[ THE CURVE ]
The thing most "AI ROI" reports get wrong
An executive showed me a slide last week. Twelve months of internal AI rollout, polished into a 30% productivity gain. The board loved it. The CFO didn't believe it.
The slide was real. The gain was real. The reason it didn't show up on the P&L is structural, and it's exactly how an AI program ends up in a strange limbo where everyone says it's working and nothing changes.
Imagine someone is at 100% capacity…doing the wrong work. AI just allows them to do the wrong work faster.
The executives I've watched extract real value did one thing differently first: they redesigned what work gets done before they automated how it gets done. They treated AI as a facilitator of portfolio decisions, not a productivity dial.
Underneath that is an asymmetry. Cutting the wrong work is cheap and reversible; the hours you free have open upside. Automating the wrong work is the opposite bet — you spend real money to produce something nobody needed, faster. The reports chase the impressive number.
The curve rewards asymmetry.
A productivity dial is reversible. A portfolio decision charts the path.
The teams that 10x with AI aren't 10x faster at their old jobs. They're doing different jobs. New approaches to creating value are coming online as the cost of intelligence drops.
──────────────────────────────
[ THE SIGNALS ]
→ Half-Baked. Columbia University's "Fourth Purpose" framing — the idea that universities should build knowledge that addresses society's problems, not just transmit it to students. It's the only honest response I've seen to AI's effect on the academic offering. Watch which universities echo this in their fall 2026 messaging.
→ Hot Take. The shift from "AI adoption" rhetoric to "AI portfolio" framing. A few mid-cap CEOs started using the word portfolio in earnings calls this quarter when describing AI initiatives. It's a small linguistic shift with a large structural implication.
→ Confession. An important AI productivity stat that isn't a gain. In a controlled trial, experienced developers using 2025 AI tools took 19% longer to finish their work — while estimating AI had made them 20% faster. That gap between felt speed and real speed is the whole game — it's exactly how a program "works" while nothing on the P&L moves. The study.
──────────────────────────────
[ THE NEXUS ]
What work am I doing today only because intelligence used to be expensive?
If you can name it, you've just found your portfolio decision.
──────────────────────────────
[ THE MONDAY MOVE ]
Before you greenlight the next AI pilot:
Pull your team's five most time-consuming recurring deliverables. For each one, ask the question from The Nexus above — are we still doing this only because intelligence used to be expensive?
The deliverable with the weakest answer is your tell. It isn't an automation candidate — it's a deletion candidate. Bolting AI onto it just lets you produce something nobody needed, faster.
So cut one. Then reallocate the freed hours to a job the old economics priced out: that analysis you've been needing, the team you've not yet engaged, the question you never had time to chase.
That's the front edge of the Exponential Work Framework: 10X isn't your old job at speed. It's a different job that used to be uneconomical.
One cut. One reallocation. Monday. That's the whole move.
──────────────────────────────
[ THE COMPOUNDING ASSET ]
This week's compounding asset isn't a video — it's a tool. Context as Strategy is the three-move guide behind everything above: how to turn a throwaway AI session into an asset you reuse for a year. → Read it.
──────────────────────────────
Return Path
This newsletter is called Signals from the Curve because there are two kinds of forecasting: the curve and the cliff. The cliff says everything changes at once. The curve says it's already changing, you just have to know where to look.
I write here every Wednesday morning about what I'm seeing in AI strategy, leadership, and the parts of work that compound. The free guide Context as Strategy is linked above if you haven't grabbed it yet.
If something here changed how you're thinking, hit reply. I read every response.
— Chris
Chief of AI & Strategy at Essential Innovations · Founder, Attainable AI