The Counterintuitive Path to AI Adoption – O’Reilly

We are experiencing a peculiar moment in the development of AI. On the one hand, the demonstrations are spectacular: agents who reason and plan with apparent ease, models that compose original songs from a text messageand research tools that produce detailed reports in minutes. However, many AI teams find themselves stuck in “prototyping purgatory,” where impressive proofs of concept fail to translate into reliable, production-ready systems.

The data backs this up: a vast majority of GenAI business initiatives fail to generate measurable business impact. The central problem is not the power of models but a “learning gap” where generic tools fail to adapt to messy enterprise workflows. This echoes what I have observed in enterprise search, where the main obstacle is not the AI ​​algorithm but the fundamental complexity of the environment it must navigate.

This is magnified when building an AI agent. These systems are often “black boxes,” notoriously difficult to debug, whose performance degrades unpredictably when faced with custom tools. They often lack memory, struggle to generalize, and fail not because of the intelligence of the AI ​​but because the system around them is fragile. The challenge goes from perfecting indications to building resilient and verifiable systems.

What makes this particularly frustrating is the thriving “AI shadow economy” happening right under our noses. In many companies, employees quietly use personal ChatGPT accounts to get their work done. This disconnect reveals that while popular demand for AI is undeniably strong, the ambitious, top-down solutions being built are failing to meet it.

The strategic power of starting small

In light of these challenges, the most effective path forward may be counterintuitive. Instead of creating complex, all-encompassing systems, AI teams should consider drastically narrowing their focus; In short, think smaller. Much smaller.

This brings us to an old but recently relevant idea from the startup world: the “wedge.” A wedge is a highly focused initial product that solves a specific and painful problem for a single user or a small team, and it does it exceptionally well. The goal is to implement independent utility: to create something so immediately useful that an individual adopts it without waiting for widespread acceptance.

The key is not just to find a small problem but to find the right person. Look for what some call “hero users”: influential employees empowered to go off script and solve their own problems. Think about the sales operations manager who spends half their day cleaning lead data or the customer success leader who manually triages every support ticket. They are your AI shadow economy and they already use consumer tools because the official solutions are not good enough. Build for them first.

This approach works particularly well for AI because it addresses a fundamental challenge: trust. A wedge product creates a tight feedback loop with a core group of users, allowing you to build credibility and refine your system in a controlled environment. It’s not just about solving the cold start problem of networks, it’s about solving the cold start problem of trust in AI systems within organizations.

From passive registration to active agent

AI teams must also appreciate a fundamental shift in enterprise software. For decades, the goal was to become the “System of Record,” the authoritative database like Salesforce or SAP that stored critical information. AI has moved the battlefield. Today’s prize is to become the “Action System”, an intelligent layer that not only stores data but actively performs work by automating entire workflows.

The most powerful way to build is through what some have called a “data Trojan horse” strategy. You create an application that provides immediate utility and, in the process, captures a unique stream of proprietary data. This creates a virtuous circle: the tool drives adoption, usage generates unique data, this data trains your AI, and the improved product becomes indispensable. You are building a moat not with a commoditized model but with workflow-specific intelligence that compounds over time.

A concrete example is the “cluttered inbox problem.” Every organization has workflows that begin with a chaotic influx of unstructured information: emails, PDFs, voicemails. An AI tool that automates this painful first step by extracting, structuring and routing this information provides immediate value. By owning this critical top-of-funnel process, you gain the right to orchestrate everything afterward. You are not competing with the Registration System; you are intercepting their data flow, positioning yourself as the new operational center.

Look at a company like ServiceNow. has was positioned not as a replacement for core systems like CRM or ERP, but as an orchestration layer (an “action system”) that sits on top of them. Its core value proposition is to connect disparate systems and automate workflows between them without requiring costly “rip and replace” of legacy software. This approach is a masterclass in how to become the intelligent fabric of an organization. Leverages existing systems of record as data sources, but captures actual operational severity by controlling workflows. Defense is gained not by owning the core database, but by integrating data from multiple silos to deliver insights and automation that no incumbent can replicate alone. For AI teams, the lesson is clear: the value shifts from simply curating data to acting intelligently on it.

Building for the long game

The path from prototype purgatory to production involves a strategic approach. But as you build your focused AI solution, keep in mind that platform players are bundling “good enough” capabilities into their core offerings. Your AI tool should be more than a wrapper around an API; You must capture unique data and integrate it deeply into workflows to create real switching costs.

By adopting a wedge strategy, you gain the foothold you need to expand. In the age of AI, the most powerful wedges capture proprietary data while delivering immediate value, paving the way to becoming an indispensable Action System. This aligns with the core principles of building durable AI solutions: prioritizing deep specialization and creating moats through workflow integration, not just model superiority.

Here’s a tactical playbook:

• Adopt the single-player beginning. Before designing complex systems, create something that is immediately useful to a person.
• Target Hero users first. Find influential employees already using shadow AI. They have the pain and autonomy to be your advocates.
• Find your “cluttered inbox.” Identify a painful bottleneck in manual data entry. That’s your wedge opportunity.
• Design for the virtuous circle. Ensure daily use generates unique data that improves your AI performance.
• Become the Action System. Don’t just analyze data – actively complete the work and own the workflow.
• Choose reliability over capability. A simple, bulletproof tool that solves a problem well gains more trust than a powerful but fragile agent that tries everything.

The teams that will succeed will not be those that pursue the most advanced models. They will be the ones to start with Hero’s single user problem, capture unique data through a focused agent, and expand relentlessly from that beachhead. In an era where employees already vote with their personal ChatGPT accounts, the opportunity is not to build the perfect enterprise AI platform, but to solve a real problem so well that everything else follows.