For advisors and their clients, the market action may be noisy, but the underlying question it surfaces is crucial: Will a given software company become the AI orchestrator of workflows, or just one of the pipes that orchestrator uses?

What the Selloff Was Signaling

To understand the fear, it helps to translate what these new agents actually do. Rather than living inside a single app as a “copilot,” Claude’s Cowork tools, built on its upgraded Opus 4.6 model, act more like an AI project manager sitting on top of many systems. Industry-specific plug-ins can draft contracts, update CRM records, generate presentations, and route tasks across tools, often with minimal user clicking. In an AI‑native workflow, the human talks to the agent; the agent talks to the software. The owner of that agent effectively owns the workflow.

The AI Project Manager Has Entered the Building

Some software platforms are explicitly trying to become that orchestration layer. Salesforce, for example, markets Agentforce as an AI agent that can understand intent and execute tasks across multiple clouds and external systems, not just surface recommendations in a single module. ServiceNow is investing in “Agentic Playbooks” that connect IT, HR, and finance workflows so AI agents can act across them with clear guardrails. These are early but concrete examples of large incumbents positioning themselves as the AI front‑door through which work flows.

Other tools may look more vulnerable. CNBC reporters with no coding experience recently used Claude Code to build a fully functioning Monday.com clone in under an hour for less than $15—a live demonstration that rattled investors and helped send the stock down more than 20% alongside a guidance miss.

Many niche SaaS tools still rely on users manually working through multiple screens and forms to deliver their value (e.g., an expense-management app). In a world where an AI agent can call their APIs behind the scenes, those tools could risk being treated like interchangeable plumbing: necessary, but with less brand visibility, weaker pricing power, and limited direct user engagement. The investor worry is that a growing slice of the software stack could be commoditized this way, even if usage technically remains high.

Orchestrators vs. Plumbing: The New Dividing Line?

For RIAs talking with clients, the practical question may not be, “Is this company using AI?” Almost everyone is. The more important question may be, “Where does this company sit in an AI‑first workflow?” Does management have a credible path to becoming an orchestration layer with agents that initiate and complete tasks across systems? Or, is the business still essentially a click‑heavy point solution waiting to be abstracted away?

None of this means “SaaS is dead.” At this point, Gartner and others expect agentic AI to automate an increasing share of routine, task‑level work more than replace core systems of record outright. But it does suggest a sharper sorting mechanism within software: Platforms that successfully orchestrate AI‑driven workflows are more likely to compound value; tools that are merely orchestrated will have to fight much harder to avoid commodity pricing.

One Way to Think About Software Exposure From Here

As you review software exposure with clients in the wake of the Claude‑driven selloff, framing holdings along this “orchestrator vs. plumbing” axis may help separate AI panic from genuine structural change.

“You only need to get burned by illiquidity once: Flexibility becomes one of the most valuable features in portfolio construction.”

一 Timothy Reilly, President

Private Credit Under the Microscope

As headlines spotlight private credit liquidity limits and potential credit stress, RIAs are reassessing how they size, fund, and diversify exposure. AFA’s 2025 RIA Private Credit Survey shows how peers are actually allocating, which vehicles they prefer, and how they’re planning next moves.