Anthropic's new plugin changes how advisors should think about AI
Anthropic's new Wealth Management plugin does meeting prep, plan drafting, and TLH screens out of the box. The question for advisors has shifted.
An analysis:
In February 2026, Altruist launched AI tax planning inside Hazel, the feature meant to position it as an AI operating layer for independent advisors. Just 60 days later, Anthropic released 10 ready-to-run agent templates for financial services, including a Wealth Management plugin for Claude that accelerates meeting prep, generates pitches, and assists with portfolio analysis and tax-loss harvesting screens out of the box. Each template is a reference architecture built from three parts: skills (task instructions and domain knowledge), connectors (governed data access), and subagents (specialized pieces of the workflow). The skills are file-based Markdown, customizable, and free to install from GitHub.
The announcement was framed as a new tool for advisors. What it actually is, is a useful lens for evaluating everything else in your stack.
What Anthropic’s wealth management plugin can do for financial advisors
Anthropic's Wealth Management plugin sits inside Claude and connects analytical workflows across an advisor's existing systems. It integrates with market data providers, CRMs, and internal document repositories to help advisors centralize information and automate portions of client prep and portfolio analysis.
To understand what changed and what didn't, it helps to think about the advisor tech stack in three layers:
- Interface—dashboards, forms, alerts, reports, chat.
- Intelligence—interpreting client data, drafting recommendations, surfacing planning opportunities, prioritizing work. This is the layer vendors have been packaging and selling. It is also the layer foundation models do natively, with a few lines of context.
- Execution—opening accounts, moving money, placing trades, managing tax lots, rebalancing, harvesting losses at scale, supervising activity, maintaining the auditable record.
AI is moving fastest through the first two. The third is different. Execution requires balance sheet, broker-dealer infrastructure, regulatory standing, and engineered systems that act on real client accounts under fiduciary control. No foundation-model release replicates that. It is not a software problem.
“The right frame for evaluating your advisor tech stack is not AI versus non-AI. It is analysis versus execution.”
A model can identify a tax-loss candidate. A platform with brokerage, custody, and tax-lot awareness can implement tax-aware management at scale. A model can draft a rebalance memo. An execution layer can place the trades. A model can summarize a tax return. It cannot become the tax professional, the custodian, the broker-dealer, or the supervisory record.
The strategic question is no longer “will AI replace software?” It is: Which features are becoming AI-native utilities, and which systems still create durable value because they control data, workflow, governance, and execution?
Why AI commoditization changes how advisors should evaluate technology
The obvious objection is that software providers can package these AI capabilities into polished advisor workflows faster and more efficiently than most individual firms could build on their own. And that’s likely true. It doesn’t undercut the argument, it reinforces it.
The intelligence layer is being commoditized whether firms build internally, buy third-party software, or combine both approaches. What’s changing is that the underlying AI capabilities are becoming increasingly portable across platforms and providers. Portability changes how advisors should think about software differentiation. Analytical workflows—summarization, planning drafts, portfolio screens, document review—can now move relatively quickly between tools, plugins, and model providers.
What does not move nearly as easily is the infrastructure layer underneath. Custody, brokerage operations, compliance frameworks, supervisory systems, and automated execution capabilities are significantly harder to replicate or replace. Those systems require operational scale, regulatory infrastructure, and years of workflow development.
The asymmetry is significant. As AI capabilities continue to spread across the industry, the most durable parts of an advisor tech stack are likely to be the systems responsible for execution—not just analysis.
What most firms can realistically build
The tools to assemble a basic advisor-AI workflow are now genuinely accessible. A motivated firm can stand up meeting prep, client review packets, document Q&A, draft emails, plan summaries, drift reviews, and tax-loss candidate screens using off-the-shelf plugins and customizable skill files. The first layer isn't that hard to build.
Sustaining it is a different problem.
The skills that make an AI workflow actually useful—investment philosophy encoded as text, tax-review processes, compliance language, approval rules—drift the moment regulations shift, markets move, or the firm's own process changes. Keeping them current isn't a one-time project. It requires rigorous, ongoing supervision to ensure the AI's outputs consistently align with the firm's fiduciary obligations, shifting regulations, and market dynamics.
If and when a firm does not have that person, the workflow can degrade quietly until it stops reflecting how the firm actually operates.
This is the same reason durable platforms invest heavily in the infrastructure underneath the AI layer. Execution capabilities—moving assets, managing tax lots, rebalancing at scale, maintaining supervisory records—require operational depth that compounds over time, not a one-time build. The firms best positioned to combine intelligent AI workflows with reliable execution likely aren't assembling it themselves. They're building on platforms that have already done the operational work.
The AI layer will keep getting easier to access and cheaper to run. What doesn't get easier is the regulatory infrastructure required to act on what the AI surfaces. That gap is where durable value lives.
The takeaway
The advisor tech stack isn’t disappearing. It’s being reorganized.
Packaged software still matters, and for many firms, buying will be smarter than building. But advisors should be increasingly skeptical of software whose primary value is packaging generic AI workflows inside a polished UI. Those capabilities will likely become cheaper, more customizable, and more portable.
Durable value will accrue to systems that do what AI alone cannot: connect trusted data, support compliance, execute transactions, implement portfolio decisions, and power a reliable client experience at scale.
Two questions are worth applying to every line item in the stack:
- Is this tool primarily generating, drafting, summarizing, or formatting information? That layer is commoditizing quickly.
- Is this platform deeply connected to data, governance, workflows, and execution? That layer is more durable.
The pace of change matters. Within a single quarter, capabilities that once looked highly differentiated became accessible enough for motivated firms to assemble themselves. And the next wave of commoditization is likely to happen even faster.
The firms best positioned for that shift are the ones building on durable operational infrastructure while treating AI as an accelerant for advisor productivity—not the foundation of the business itself.
The most durable advisor technology platforms are likely to be the ones that combine intelligent workflows with the infrastructure required to actually execute on them.
The key question for advisors is simple: What in your stack is truly durable, and what is ultimately just a UI layered on top of a foundation model?