Briefing #16: The AI You'll Never See
AI agents at work, behind the scenes
Note: This briefing was originally published on LinkedIn on November 7, 2025. It has been migrated to our new home on Substack to create a complete archive. Multi-format features like video and audio commentary are available for all new briefings published from April 2026 onwards.
For the last two years, the business world has been captivated by AI as a new front-end. Owing to the success of chat services like ChatGPT and Google Gemini, we’ve collectively focused on the chatbot as AI’s definitive application, treating AI’s ultimate purpose as a “knowledge worker” we can talk to.
This has been reinforced by recent news. There’s a new wave of agentic browsers from companies like OpenAI (who recently announced “Atlas”) and Perplexity (who earlier this year launched their own browser called “Comet”). These browsers let AI chatbots like ChatGPT sit alongside you as you browse the web. They can see the pages you’re browsing, help you do tasks on web pages, and even autonomously browse the web for you.
But this entire focus on the browser — the human-facing web — is missing the bigger picture. The web was built for human consumption. The real opportunity lies in the data and execution layer beneath it, which was never designed for humans in the first place.
This new autonomous “execution layer” shifts AI’s role from a “knowledge worker” that talks about work to a “strategic implementer” that silently does the work, enabling new customer experiences and enterprise opportunities.
Here are just three examples:
1. Beyond “Checkout”: The Rise of Agentic Commerce. We’re used to thinking of web transactions as a human clicking “buy.” With agentic AI, these transactions can now occur autonomously. Payments giants like Visa and Mastercard are building foundational “Agent Pay” infrastructure. This delivers the plumbing for AI agents to be registered, verified, and given the authority to act on a consumer’s or a business’s behalf.
Think of an agent that can be a virtual buyer, not just finding the best price for a complex B2B component, but negotiating the terms, executing the purchase order, and scheduling the logistics with the supplier’s agent. This represents a foundational new layer of commerce infrastructure, operating far beyond a simple “buy” button.
2. Beyond APIs: Intelligent System Integration. For decades, connecting core systems like an ERP and a CRM has been a brittle, expensive, and time-consuming process. It’s been the “dirty work” that cripples execution. Organizations have traditionally had to rely on rigid APIs and manual data mapping, hoping nothing breaks, and making data migrations and data transformation an enormous, capital-intensive endeavor.
Agentic AI reframes this entirely. Some organizations are finding success cutting multi-month (or even multi-year) integration projects down to initiatives completed in weeks. Underlying these integrations are AI agents capable of autonomously analyzing the data schemas of systems, mapping fields, and managing the flow of data. Instead of integration as a one-time process, AI agents are becoming the integration, adapting to context and changes in real-time.
3. Beyond Databases: Autonomous Data Migration. Ask any CIO about their biggest execution nightmare, and “data migration” will be near the top. It’s a high-stakes, high-risk process defined by manual validation and the constant threat of data corruption. This is perhaps the most powerful example of AI creating value “beyond chat.”
AI startup DualEntry boldly claims to use agentic AI to complete a full migration of a company’s legacy financial data to their platform in 24 hours, an expensive process that is usually thought of as taking months, or even longer. In agentic platforms such as DualEntry, AI agents are tireless “execution workers” that can validate, migrate, and remediate data at scale in ways that are hard to replicate with human teams alone.
Here’s the bottom line: there’s a web of possibilities for your organization that extend well beyond the AI chatbot.
As a starting act, the AI chatbot is a compelling parlor trick. But there’s a significant opportunity in AI’s next act, moving from AI that assists knowledge work to autonomous AI that solves the implementation and coordination challenges that used to only be possible through manual intervention.
That’s the real revolution. And it’s already happening in the plumbing.