontal or vertical, I already have bingo and it's a success.

So I don't have to do all of them. And if you go to mark.dev/bingo

Speaker 3: Okay. Let's check it out, guys.

Speaker 4: You can it's still since it's just well, now February, not a lot has happened there. But

Speaker 5: it's a it's a really nice way to actually do some of the New Year's resolution. I always get the press at the end of the year like I have done, like, a quarter of them.

Speaker 3: Yeah. I love it. Alright. So alright. This is neat, man.

I I miss Yep.

Speaker 4: And each of them can be either, like, you did it or you didn't do it or it can be progressive. Like, read six books and you are, like, one books, two books, three books in. And I think I also have, like, sub tasks. If we can make that work, like, if one one, let's say, one bingo item has a few sub items as well. Like, don't have an example now, but that would also be cool.

Speaker 3: Gotcha. Okay. New Year's resolution. Bingo card generator. Alright.

That's what we're doing. This is gonna be amazing. This should be fun. What color are you guys feeling? Purple, pink?

Speaker 5: I go I go purple. Blue. Or purple?

Speaker 3: Purple. There we go.

Speaker 4: Direct is purple. Nice.

Speaker 3: Direct is purple. Alright, guys. Alright. So I'm sure you've seen the show. We're gonna start the clock.

We got ten minutes to plan and build this thing. Let's do it. Alright. So the first thing I usually do here is cover requirements. Right?

So what are the requirements we need out of this? Right? We need to generate bingo cards. Like, what do you what were you calling those?

Speaker 4: Like, items probably or

Speaker 3: Okay.

Speaker 4: Goals. Yeah. Items.

Speaker 5: Yeah. Like a grid of of of items. Mhmm.

Speaker 3: Yeah. Alright. So we got some goals. Those are what kinda fields are you tracking on those? Just the name of the goal?

Speaker 4: Yeah. A name description and then the status.

Speaker 3: Status of the goal. Progress. Progress. Is it are you status and progress interchangeable?

Speaker 4: Yeah. I guess if you like the if the progress is under percent, the status

Speaker 3: Ah, okay. Yeah. Yeah. Yeah. Got it.

Okay. And then we've got if you got goals, you got what? Items underneath the goals? We want, like, subtasks, like, if it's

Speaker 4: You you can have subtasks. Let's see if there's one that has subtasks. I don't remember.

Speaker 3: Task. It's called test.

Speaker 4: Alright. So that the task would play into into progress as well, I guess.

Speaker 3: Into goal. And then the task completed increases progress. Cool. Alright. And task needs what?

Name? Description? No. Just name? Date date probably.

Speaker 4: Maybe the, the item can have a a completed ad. Yeah. They completed as well for the task for the, item on top. Yeah.

Speaker 3: Alright. And then we we wanna try to get a front end set up for this as well. Yeah. Alright. And we we need a front end to display the pingo cards.

Alright. This could be a stretch in seven minutes now. Let's see how we do. Alright. So what are we using today?

Right? We've got a blank directus project. We've got Claude code over here. Let's dive into it. Alright?

I'm going to I'm not sure what you guys have been coding with. I've been using Super Whisper. I dig it. Alright. How are you doing?

Alright, guys. We are building a New Year's resolution bingo card generator. I'm gonna copy and paste the data model that we want. You have access to a direct assistance. I want you to create our schema for that.

We're also going to be building a front end to display the bingo cards. Let me know what questions you have. Let's create a plan. Alright. So this is crunching the transcript for that right now.

Cool. There we go. I'll just, copy and paste this. Hopefully, we'll get some something good out of it. And we're gonna ask Claude Coad to plan.

Speaker 5: Alright.

Speaker 3: So now we've got the schema. So we've got the direct us MCP connected to this thing. And I I think you guys have had a chance to try this out already. Right? Yeah.

Speaker 4: I think Avro has. I haven't.

Speaker 5: Yeah. Play with it in the morning. It's gonna create the collections, the scheme is for you.

Speaker 3: Yeah. Alright. So it's got a fresh direction. No custom collections. Alright.

And I can zoom in just a little bit more so we could see this. What is the plan? And this is probably one of my favorite parts about this thing where it will prompt you for questions. Direct us flow, that's what we wanna do there. Vanilla JS.

Yeah. That's what we'll do. What do you guys think? Five by five grid? Four by four?

Speaker 4: We we can do also four by four so we don't have to come up with 25 things.

Speaker 3: Amazing. Right? We got five minutes left.

Speaker 5: You you can say to the MCP, hey, cloud, get, your twenty twenty six, bingo

Speaker 4: Oh, that was cool.

Speaker 3: Yeah. Yeah. Alright. Public read, that's fine. Anyone can view those.

Cool. Alright. And now, hopefully, this thing should have a plan.

Speaker 5: I wonder which resolutions Cloud Cove could have.

Speaker 3: I don't know. Let's see. We'll we'll spin that up in an in a new find out. Alright. Cool.

Right? Here's the direct to schema. There's our it's gonna create a flow. It's gonna create the front end. Sounds good.

Let's let's roll with it. Right? I don't know what we're actually doing other than just talking this through here, but, I'm curious to see just how this thing works. I've you know, of course, like, spent a ton of time testing and building the MCP, but I've not spent a ton of time using it with the the latest Opus four five model. Alright.

So it is checked the existing schema. Now we are it should start implementing. Yes. Please just start jamming on here. And if I refresh, now we should see some collections start to come in to the direct instance.

We should see some collections. Start to come into the direct instance. There we go. Okay. Alright.

Oh, nice. I was just worried that I did something wrong. So we got our goals. We got our tasks. Amazing.

Right? Now I could go in. We could potentially create some new ones if we need. One of the things that I like about this is it, like it seems like the anthropic models do a better job of, like, actually putting together a cohesive form than than, like, the OpenAI ones. So it's going through creating relations and fields.

Alright, guys. So in this other one, create, some New Year's resolutions for yourself, Claude. Alright. You guys have any more guidance for this thing?

Speaker 4: They should follow this the smart principle, probably.

Speaker 3: Follow the smart principle. What's the smart principle?

Speaker 4: Now you got me. So it's like measurable, achievable.

Speaker 3: I know what you're talking about now. Yeah. The smart goals.

Speaker 1: Yeah. Yeah. Yeah.

Speaker 3: And include the add them to the goals and tasks inside.

Speaker 4: For the for me, the most important one is always measurable. You have to be able to measure what you do. If not, you lose yourself.

Speaker 5: You lose yourself. That's so funny.

Speaker 3: That is very poetic. I love it, man. Alright. So it looks like okay. Yeah.

I was just making sure we've got the relationship created correctly there. Alright. It is going to so we got two claws going. We got two minutes here. Let's see.

I can see their goals and tasks.

Speaker 5: Alright. This is the next development, man. Right? This is

Speaker 3: the next development. Yeah. This thing is going to yeah. I need to enter YOLO mode so we can actually have this thing not stop to do these calls. But, behind the scenes, right, it is building this progress calculator flow.

And and flows are Sure. A a nice piece of functionality. It can be a little time consuming to set up, like, complex flows via the UI. So having direct us put these together, is, yeah, definitely time saving. Right?

That's probably, like, five, ten steps there. Yes. Create those items. Alright. Let's see what we've got.

Are we gonna get to the front end for this thing? I don't know if we are, man.

Speaker 4: I should've had Bryant.

Speaker 3: Should've had, Claude do that first. It's connecting the operations. Claude, you need to go faster, my friend. Alright. So what are the what are the goals that Claude set for itself?

This should be interesting.

Speaker 5: Put that description statement.

Speaker 4: I'll reduce

Speaker 3: average response latency by 20%. Achieve 95% task completion rate without clarification. What an interesting goal. Here's the the individual tasks. And, oh,

Speaker 5: and that was done there.

Speaker 3: The HTML.

Speaker 5: The front end.

Speaker 3: Now it's doing it. No. Let me open this test project up. Is it going to have enough time? Yes.

Proceed. New Year's resolution. Bingo. Oh, no. We ran out of time.

It's so close. MCP connection should have access. No need to set up. I think, you know, this was so close, guys. I'm just going to it's against the rules, but you know what?

We can make up our own rules here. I am just going to give access here to see and see if this will actually finish. Of course. There it is, man. The API permissions got us.

We could see the bingo card here. There's the individual task. Ten minutes, full working back end with permissions, so close to a working front end.

Speaker 4: It did pretty cool.

Speaker 3: This is this is very cool. Right?

Speaker 5: Even even with the subtask because that that wasn't an extra thing. Like, now it's the only iteration. Like, put the progress in the front end and

Speaker 3: Yeah. I'm very curious to see. Right? It's already got it looks like it maybe did it miss some of the flows? Right?

So the thing to take away here is obviously, like, you could build incredibly quickly with Directus and MCP, and this is not loading, probably because of my computer. Just hates running all these Docker containers locally. What is going on?

Speaker 5: How many do you have?

Speaker 3: There's probably, like, five or 10 running at the moment, like, different instances. And I'm sure if I, like, killed the camera, it would probably stop doing this. I don't I don't know what's going on here. Local host 8055. I at least want to end this episode on a high note and show something.

Come on. Alright. So we could see the flows. Did they yeah. It actually connected the flow.

So I'm just curious. Right. Just wanting to see. Right? Build a mastering five new programming frameworks.

Let's say we completed this right now. Does this flow actually work? And So it it

Speaker 5: it could increase the progress of the task of the goal.

Speaker 3: It should. It should. And, of course, doing a hard refresh here is not not a great idea.

Speaker 2: Alright. Well, gents,

Speaker 3: you know, I'm not sure whether to put a, like, a thumbs up stamp on this one. Thumbs down stamp. So we just do I think yeah. This was, we we got most of the functionality

Speaker 5: here. We

Speaker 3: just didn't get, the front end all the way there.

Speaker 4: Oh, Brian, you are lagging quite

Speaker 3: Of course, I did.

Speaker 5: Yeah.

Speaker 3: Because it does I think you

Speaker 4: you get a a thumbs up, Brian, because it we got a working thing at the end, and you had the the grid showing everything with the progress. So I think you get a thumbs up.

Speaker 3: Yeah. Alright, guys. My computer is struggling. So we are going to sign off for this episode. Mark Alvaro, I've heard a little rumor that there might be a podcast coming up, so I'm super excited for that.

Thanks for joining me for this episode of one app in ten minutes.

Speaker 0: We want to take a moment towards the end of the change log for thanking our amazing community contributors who give their time to improve the director's project. In January, we had 14 contributors, and so we'd like to say a huge thank you to Oscar for removing the deprecated webhooks functionality across the stack, Abdullah for removing the comment tab from the activities page, Thomas for adding concurrency control for file uploads via a new files max upload concurrency environment variable, 'kiki' for fixing an issue that would cause some draw header icons from being displayed too large, 'pancaj' for fixing incorrect initial slider fill position when the midpoint is not a valid stepped value, and for fixing markdown editor layout when a minimum input height is applied. VDR for fixing sticky column background in many to many list interface. Fan for improving system permissions collection picker to support easier multi selection. Ty for replacing the local use local storage composable with the view use equivalent, Daniel for disabling text highlighting for druggable view elements in Chrome and Firefox, Clint for fixing permission cache to respect cache system TTL, Bruno for fixing conversion of fields from object notation to dot syntax in SDK subscription queries, Arthur for fixing an issue where the Supabase storage driver would fail if the root folder is the empty string, and Joseph for adding support for specifying a KMS key ID in s three storage when using AWS KMS server side encryption.

Thank you again to these individuals. You can see their specific pull inside of the full release notes on GitHub. Lastly, we also want to take the time to thank our GitHub Sponsors of January who financially contribute to Directus' development. Thank you to Wayfan, Mike, Fergus, Omar, Marcus, Mission Control, Utomic, Steven, James, Manuel, Andreas, John, Burb, Adam, Jason, Yuya, Valentino, Jens and Wayne. The money we are given from our GitHub sponsors goes straight back to community members who build tooling and extensions for the director's ecosystem.

Thank you again for being part of that. Alright, that is it for this month's changelog, if you are still here still watching thank you very much for spending the time with us. If you have any questions head on over to the directors forum, have a great rest of your day and see you soon.