Is In-Housing About To Accelerate Now That AI Narrows The Skills Gap?
One of the main reasons brands hired agencies rather than building capability internally was that agencies had specialist production skills a lean in-house team couldn't easily match. AI narrows that gap significantly. It doesn't close the other gap that made agencies valuable.
In-housing has been a steady trend for a decade, driven mostly by cost pressure and a desire for more control over brand consistency. What historically slowed it down was skill scarcity: an in-house team of generalists genuinely could not produce work at the craft level a specialist production agency could, at least not without hiring a specialist team of their own, at which point the cost savings became less obvious. AI tools meaningfully close that specific gap, letting smaller in-house teams produce work at a technical quality that used to require dedicated specialists.
What AI actually removes as a barrier
The production bottleneck, needing a video editor, a motion designer, a retoucher, each with years of tool-specific expertise, is exactly the kind of narrow technical skill AI assistance compresses fastest. A generalist in-house marketer with AI tools can now credibly produce work that would previously have required commissioning outside specialists, at a fraction of the turnaround time.
What it doesn't remove
What AI does not replicate is the breadth of pattern recognition an agency builds by working across dozens of brands and categories at once. An in-house team sees one business's problems repeatedly, deeply, but narrowly. An agency sees the same category of problem show up differently across many clients, and that comparative view is where a lot of genuinely useful strategic judgment comes from. No AI tool gives an in-house team that same breadth of exposure, because the tool only knows what it's shown, not what an agency's other clients are quietly learning in parallel.
AI makes it easier to build things in-house. It doesn't give an in-house team the same view across a category that an agency accumulates by default.
What this likely means for the split
The realistic outcome is not a wholesale collapse of the agency model, but a further narrowing of what brands keep in-house versus what they still pay an agency for. Execution and production, the layer AI compresses fastest, is the part most likely to keep moving in-house. Strategic thinking that depends on cross-category pattern recognition is the part most likely to stay with an agency, precisely because it's the part AI genuinely cannot substitute for.