Business & Beyond You Don’t Need to Understand Everything About AI - Here’s What Matters | Rohin Wood (BCG X ANZ Lead, and Managing Director & Partner BCG)
Mar 22, 2026
Rohin Wood, BCG X ANZ leader with a PhD in engineering, explains practical AI for business. He discusses how agentic systems change commerce and workflows. He covers why pilots stall and how to scale transformation. He highlights tooling that democratizes AI and rising cyber risks.
AI Snips
Chapters
Transcript
Episode notes
Agentic AI Works Best When Constrained
- Agentic AI is when algorithms decide what tasks to do; practical deployments constrain agents heavily rather than unleashing unconstrained self‑organizing swarms.
- Rohin contrasts constrained task sequences (working today) with sci‑fi open agents (largely theoretical).
Treat AI Interfaces As Strategic Distribution Channels
- Prepare for AI platforms to become primary entry points that own the screen and user relationship, threatening traditional channels.
- Rohin urges businesses to consider how owning the interface and data could remake retail, loyalty, and advertising.
Most LLM Work Is Code And Companions Today
- Usage patterns show large proportions of LLM tokens are for code and virtual companions, indicating where current economic value concentrates.
- Rohin cites a token analysis finding ~50% for code and ~25% for virtual companions.