The Expectation Reset

Core thesis

AI has reset customer expectations across seven dimensions. Products built before 2023 are now judged by standards they were never designed to meet. The expectation reset is silent. Customers do not file tickets saying "your product fails the new expectation standard." They leave. They churn. They tell their colleagues your product feels old. The reset applies to every product, not just AI products. A project management tool, a CRM, an analytics dashboard: all are compared to the best AI-powered experience the user has ever had, not the best experience in the product's category.

Reset 1: From "a place for me to create" to "do the work for me"

Customers no longer want tools. They want outcomes. A marketing manager does not want a campaign builder. She wants a campaign. A data analyst does not want a query interface. He wants the answer. This reset is the most fundamental and the most disruptive to existing SaaS products. Products designed as creation surfaces, where the user brings the intent and the product provides the canvas, are now compared to products where the user states the outcome and the product produces it. The Writer 2026 survey found that 97 percent of executives deployed AI agents, but only 29 percent saw ROI. The gap is often that the agent was deployed as a feature inside a creation-surface product. The user still had to do most of the work. The expectation had already shifted past what the product delivered.

Reset 2: From "one size, I customize" to "custom made for me"

Personalization is the baseline, not a differentiator. Customers expect the product to know who they are, what they have done before, what their preferences are, and to adapt accordingly. A customer who logs into a dashboard and sees the same default view as every other user now experiences this as a product failure, not a neutral experience. The data supports this. A 2025 Zendesk survey found that 71 percent of consumers expect personalization, and 76 percent get frustrated when they do not receive it. AI makes personalization cheap enough to be table stakes. Products that do not personalize are not competing on features. They are competing on whether the user feels known. Feeling unknown is churn.

Reset 3: From "I will do the busy work" to "the busy work is done for me"

Busy work is any task the user must complete that does not require their unique judgment. Data entry, formatting, categorization, summarization, scheduling. Before AI, these tasks were accepted as part of using software. After AI, they are experienced as the product wasting the user's time. The LangChain State of Agent Engineering survey found that the top three agent use cases in production are customer service at 26.5 percent, research and data analysis at 24.4 percent, and internal workflow automation at 18 percent. All three are busy-work elimination. The products winning in each category are not the ones with the best busy-work interfaces. They are the ones where the busy work simply does not exist. The user states the outcome. The product handles the steps.

Reset 4: From "I will pay per seat" to "I will pay for output"

Per-seat pricing assumes value scales with users. AI breaks this assumption. When an agent does the work of five people, the customer sees five seats of cost and one unit of output. The math stops working. Outcome-based pricing replaces feature-based pricing: the customer pays for reports generated, campaigns run, tickets resolved, not for seats that might produce value. Gartner's April 2026 CEO survey found that 28 percent of CEOs say transactional revenue is most at risk from AI, because agents can bypass intermediated systems and enable real-time pricing. Products still priced per seat are vulnerable to any competitor that prices per outcome. The shift from per-seat to per-outcome is not a pricing change. It is a business model migration that most per-seat products cannot execute without rebuilding their revenue engine.

Reset 5: From "I expect to wait" to "I expect it now"

Latency tolerance has collapsed. Users judge products in seconds, not minutes. A human customer service agent taking four hours to respond was acceptable in 2022. In 2026, an AI agent that takes four seconds feels slow. This reset applies far beyond customer service. Dashboard load times, report generation, search results: every interaction is compared to the speed of a direct LLM response. The LangChain survey found that latency is the second-largest barrier to production, cited by 20 percent of teams. But latency is not just a technical problem. It is an expectation problem. A product that takes ten seconds when the user expects two seconds feels broken, even if ten seconds is objectively fast. Speed is relative to the user's most recent AI experience, not your historical benchmark.

Reset 6: From "I will learn this workflow" to "the interface adapts to me"

Onboarding friction is now churn. Users expect the product to adapt to their mental model, not the other way around. A product that requires a tutorial, a setup wizard, or a training video before the user can complete the first task is now competing against products where the user types a sentence and gets a result. The Deloitte 2026 report found that 53 percent of organizations are investing in workforce AI education. But the best AI products are not educating users. They are eliminating the need for education. They have no onboarding flow. They have a text input and an output. The user learns by doing, not by watching. Every product that still requires a user to learn its workflow is burning adoption rate against competitors that do not.

Reset 7: From "the tool has no context" to "the tool can see what I am doing"

Blank-slate products feel broken. Users expect the product to have context: what they were working on, what they have done before, what matters to them. A CRM that opens to an empty dashboard is failing the context expectation. A project management tool that asks the user to populate a project from scratch is failing the context expectation. Context is the new differentiator. The Writer survey identified that capturing and owning organizational context is the next competitive moat. Generic AI creates generic companies. Products that accumulate context with each interaction build a switching cost that competitors cannot replicate. Products that treat each session as a blank slate are training users to leave.

The silent churn problem

The expectation reset creates silent churn. Customers do not complain. They do not submit feature requests for "be more like an AI product." They simply stop logging in. They try a competitor that meets the new expectation. They downgrade. The product team sees a retention curve gently declining and attributes it to normal churn. It is not normal. It is the product failing expectations the team did not know had shifted. The Writer 2026 survey found that 84 percent of product teams are concerned what they build will not succeed in the market. The concern is justified. But the failure mode is not a dramatic launch collapse. It is a slow leak of users who found a product that meets the new expectations while yours still met the old ones.

Competitive dynamics of the expectation resets

The expectation resets create asymmetric competition. A startup with a thin AI wrapper can beat an incumbent with a decade of feature development because the startup meets expectation resets 4, 5, and 7 while the incumbent meets none. The incumbent has more features. The startup has fewer steps between intent and outcome. Customers choose fewer steps. I have watched this play out across CRM, analytics, project management, and customer support. In each category, the products gaining share are not the ones with the deepest feature sets. They are the ones that feel like AI products: instant, contextual, proactive, adaptive. The Gartner survey found that 80 percent of CEOs expect AI to force operational capability overhauls. The products that wait for the overhaul to be demanded will lose to the products that already shipped it.

Industry data on expectation shifts

The data on expectation shifts has accumulated rapidly. The LangChain survey found that 57.3 percent of organizations have AI agents in production, up from 51 percent the prior year. Every agent in production is training a user to expect agent-quality interaction from every product they use. The Deloitte report found that worker access to AI rose by 50 percent in 2025. More access means more exposure, which means higher expectations. The Writer survey found that 75 percent of executives admit their AI strategy is more for show than for substance. But users do not care about strategy. They care about the experience they had with the best AI product they used this week. That experience becomes the floor for every other product they touch.

How to audit your product against the resets

Take your product and score it against each of the seven resets on a three-point scale. Score 1: your product meets the old expectation. Score 2: your product is moving toward the new expectation but has not arrived. Score 3: your product meets the new expectation. Do this audit with real users, not internal stakeholders. Show them your product and ask them to complete a task. Then show them a competitor that scores a 3 on a specific reset. Ask which they would choose. If your product scores mostly 1s, you have 12 to 18 months before the market moves past you. If it scores mostly 2s, you have a window to close the gap. If it scores mostly 3s, you are setting the expectation rather than chasing it. The audit is uncomfortable. It is also the most valuable 90 minutes a product team can spend.

Getting started: the expectation audit

This week, pick one of the seven resets. The one where you suspect the gap between your product and user expectations is largest. Find three users. Show them your product. Ask them to complete the primary task. Time how long it takes. Count the steps required. Ask them what felt slow, what felt confusing, what felt like busy work. Then show them any AI product that meets reset 5, instant, or reset 7, contextual. Ask them to compare. Document the gap. Pick the single largest gap and run a one-week prototype to close it. Test the prototype with the same three users. Measure whether task time decreased and satisfaction increased. That is one cycle. Run it once per month. After six months, your product will meet expectations your competitors have not yet noticed have shifted.