The Innovation Vista Thesis

Jeff Roberts, Founder and CEO, Innovation Vista · Last reviewed September 28, 2026

Most of what gets written about AI and business is a warning or a promise. This page is neither. It is the argument Innovation Vista has been making about what AI is doing to mid-market companies, stated once and in full. The predictions that follow from it are dated, sourced, and scored on a page of their own, and we keep the ones we get wrong. This page is the argument.

The Thesis in One Paragraph

Efficiency is where every company starts with AI, and efficiency alone is a trap: it answers the questions you already had and leaves the open ones untouched. The productivity AI creates does not flow to the company that deployed it. It is intercepted twice on the way, once by employees who quietly keep the hours it frees, and once by vendors who change the unit of pricing so that cheaper delivery arrives at yesterday’s price. The deflation everyone expects therefore does not show up on schedule. It arrives in 2027 and 2028, when contracts are redesigned and labor markets finish adjusting, and when it comes it brings a policy response and a money response with it. Which side of that repricing you land on is set less by your ambition than by your industry’s shape: in some sectors AI is a wedge for market share, in others a tool for margin, in others a hub for consolidation, and in a few a moat that ends competition. And inside any shape, AI does not skip stages. Companies that stabilize, then optimize, then monetize, under governance built to make them faster rather than slower, capture the dividend. Companies that leap to the exciting part do not. That is the whole argument. Everything below is a link that supports it.

The Argument in Six Moves

Move one. Efficiency is the entrance, not the destination. Every company begins its AI work by doing what it already does faster, and every efficiency framework you own will help with that. None of them can answer a genuinely open question: whether to enter a market, exit a line, change what you sell, or change who you sell to. Innovating Beyond Efficiency is the book and the method for those questions. It begins where the efficiency frameworks stop.

Move two. The dividend is intercepted twice. When AI makes a task cheaper, the saving passes through two sets of hands before it reaches you. Employees find the freed hours first and, in most organizations, keep them. Vendors find the cheaper delivery next and switch the pricing unit, from hours or seats to fixed bids and outcomes, anchored at the price you were already paying. The productivity is real. Your share of it is a negotiation you may not know you are in. Your AI Dividend Is Going to the Wrong Bank Account and Vendor Partnerships in the Era of AI Arbitrage.

Move three. The Great Repricing is a 2027 and 2028 event, and it has three legs. Labor first: white-collar payrolls that do not recover their prior peak and a college wage premium that keeps narrowing. Policy second: an income-support response that becomes law or a platform plank once displacement is undeniable. Money third, conditional on the second: the financing of that response is inflationary, which is the leg we owe to Lyn Alden’s work on debasement and extend to the AI case. The buyer-side deflation everyone expects is delayed by move two and arrives only when contracts are redesigned, which we place in the same window. The Great Repricing, the earlier labor-only version, and Doom & Gloom or Wine & Roses? as the map of where this sits among the forecasts.

Move four. Your industry’s shape decides what AI is to you. Three questions set it. Can buyers specify what they want? Do they shop for it? Will they sole-source it, collapsing onto one or two providers, or will capacity, geography, licensing, or plain preference keep the market split? The answers sort industries into four shapes. In a Rising Bar sector AI is a wedge: the bar rises because of it, and it wins share for whoever moves. In a Slow Melt sector AI is a tool: margin, not share, and no reason to pretend otherwise. In a Rollup sector AI is a hub, the integration layer that makes acquisitions add up. In a Falling Floor sector AI is a moat, and the floor keeps falling until competition stops. You do not choose your shape. You can read it. The Four Shapes of the AI Dividend and The 2026 AI Dividend Map, refreshed every year.

Move five. Your next customer arrives as software. Buying agents are being deployed faster than selling interfaces are being built. For a while, the mid-market seller with a transactable, agent-facing surface, a dock, will be visible to counterparties its competitors cannot see. That asymmetry closes eventually. The order in which it closes decides who gets the early orders. Your Next Customer Won’t Be Human · MCP and the Rise of Agent Docks.

Move six. AI does not skip stages. Whatever your shape, the sequence is Stabilize, then Optimize, then Monetize, and companies that leap to the third stage from an unstable base spend the dividend on cleanup. Governance is what makes the sequence fast rather than slow; guardrails are the reason a freeway carries traffic at speed. Governance is a Freeway, Not a Roadblock, AI Doesn’t Skip Stages, and the Stabilize, Optimize, and Monetize pages.

What Would Change Our Mind

If sector-level price deflation shows up in professional and business services during 2026 while vendor margins hold, the interception thesis is wrong and we will say so. If advisory revenue at the large firms grows through 2028, the expert-direct thesis is early or wrong. If the fast-follower discount reappears, meaning the bottom quartile closes the gap by buying what the leaders built, then the first-mover claim fails and our paralysis articles were bad advice. Each of these is a real possibility. That is what makes the argument worth reading, and it is why every claim above that can be tested has a date on it.

The claims that test this argument, dated, sourced, and scored, are on the ledger. If you want to know which shape your industry sits in, the 2026 AI Dividend Map is here. If you want the method for moving through the stages without skipping one, that is the book and the work.