Case study
Premium TV drama: an end-to-end AI policy that saved five second unit days
We wrote a production company's end-to-end AI policy, built on responsibility, security and ethics, then put it to work on a premium TV drama. Using generative AI within that policy saved five second unit days, and the money went back into the show.
About the client
An independent film and television production company with offices in the UK and the US. The show is a premium scripted drama with a major international cast, shooting largely on location in Europe, with distribution partners in the US and the UK. We can't name the company or the production yet.
The challenge
The company wanted to use generative AI properly: across the business, and on a live production where it could make a real difference to the budget. But it had no AI policy, and the show sat across several sets of rules at once:
- UK and EU law, including copyright and data protection
- The AI standards of a major US studio partner, supplied as a third-party AI tools addendum
- The AI guidelines of the show's distribution partners
- The position of the performer unions in each territory
None of these were written with the company's situation in mind. It needed one clear policy its people could follow, and that its partners could trust.
Our approach
We wrote the company's AI policy end to end: AI guidelines and guardrails for the whole business, and a practical how-to guide for the people using the tools every day. It rests on three principles.
- Responsibility. Clear rules on what AI is and isn't used for, who approves it and where human creative control is non-negotiable. Cast and crew are protected: no principal likenesses, no voice cloning and no digital replicas.
- Security. Approved tools only, with scripts, footage and client material kept on platforms and accounts the company and its partners control. AI only gets the access it needs.
- Ethics. Respect for performers, consent and licensing, and compliance with UK and EU law and each partner's standards. Where the rules differ, the policy follows the strictest one, not the most convenient.
Then we trained the company's management and production teams to use Claude within the policy, so the guidelines became how people actually work.
Putting it to work on the show
The first test was the drama itself. The producers wanted generative AI for a handful of low-level shots: background crowd and b-roll that would otherwise have needed second unit days or stock footage.
- We mapped every planned shot against the studio's AI addendum, UK and EU law and the new policy
- Generation ran on a platform licensed through the distributor's own account, so outputs and terms stayed inside a relationship the distributor already controlled
- Principal cast stayed entirely out of scope
- Every generated shot is logged, with what it replaced and how it was made
- Disclosure and approval records are in place, so the producers can show their partners exactly what was made, how and with what
The same framework is now guiding the company’s next production, adapted to that project’s partners.
Outcome
Second unit days saved
Generative AI replaced five second unit days on the drama, and the production reinvested that money in other parts of the show.
A complete AI policy
Company-wide AI guidelines and guardrails plus a how-to guide, built on responsibility, security and ethics, and adopted across the business.
Within the law and partner standards
An approach that respects UK and EU law and the standards of the studio and distribution partners on the show.
Teams using AI the right way
Hands-on Claude training for management and production, grounded in the new policy.
Bring us the complexity.
Tell us what’s tangled, the AI you want to build, the IT you’re tired of worrying about, or both. The first call is free and jargon-free.