The lab is the studio.

Agencies have R&D decks. Studios have innovation officers. We have neither. The research happens where the work happens, on the same machines, on the same days, by the same people.

This collapses a familiar feedback loop. A director's question — can we hold this character across eight spots — becomes a tooling problem on Monday and a production decision by Friday. Nothing has to be "productized" before it ships. Nothing has to be pitched up a chain. The path from question to artifact is one room wide.

What the industry usually does

The orthodox model dates to the 1970s lab-vs-studio split. Xerox PARC, Lucasfilm Computer Division, MIT's Architecture Machine Group — all great labs, all eventually estranged from the production they were supposed to feed [1]. The pattern is consistent: research output piles up, very little of it crosses the wall, and the studio keeps shipping with whatever last shipped. Christensen described the same gap in 1997 and gave it a name [2].

The defense for the split is real. Research needs slack; production needs throughput. Put them in one room and the throughput wins every time, because there is always a deadline.

What we do instead

We accept that throughput wins, and we treat the slack as a budget the project pays. Each engagement carries a research line: a question the team owes back to the studio at wrap. Sometimes the answer is a tool. Sometimes it is a one-page note in the archive. Sometimes it is the negative result that closes off a direction for the next director who would have tried it.

• Tools are owned by the project that birthed them and inherited by the next one that needs them.
• Notes go into the archive on commit, not after the wrap party.
• Negative results get the same shelf space as wins, because they save the next bid.

The trade-off is honest: we publish less than a pure research lab and we ship slower than a pure production company. What we get in return is a studio that does not pretend the tools are settled.

Why this matters now

The stack underneath moving image is changing on a quarterly cadence. New samplers, new control models, new audio pipelines — by the time a traditional R&D group has a memo, the memo is stale [3]. The only honest way to keep up is to do the research in production and ship the result in the same week. Tools age in months now [4]; org charts have to follow.

The takeaway

Research is not a department. It is a discipline applied to whatever the studio is already doing. Run it any other way and the work and the thinking drift apart, and the work loses.