Interesting new entry in the Claude connector directory: Aiwyn Tax (formerly Column Tax) is live as of yesterday. Sounds niche at first, but it’s actually a pretty sharp look at where Anthropic is steering its connectors ecosystem.
What does it actually do?
The connector plugs Aiwyn’s tax engine directly into Claude. According to the announcement, users can:
- run basic US 1040 calculations
- play through scenarios and what-if cases
- upload W-2s and other tax documents
- generate finished tax forms as PDFs
Under the hood sits an engine that has processed over a million real-world 1040 returns and is validated against thousands of scenarios via an internal framework called TaxCalcBench. That’s a lot more depth than asking Claude to reason through the tax code in its head.
Why it’s more than a nice integration
Aiwyn’s founder Michael Bock dropped a pretty loaded claim on X: “This is the last year anyone has to pay for TurboTax.” That’s marketing, sure — but the trend is real. Specialized calculation engines are migrating into chat interfaces and turning into part of the conversation, instead of being their own separate app.
Strategically, this matters for Anthropic. Connectors like Microsoft 365, Google Workspace, DocuSign, LegalZoom — and now Aiwyn — all point the same direction: Claude as a control surface for daily work, not just a chatbot. Tax in the US is a brutally competitive, heavily regulated space. Pulling in a vetted partner rather than trying to own it themselves is smart division of labor.
My take
For those of us outside the US, a 1040 connector isn’t immediately useful. What’s interesting is the pattern: external partners bring verified calculation logic, Anthropic brings the interface and the trust. If this model works, we should expect a similar wave for ELSTER, DATEV, or European DocuSign alternatives. The moment someone ships a connector there, things get genuinely exciting.
Until then, it’s a nice illustration of how the Claude ecosystem is growing in 2026: quietly, through partners, rather than through big model announcements.
Sources: