Unautomatable: The Skills That Still Matter When AI Does the Rest

When intelligence itself becomes infrastructure, what is a serious career still for?

For two decades, Amit Singhal has built and led engineering organisations inside the kinds of institutions that cannot afford to be casual about technology — telecoms, customer-support software, payments at the scale of trillions a day. Unautomatable is a careful, first-hand argument about what changes for the people inside those institutions when the cognitive parts of their work become cheap and the dashboards stop tracking what actually matters.

This is not another book of forecasts about AGI, productivity tips, or "AI-proof" job lists. It is a structural diagnosis and a practical orientation: an account of why intelligence has stopped being scarce, why working harder feels useless, why the trades may outlast the analysts, and what kinds of judgment, ownership, taste, and moral authority remain irreducibly human in environments where so much else has been industrialised.

Across seven parts and thirty-six chapters, Singhal lays out:

• Why the bargain between productivity and leverage has broken — and what is silently hollowing out senior roles even as compensation rises.
• Why juniors today feel lost despite being more capable than any cohort before them, and why seniors feel exposed despite their experience.
• Why some of the most secure work of the next twenty years lives outside the office — in the trades, in care, in repair, in the embodied judgment that machines cannot supply.
• What unautomatable actually means, in eight careful cuts: judgment under irreducible uncertainty, the ownership of consequence, the context the system cannot see, sense-making, taste and restraint, moral authority, legitimacy, and trust.
• How to position a career around the durable advantages instead of the ones that are quietly deflating.
• Why AI is making organisations more fragile, not more robust — and which leaders are still worth following.

Drawn from conversations, decisions, and mistakes inside platforms that handle ten trillion dollars of corporate flow on a normal day, Unautomatable is the book for engineers, leaders, advisers, and parents quietly wondering whether they will still matter in a world that thinks for them. The answer, carefully argued: yes — but only if you choose, deliberately, what the next decade is going to ask of you.

For readers of The Decade Intelligence Changed, Working Backwards, The Hard Thing About Hard Things, and Working in Public.