Tech Landscape #421
Lots of AI creative tool upgrades, the OpenClaw era, and Meta states the state of VR.
Hello!
Lately, because of work, I’ve had to watch a lot of ads made with AI — and pretty much all of them are disappointing. Not because of the level of craft they’re made with, necessarily, but because with (very) few exceptions they’re just imitations of the ads we make now, but cheaper.
They’re in the “radio on the TV” phase; early broadcasters didn’t immediately comprehend the unique visual capabilities of TV so they put radio announcers in front of cameras and reused radio scripts. It was a transition phase until people realised this was a new medium and they could do things in a new way.
I want to see brands using AI to do new and unexpected things. There are so many amazing creators posting videos to social media that are way more inventive and exciting and use this technology so much better, but brands (and agencies) largely are lagging behind. I really hope we pass this phase soon.
Anyway, let’s get on with it. Hope you’re well!
Synthetic Audio-Video
Sora 2’s best capabilities are more broadly available; custom characters and objects, clips up to 20 seconds, 16:9 / 9:16 aspect ratios, and more are all available through the API. x.com/OpenAIDevs
This will mostly impact people who use Sora through third-party tools, rather than in the official app where these features are already available (still not in the UK, though). BTW, remember how everyone lost their minds over Sora 2 when it launched? That passed quickly.
When Disney and OpenAI announced their partnership on Sora 2 part of the agreement was to “make a selection of… fan-inspired Sora short form videos available to stream on Disney+” — which just added Verts, a vertical video feed on its mobile app, where those videos will appear.
Runway launched Characters, real-time expressive conversational avatars generated from a single image, with full control over voice, personality, and actions. runwayml.com
xAI’s Grok Imagine can now take up to 7 reference images, on iOS, Android, and Web. x.com/grok
Canva introduced Magic Layers, an AI feature that converts flat images into fully editable, multi-layered designs. canva.com
This is very cool; it separates the elements into layers and in-paints the empty spaces. Adobe should definitely be working on a version of this.
Adobe launched AI Assistant for Photoshop, a conversational tool that lets users edit images using text or voice prompts, available in public beta on Web and mobile. It also updated the Firefly Image Editor with Generative Fill, Remove, Expand, and Upscale. blog.adobe.com
I’ve tested it briefly and it seems to work as intended, but I’m not sure it makes editing any easier.
Adobe also added Rotate Object (beta) to Photoshop (beta), which uses AI to transform a 2D image into 3D that can be rotated and positioned better within an image, and then blended with the Harmonize tool. community.adobe.com
Hedra launched Agent, letting users create images, videos, and full content workflows through a chat interface. hedra.com
Freepik added Speak, an image/video lip-syncing feature. x.com/freepik
It’s powered by VEED’s fabric 1.0 model (TL 398). You can see in my tests ⬇️ that lip-syncing and motion are generally good, although there are some odd artefacts (the text on the paper, the subtitles).
Higgsfield added Audio, enabling lip-syncing, dubbing, and more. threads.com/@higgsfield.ai
Moonvalley is ending public subscriptions and pivoting to an industry / B2B tool. It’s a bit of a shame; Moonvalley was trained only on licensed data, and while it wasn’t the best AI video tool it was the best ‘commercially safe’ alternative to Firefly Video and I’d hoped to see it develop further.
Assistants & Search
Claude introduced interactive visuals (in beta), to generate charts and diagrams directly within chat responses. claude.com
Google updated Maps with Gemini features: Ask Maps for conversational search and recommendations, and Immersive Navigation for routes and directions. blog.google
Microsoft launched Copilot Health, a secure and dedicated space within Copilot that provides personalized medical insights by aggregating data from health records and wearable devices. Currently available via a waitlist for adults in the U.S. microsoft.ai
The OpenClaw Era
Jensen Huang of NVIDIA recently described OpenClaw as “the single most important release of software probably ever”, and you can certainly feel its influence.
Genspark launched AI Workspace 3.0, an updated platform featuring “Genspark Claw,” an autonomous cloud-based AI employee that can run multi-step workflows and manage tasks across third-party applications. mainfunc.ai
Perplexity launched Personal Computer, an OpenClaw copy that runs on a dedicated Mac Mini, connected to local apps and Perplexity’s secure servers. perplexity.ai
Meta bought Moltbook, the bot-only social network, in an ‘acquihire’ deal for its two creators. axios.com
Social
Digg shut down suddenly and announced ‘significant downsizing’, blaming bot manipulation and the difficulty of growing an audience. digg.com
It also says “positioning Digg as simply an alternative to incumbents wasn't imaginative enough”, with which I concur.Discord announced updates for game developers and players, including improved Account Linking, expanded Social Commerce for in-chat gifting, and Discord Official which lets game developers claim and customise their game’s server. discord.com
Facebook Marketplace introduced new Meta AI tools including automated listings from images, and auto-generated replies to buyer inquiries. about.fb.com
Meta Edits added Freeze Frame, SoundFX, and a Teleprompter, and 11 new Effects, customisable overlays, and an updated ideas tab.
Meta Horizon
Meta gave a ‘state of VR’ talk at GDC which is worth a look. It says that usage hit an all-time high in 2025, developer revenue was slightly up, over 100 titles earned $1m+, and in-app-purchases showed significant growth. There are new dev tools coming, but the cost of making AAA titles is too high which is why it shut down a bunch of studios. The core audience is maturing and might need different types of games, hand tracking is a preferred input method for “mainstream adults”, and there are new devices in development. developers.meta.com