Google DeepMind is investing $75 million in independent film studio A24 to embed its researchers inside productions and build AI tools shaped by filmmaker feedback, rather than handing the industry generic software. Strip away the Hollywood specifics and the deal is a clear template for how any business should adopt AI in work that depends on craft and trust.
What Google and A24 Actually Agreed To
In June 2026, Google DeepMind announced a first-of-its-kind research partnership with A24, the studio behind a string of acclaimed and cult films. The reported terms are a $75 million investment from DeepMind, in line with what venture investors put into A24 in its last funding round, paired with a working collaboration rather than a simple licensing arrangement.
The structure is what makes it notable. DeepMind researchers will embed inside A24 productions and use real filmmaker feedback to build and refine tools, starting with AI-generated storyboards. Two guardrails are explicit in the reporting: the deal does not give Google access to A24's content library or its data, and there is no mandate forcing filmmakers to use the resulting tools.
A24 partner Scott Belsky framed the output as deliberately different from what most people picture when they hear generative AI. The new tools, he said, will not look anything like the prompted generation type of AI that people feel uncomfortable with, and are meant to preserve creative control and support risk-taking.
Why "Shaped by the Creators" Is More Than a Slogan
DeepMind's framing for the partnership is that the tools of the future are shaped by the creators who use them. That phrase is easy to dismiss as marketing, but it describes a genuine design choice that most AI deployments get wrong.
The default way companies adopt AI is to buy a finished product, push it out to teams, and hope usage follows. The A24 model inverts that. Builders sit next to practitioners, watch the actual work, and let the workflow dictate what gets built. The tool is a byproduct of understanding the job, not a generic capability looking for somewhere to land.
This matters because craft work resists generic automation. A storyboard is not just an image; it encodes a director's intent, pacing, and visual language. A tool that ignores those constraints produces output a professional has to throw away. Our take: the same is true of a financial analyst's model, a lawyer's brief, a nurse's intake notes, or a support agent's escalation judgment. In any function where the output carries professional judgment, the distance between a generic AI feature and a useful one is the distance between the vendor's assumptions and the practitioner's reality. Closing that gap is design work, and design work requires the practitioner in the room.
The Backlash, and the Strategic Bet Underneath It
The deal did not land quietly. A24 has a fiercely loyal audience built on a reputation for protecting artistic vision, and many of those fans reacted to the Google tie-up as a betrayal, with some publicly threatening to cancel their studio memberships.
A24's defense is the most instructive part of the story for business leaders. The studio said it would rather have a seat at the table than on the sidelines, and that the partnership exists so the studio can help dictate what tools get built for artists rather than having tools handed to them later. In other words, A24 concluded that AI is going to reshape filmmaking whether it participates or not, and that shaping those tools from inside is safer than inheriting whatever the broader market produces.
That is a posture decision, not a technology decision, and every company faces a version of it. You can wait for AI tools to arrive in your industry, built on someone else's assumptions about your work, or you can invest early in shaping tools around your own processes and standards. The first path is cheaper in the short term and cedes control. The second costs more attention up front and buys influence over the outcome. A24 took the backlash because it judged the influence to be worth more than the comfort.
What This Means for Your Business
You do not need a $75 million research lab to apply the lesson. The A24 deal validates a deployment model that works at any scale, and it lines up with what the adoption data has been showing for the past year.
Most failed AI initiatives fail at adoption, not at the model. When the majority of enterprise workers quietly avoid or resist the AI tools their employers deploy, the root cause is rarely the technology. It is that tools were chosen and configured without the people who have to use them, so they do not fit the actual work. A24's structure is the direct antidote: practitioners shape the tool, so the tool fits the practitioner.
The harder, more durable work is the opposite of a generic rollout. It is building AI tools around the specific contours of an existing workflow so the output matches how people already work and the standards they are held to. That is why so many off-the-shelf AI features stall after the pilot. They solve a generic version of a problem that, up close, is never generic.
There is also a brand and trust dimension that A24 is navigating in public. Creative and knowledge work carry reputational weight, and audiences and clients increasingly care how AI is used behind the scenes. The same tension is showing up in how AI is reshaping content and creative marketing, where the line between augmentation and replacement determines whether an audience trusts the result. Co-designing tools that keep humans in control is not only a productivity strategy; it is a way to adopt AI without quietly eroding the thing that made the work valuable in the first place.
How to Co-Design AI Tools With the People Who Use Them
The A24 model is repeatable. Here is how to run a smaller version of it inside your own organization.
-
Start with one real workflow, not a tool shopping list. Pick a specific, high-friction process owned by people who can articulate what good looks like. The goal is to improve that job, not to deploy AI in the abstract.
-
Put builders next to practitioners. Whether the builders are an internal team or an outside partner, they should observe the actual work before proposing anything. The insight that makes a tool useful usually lives in details practitioners never think to write in a requirements doc.
-
Design for control, not autonomy, at first. A24's stated priority is preserving creative control. Mirror that. Tools that assist and keep a human in the decision earn trust faster than tools that try to automate the judgment away, and trust is what unlocks adoption.
-
Iterate in short cycles against real output. Test the tool on work the team just completed and compare quality and time honestly. If the output needs heavy rework, the tool is not ready, regardless of how impressive the demo looked.
-
Move from pilot to production deliberately. Many promising projects die in the gap between a working prototype and a dependable system. Treat the transition from pilot to production as its own phase with its own owners, rather than assuming a good demo will scale itself.
Common Mistakes to Avoid
Buying the tool before understanding the work. A polished product chosen by leadership and pushed to teams reliably produces the resistance pattern the adoption data describes. The sequence matters: understand the workflow, then build or buy.
Confusing a demo with a deployment. Creative and judgment-heavy work is full of edge cases that demos skip. Evaluate on three real jobs you just finished, not on a curated showcase.
Automating the judgment instead of supporting it. The fastest way to lose practitioners is to signal that the tool exists to replace their discretion. A24's framing, preserving control and supporting risk-taking, is the opposite signal, and it is the right one.
Ignoring the trust and brand stakes. How you adopt AI is visible to employees, customers, and audiences. Treat the perception of the rollout as part of the rollout, not an afterthought.
Key Takeaways
- Google DeepMind is investing $75 million in A24 to embed researchers in productions and build AI filmmaking tools shaped by practitioner feedback, starting with AI-generated storyboards.
- The deal explicitly excludes access to A24's content library and data and does not mandate that filmmakers use the tools, keeping proprietary creative work separate from the research.
- A24's "seat at the table" defense reframes AI adoption as a posture decision: shape the tools that will reshape your industry, or inherit tools built on someone else's assumptions.
- The model generalizes. In any craft-heavy or judgment-heavy function, co-designing AI around the real workflow drives stronger adoption than deploying generic software top-down.
- The biggest risk in enterprise AI is adoption, not capability. Putting practitioners in the design loop is the most reliable way to close that gap.
The businesses that move early on co-designing AI into their workflows will have a meaningful advantage. If you want to be one of them, let's start with a conversation.