There's a flood of studios now claiming to "do AI video." Most of them are selling speed and price. Here's what actually separates the ones worth hiring, and why the criteria matter more than the showreel.
A year ago, "AI video production" was a niche line on a handful of websites. Now it's everywhere. Search for a studio in London and you'll find dozens promising cinematic results "at a fraction of the cost," delivered "in days, not weeks." Some are excellent. Many are repackaging a free text-to-video tool and a markup.
The problem is that from the outside, they look identical. Everyone lists the same tools: Veo, Runway, Kling, Midjourney, HeyGen, ElevenLabs. Everyone has a slick reel. So how do you tell the studio that will protect your brand from the one that will quietly embarrass it?
This is the question we get asked most, usually by marketing and brand leads who've been burned once already. So here's our honest take on what to look for. We run an AI-augmented studio ourselves, so consider this a view from inside the tent. The criteria hold regardless of who you eventually hire.
The market has split into two camps, and they're easy to confuse.
The first camp sells output: cheap, fast, high-volume content, produced by pointing AI tools at a brief and shipping what comes back. It's genuinely useful for some jobs, such as disposable social tests, internal comms, and throwaway variants. If that's your need, optimise for price and turnaround, and don't overthink it.
The second camp sells outcomes: brand films, campaigns, and content where the work has to land emotionally, hold up next to your other marketing, and not become the reason you trend for the wrong reasons. This is a different discipline entirely, and most of the studios competing on price simply can't do it.
The mistake brands make is briefing the second job and hiring the first camp to do it. The criteria below are about avoiding exactly that.
1. They lead with creative direction, not the tool list.
Tools are table stakes now. Anyone can subscribe to the same generators you can. What you're paying a studio for is judgement: the decision about what to make, why, and how it should feel. If a studio's pitch is mostly a list of the AI platforms they use, that's a tell. The interesting question isn't "which model do you use?" It's "who's directing it, and what have they directed before?" Our own answer is a core in-house creative team working alongside a roster of trusted freelance specialists, including AI directors we've vetted and rostered specifically for this kind of work. Direction is the job. The tools just serve it. On Escape Convention, a cinematic car spot for KGM, we were handed an AI-generated storyboard and deliberately chose not to follow it to the letter. We rebuilt it as a written narrative first, set the tone and pacing, and only then generated to it. The idea led the tools, rather than the other way round.
2. They can close the last 15%.
AI will get a piece of work roughly 85% of the way there, fast. That last 15%, the timing of a cut, the grade, the sound design, the tiny details that separate "convincing" from "uncanny," is where craft lives, and it's the part AI can't do for you. We call it the 85% Problem internally, and it's the single biggest predictor of whether a piece of AI work feels premium or cheap. Ask a studio to walk you through how they finish work, not just how they generate it. If they don't have a strong answer, you've found the camp they're in. You feel that finishing in something like From Confusion to Confidence, an AI portrait series we made for SD Worx that lives entirely in the subtle stuff: the shift in an expression, the movement of the eyes, the lighting on a face as overwhelm turns to calm. None of that comes out of a prompt. It comes out of judgement.
3. They're straight with you about where AI is used, and where it isn't.
The AI backlash is real and it's measurable. Industry research in early 2026 found roughly one in five videos recommended by YouTube's algorithm now qualifies as AI "slop," and that the majority of viewers say uncanny-valley elements actively pull them out of the experience. Consumers are getting fast at spotting synthetic content, and brands from Coca-Cola to Mango have taken public hits for using it clumsily.
A studio worth hiring will have a clear point of view on disclosure, provenance standards like C2PA, and, crucially, when not to use AI at all. This is the kind of standard-setting we take seriously enough to help shape it: BearJam holds a seat on the APA's AI Taskforce, and we built and published a free 87-term AI Video Glossary precisely because the category needed a shared, honest vocabulary rather than more hype. It's also why, when Surrey's Office of the Police and Crime Commissioner asked us to create a deepfake of their own Commissioner to warn the public about AI-enabled fraud, we built it through an approved, transparent pipeline using only publicly available footage, with the educational purpose made unmistakable. Handled that way, it became the OPCC's most successful piece of content to date. Handled carelessly, the very same technique is exactly what we warn clients against. Reflexive AI-everything is now a brand-safety risk, not a selling point.
4. They have a real track record with real brands.
A reel can be assembled from experiments. A client list can't be faked. Look for evidence the studio has delivered to brand standard, at scale, for organisations whose reputations matter to them.
For context on the bar we mean: we've produced work for Netflix, Revolut, UBS and McDonald's, and we hold long-term anchor relationships with the Wall Street Journal and NBCUniversal, the kind of multi-year trust you only earn by delivering repeatedly rather than once. Our AI-led Visions of the Future campaign for Osprey ran as one hero film and 16 tailored versions across every major channel, reached 3.1 million people and lifted website traffic by almost 25%. Ask any studio you're considering who signed off their work and whether it ran. The difference between "we made this for fun" and "this went out for a global brand and performed" is the difference you're paying for.
5. They use AI to raise the ceiling, not just lower the floor.
Here's the most important distinction, and the one most buyers miss. There are two reasons to bring AI into production. One is to do the same thing cheaper. The other is to do things that simply weren't possible before: more ambitious ideas, more versions, more markets, more iteration within the same budget. The first is a race to the bottom. The second is where the real value is.
We've been running both hybrid AI and live-action pipelines and fully AI-generated production for over two years now, long enough to know that the studios still framing AI purely as a cost saving are competing on price, and you'll get what you pay for. 80 to Infinity is a good example of the alternative: SD Worx wanted to imagine the future of HR 80 years out for their anniversary, an idea that as live action with VFX would have carried a budget and timeline big enough to kill it on the spot. As our first fully AI film, it made the whole ambition affordable and got made. It cuts the other way too. On Keepr we blended live action with generative AI to land an emotional, sub-30-second brand story that neither approach would have nailed on its own. The studios worth your time talk about what AI lets them attempt that they couldn't before.
6. There's a human accountable for every frame.
When something ships under your brand, someone needs to own it, to have looked at it, judged it, and put their name to it. Slop happens when no one is accountable for the output, when the machine's first draft is the final draft. Ask who the named creative lead on your project will be. The answer should be a person, not a workflow.
For transparency: we're BearJam, a London-based studio that's been producing video since 2012 and has spent the last few years rebuilding around an AI-augmented model.
We've built our whole approach around the principle behind these criteria, something we call Craft Intelligence: AI to amplify human creative direction, never to replace it. It's why the last 15% is a non-negotiable line on our quotes rather than a corner we cut, and why we're as comfortable telling a client not to use AI for something as we are recommending it.
It also shapes what we do with the work. Through our Nature Impact partnership, we commit 1% of revenue from AI projects to UK biodiversity restoration via the Kensham Habitat Bank in Kent, because we think a studio's standards should show up in more than its showreel. The approach has been recognised in the industry too. Our AI and hybrid work has won real awards: the Osprey campaign above took Best Video Campaign at the International Content Marketing Awards, EV Marketing Campaign of the Year, and a Brand Film Award, and our fully AI SD Worx film won at the Lens Awards. Our founder and Executive Creative Director, James Hilditch, is also shortlisted as AI Communications Leader of the Year at the inaugural AI Comms Awards, where our Nature Impact partnership is shortlisted alongside it for Best Use of AI in ESG, CSR and Sustainability Communications.
That's our bias, stated plainly. But the test still stands on its own: whoever you're considering, run them through the six points above. The studio that answers them well is the one that'll protect your brand. The one that changes the subject to price is telling you something too.
The UK AI video market is loud, crowded, and genuinely hard to read from the outside. Strip away the tool lists and the turnaround promises and it comes down to a single question: is there craft and accountability behind the AI, or just the AI?
Get that answer right and AI video is one of the best things to happen to brand storytelling in a decade. Get it wrong and it's the fastest way yet to put something cheap-looking out under your name.
BearJam is an AI-augmented video production studio in London. If you're weighing up a project and want a straight conversation about whether, and how, AI should be part of it, get in touch.