Marketing teams: Layer just connected to Claude!
Tell your AI the campaign you need →
Layer runs the creative workflows →
assets come back on-brand and channel-ready.
No extra headcount.
No agency wait time. 🧵
When I became a manager after being an engineer, I had to make a significant mindset shift. I had to hold myself back from forcing a specific solution, trust the outcome and not dictate the way, not get frustrated because “I would have done it differently”. This is almost exactly the mindset shift needed for making agents work - and now almost everyone is getting a taste of being a manager and are forced to think about a) how do I set the right framework and guidance for my age t to succeed and b) how do I measure the success of the outcome and only look at that rather than micromanage and block progress.
Agents coding now gets you 95% of the way (up from maybe 50% a year ago) - but years of software experience is what's needed to get it to 100%. AI is a force multiplier for professionals with good foundational skills, not a replacement!
This human can make a decision far in advance of the triggered action or schedule - but the framework of the decisions and actions is set and controlled by the human.
No matter how hype-y AI news gets, always remember - there will always be a human making a decision and triggering an action (or setting up a schedule to trigger actions) at the top of all the abstractions.
Trying out an agent based workflow to automate something that's quite manual today - wish me luck!
We use (and love) @TrustVanta for @layer_ai's compliance program, and this includes vendor management. For every new vendor we need to fill in a bunch of publicly available info, and for every vendor every year we need to conduct a security review where someone needs to collect several documents such as SOC 2 reports, ISO27001 reports, questionnaires, etc. As you can imagine this gets quite repetitive and cumbersome over time.
In comes @AnthropicAI Claude - especially the Claude Browser Extension. Here's the plan of action that it executed:
Approach to follow
1. Review the current vendor page in Vanta to understand the vendor and existing information
2. For each vendor: research public information (website, HQ location, support contact) using Google
3. Search for publicly available security reports (SOC 2, ISO 27001) for the vendor
4. Update vendor fields: set business owner based on employee directory, select appropriate category, add website, set Auth Method to Google Workspace, set HQ country, add point of contact if available, populate services description
5. Navigate to the security review page and upload any found security documents
6. Set an inherent risk score under Risk Management
7. Navigate to the next vendor and repeat until all vendors are reviewed
Excited about the potential here to save our team a lot of time.
Getting increasingly excited about continuous media generation driven by agents! Agents have become so powerful in software, and once a few missing pieces are complete there's no reason why media generation cannot be the same
Announcing Layer Workflows: The shift from Generative to Operational
Over the last year, it became clear to us at Layer that studios don't need more assets for the sake of assets — ultimately they need lower CAC and higher LTV.
📰 Creative has become the only real lever left in UA, with media buying becoming a global commodity.
♻️ Best UA teams found success via continuously iterating on new ideas.
👟 Content hungry LiveOps events has become table stakes for maximizing LTV.
That is why we built Workflows.
Layer is no longer just about one-off generations; it is about running continuous creative operations. It allows transferring know-how from AI experts to art teams, allowing studios to take control again, and unlocking the scale, consistency, and speed the industry has been waiting for.
Game studios are racing to create faster, but GPU infra was holding @layer_ai back.
They were running their own clusters: low utilization, high cost, constant maintenance.
Then they switched to fal.
⚡️ 24hr model deployments
🏎️ 2× faster inference
💸 30% lower GPU spend
🎯 Now powering 300+ studios (Zynga, SciPlay, IGT & more)
New model: Imagen 4 by Google
Imagen 4 is the hottest text-to-image model available - and for good reason. With a focus on photorealism, super fine details, and accurate typography - Imagen 4 is a great tool for a wide range of usecases.
Perfect for game studios working across a range of genres, Imagen 4 supports everything from photorealism to stylized cartoons and anime. It’s also quite good at generating accurate dynamic scenes while avoiding annoying artifacts or inconsistencies.
Give it a try today on Layer: https://t.co/UhjNiwR4xH
#gamedev #genai #gamedevelopment #gametools #content #ai #layerai #imagen4 #google
Direct AI video like a pro: First Frame/Last Frame is now LIVE 🎬
With this awesome new feature, you can now define the opening and closing shots of video generations with key image “frames”. No more guesswork. Just smooth, consistent videos that hit exactly what you imagined.
Why is this important? You can now create accurate and consistent videos over longer periods of time.
Perfect for:
📣 Short 15-20 second ads for marketing campaigns
👾 Accurate in-game cutscenes with character movements.
🏞️ Longer animated backgrounds for game scenes.
🔥 Social media assets.
Getting started is easy - just head your Layer workspace, select Video forging, and Kling 1.6 or 2.0 as a model.
Check out our docs for all the details: https://t.co/GvD0Ot9p6b
#gamedev #genai #aivideo #kling #video #gametools #gamedevelopment