Deconstructor of Fun proposes metered AI budgets as workflow redesign beats tool licenses

McKinsey's 2026 State of AI survey finds 80% report personal productivity gains from AI. Only 37% say AI has lifted company EBIT, unchanged from a year earlier. Consultant Michail Katkoff argues studios overspend on tool licenses and underspend on workflow redesign. His model for a $1M mid-size mobile studio puts 30% into change management, the largest single line. AI pricing has shifted from seat licenses to usage-based billing, pushing studios toward capped, metered budgets.

Unity's 2026 developer report finds only 5% of developers don't use AI at all.
GDC's 2026 survey finds just 36% of developers personally use AI tools.
52% say generative AI has hurt the industry, up from 30% a year earlier.
McKinsey's prior survey found workflow redesign had the strongest correlation with EBIT impact of any attribute tested.
"You are not buying tools. You are buying a different way of working." - Michail Katkoff, Deconstructor of Fun

Katkoff's $1M model splits spend: UA creative pipeline 25%, code assistance 25%, text/localization/voice 10%, art/assets 10%, change management 30%.
Playables reached 13.3% of mobile ad impressions in 2026, nearly doubling year over year per Sensor Tower.
Top studios now run about 100 playable concepts a month.
UA creative is ranked as the single highest-return category in the stack.

GitHub moved Copilot to usage-based billing in June 2026.
Uber burned through its entire 2026 AI budget in four months.
Uber now caps agentic coding spend at $1,500 per engineer per month.
Katkoff's prescription: budget AI like cloud spend, metered, capped, with one owner reviewing weekly.

McKinsey's 2025 data shows over half of profitable AI adopters had rebuilt workflows, not just added tools.
Their leaders were three times more likely to use AI themselves than peers.
AppsFlyer paused its OKRs for a quarter and ran 1,300 employees through an AI builder course.
King reportedly cut roles and replaced them with AI tools, leaving staff morale "in the gutter."
Krafton committed roughly $70M to a GPU cluster plus about $21M a year for AI enablement, then froze hiring.

Real AI spend runs $1.25M to $1.3M for every $1M planned, per Katkoff's client work.
The gap is funded mainly by junior roles that studios no longer open.
Stanford payroll data shows employment among 22 to 25 year olds in AI-exposed roles is about 19% below trend.
IBM reportedly tripled its junior intake this year, betting on talent pipelines three to five years out.
METR's 2025 study found experienced developers were 19% slower with AI while believing they were faster.

For Publishers & Developers: Assign one owner to review metered AI token spend weekly against per-engineer caps.
For Investors & VC: Add workflow-redesign evidence and junior-hiring freezes to studio diligence checklists.
For Service Vendors: Pitch playable-concept production capacity to mid-size mobile studios scaling creative volume.