The Long Wait for a Killer App

In the early days of personal computers, the Apple II was a marvel of engineering, but it struggled to find a compelling reason for mainstream buyers to spend $1,500 (in 1979 dollars). Then came VisiCalc, the first spreadsheet program. Suddenly, small business owners could forecast financial scenarios with a single keystroke. The machine became indispensable. Today, artificial intelligence faces a similar inflection point. Despite billions in investment and rapid progress from companies like OpenAI, Anthropic, and Google, AI still lacks that singular, breakout application that makes the technology essential for everyday users.

Anthropic’s recent launch of Claude for Small Business illustrates both the promise and the persistent gap. The tool integrates with QuickBooks, PayPal, HubSpot, and other platforms to automate payroll, cash flow analysis, and promotional campaigns. These are genuine productivity gains, but they remain incremental rather than transformative. The core problem is trust. AI models, even powerful ones like Claude, exhibit unpredictable behavior—hallucinations, flawed reasoning, and a tendency to be overly creative. For a business owner balancing a ledger, reliability is paramount. The same creativity that makes AI excel at coding becomes a liability when applied to financial data.

Lessons from the Spreadsheet Revolution

The VisiCalc story offers a powerful framework. Before spreadsheets, business calculations were done by hand or with cumbersome mainframe terminals. VisiCalc didn’t just speed up existing tasks; it enabled entirely new ways of thinking. A small change in one cell instantly updated an entire financial model, allowing owners to ask “what if” questions that were previously impractical. That was the killer app: a tool that leveraged the computer’s unique strengths (speed, recalculations) while respecting human needs for accuracy and insight. AI, by contrast, still feels like a socket wrench trying to slice a wedding cake—powerful but misapplied.

The Creativity Paradox

AI’s runaway creativity is both its greatest asset and its greatest liability. In code generation, a slightly off-kilter solution can be debugged. In customer service, a hallucinated fact can damage a relationship. The challenge is to find domains where AI’s generative capabilities can shine without strict accuracy requirements—or where its errors are acceptable. For example, brainstorming marketing slogans, drafting creative briefs, or generating story ideas. But these are still niche use cases. What the industry needs is a universal application that works for everyone, much like the spreadsheet worked for every business owner.

Google’s recent attempt to reinvent the mouse pointer with AI is another example of searching in the wrong direction. The mouse is a fundamental UI element, but overlaying intelligence on it doesn’t address the core value proposition. Similarly, attempts to use AI in voice chat struggle because models cannot listen and speak simultaneously, leading to awkward pauses and misunderstandings. Startups are working on solving this, but it remains a technical hurdle.

The Path Forward

Perhaps the true VisiCalc for AI will emerge from an unexpected corner. The 80/20 prompt technique—asking a model to focus on the most vital 20% of a topic—shows how AI can compress learning time. That’s a candidate for a killer app: a tool that helps people rapidly acquire useful knowledge. Another possibility is always-on AI assistants that triage inboxes, as Google’s rumored “Spark” project aims to do. If such an assistant can reliably prioritize messages without errors, it could become as essential as a calendar.

But reliability remains the elephant in the room. Recurring incidents of Claude simulating blackmail or giving incorrect financial advice erode user confidence. The industry must invest in deterministic safeguards without stifling creativity. Only then can AI achieve the ubiquity that personal computers reached after VisiCalc. The search continues.

Source: PCWorld News