Share with your CIO
Intuit is treating AI as an internal accelerant, not an existential threat, and its numbers are starting to make the case. Ashok Srivastava, the company’s chief AI officer, reports a 40% increase in development speed over the past year and a 12x increase in development velocity over five years, driven by agentic coding tools. QuickBooks Live subscriptions doubled year-over-year after AI integration, and the AI-powered QuickBooks Capital lending platform posted 73% year-over-year growth. Srivastava is also pushing back against tokenmaxxing, the practice of maximizing AI token consumption to signal adoption, calling it expensive theater disconnected from customer value.
What this means for your business
The SaaSpocalypse narrative, the idea that AI agents will simply replace the workflow software layer entirely, assumes incumbents are standing still. Intuit is not a startup with one product to protect; it runs QuickBooks, TurboTax, and Credit Karma across millions of small business and consumer accounts, and it’s using that installed base as the distribution advantage that new AI entrants don’t have. If you’re a CIO evaluating whether your existing SaaS vendors are AI threats or AI partners, Intuit is the clearest case study yet that incumbency plus data plus disciplined execution is a defensible position.
The tokenmaxxing rejection deserves more attention than it’s getting. A growing number of enterprises are measuring AI adoption by token consumption, essentially counting how much they’re spending on model calls as proof of progress. That’s the FinOps equivalent of measuring software project success by lines of code. Srivastava’s framing, that token spend divorced from customer outcomes is waste by another name, is exactly the discipline most AI budgets currently lack. CIOs who haven’t yet established an outcome-linked unit of AI value, not tokens consumed but customer results produced, are building a cost center, not a capability.
The real stress test for Intuit’s position is whether trust and compliance hold as a moat once AI commoditizes the underlying model layer. Srivastava says customers doubled down on QuickBooks Live specifically because they wanted Intuit’s involvement alongside the AI, not just the AI alone. That’s a signal worth tracking. If customers in regulated, high-stakes domains, financial compliance, tax accuracy, lending decisions, still pay a premium for a trusted human-in-the-loop wrapper, the incumbents win the next decade. If the models get good enough fast enough that the wrapper stops mattering, they don’t. The renewal decisions your SMB-serving business units make over the next 18 months are the leading indicator.
Concept deep-dive: Tokenmaxxing
Tokenmaxxing is the practice of directing employees or systems to spend as many AI tokens (the units of text a model processes, roughly analogous to word-chunks that determine compute cost) as possible, treating high consumption as evidence of AI adoption. It emerged as organizations struggled to measure AI ROI and defaulted to input metrics instead of output ones. The business risk is straightforward: it optimizes for spend rather than results, creating AI costs that grow without a corresponding improvement in what customers actually experience.
Based on reporting from Intuit’s chief AI officer on the SaaSpocalypse and disciplined AI, originally published 2026-06-11 03:00:00.
