Hello fellow keepers of numbers,

We’re kicking off tax season with some fun news in the AI tax prep space with a partnership between Juno and K1x. Anthropic opens up Claude in Excel to Pro plan users, and I’ve included a cool demo of it here. Gallup releases their 2025 Q4 AI in the workplace results. Plus, Anthropic launched apps in Claude, which is cool, maybe?

THE LATEST

K1x and Juno announced a new integration that sends K-1 data directly from K1x into Juno’s tax workflows. K1x is an AI-powered tax data operations platform that specializes in extracting and validating complex tax data from forms like K-1s, K-3s, and 1099s, while Juno is a modern tax prep and review platform built by CPAs. The goal is to remove manual K-1 data entry and clean-up work from the tax prep process.

With the new integration, firms can route K-1 PDFs into K1x, which extracts and validates both core and supplemental data with “field-level precision,” then feeds that data straight into Juno workpapers and into connected tax software. The integration is designed to cover the full K-1 workflow from document intake through return preparation in a single system, reducing review discrepancies and rework. Early users report that work that used to take hours now takes minutes, particularly for K-1-heavy clients.

K1x highlights that its platform is already used by 20 of the top 25 U.S. accounting firms and over 40,000 organizations, positioning this as enterprise-grade tech now wrapped in a small-firm-friendly workflow through Juno. The integration isn’t fully open yet. Firms can join a waitlist through Juno to be notified when it becomes generally available and get access to an introductory webinar.

Why it’s important for us:

This is a really cool partnership that could make a bit of noise in the tax world. K1x is a large, mature software used by a lot of the top firms, which makes this news a much bigger deal for Juno.

K-1s can get extremely complicated, and even K1x has some limitations. There are definitely places where this process can break down: K1x could miss information on a complex K-1, K1x’s integration with Juno, Juno’s integration with the tax software. It remains to be seen how well this’ll work end-to-end. But the promise is great, and even if it works only moderately well, it’s likely still going to be a massive time-saver.

K1x has some other competitors in the space: Additive (owned by Thomson Reuters) and Abacus are two that come to mind. Both offer integration with tax software, so this could’ve been a move made by K1x to better compete in the space long-term.

AI-powered tax prep software is also gaining some steam. Black Ore is one of the hot new software in the space, and Juno is obviously making moves now as well. There are quite a few others to watch as we progress through 2026.

Most firms have already made plans for their 2026 tax busy season. If I had to, I’d predict that 2027 will be a major adoption year for AI-powered tax prep software and supporting tools like K1x, Additive, Abacus, or others.

Anthropic expanded its Claude in Excel beta to all Claude Pro, Max, Team, and Enterprise plans. Claude in Excel is an add-in that adds a chat sidebar to Excel so the model can read, explain, edit, and create workbooks directly inside the spreadsheet, instead of copy-pasting back and forth.

The integration is powered by Claude Opus 4.5 and is designed to preserve formulas and workbook structure when it makes changes. Users can ask Claude to debug formula errors, update assumptions across a model while keeping dependencies intact, build new sheets, or fill existing templates. It also supports pivot tables, charts, and cell-level citations so you can click from an explanation straight to the referenced cells.

Claude in Excel includes some guardrails like “ask before edits” and optional session logging, which can create a “Claude Log” tab that tracks the actions it takes in a workbook. Anthropic also warns about prompt-injection risks in untrusted spreadsheets and suggests users proceed with caution.

Why it’s important for us:

This is a long-awaited promise of AI. Creating and editing Excel schedules with AI has been possibly the most requested feature since AI was handed to us normies.

Claude in Excel is seriously powerful. It’s light-years ahead of where Microsoft’s Agent Mode in Excel is currently. And Microsoft hasn’t given us much reason to believe that gap will be narrowing anytime soon.

Opus 4.5 is the smartest model in the space right now for most things, including Excel. Claude in Excel is creating multi-tab workbooks with complex formulas and formatting with only a simple prompt. It can also answer questions about extremely complicated workbooks, and can even help create macros to set up automations inside workbooks.

Claude has officially built a lead over other AI providers. And they’re clearly the leader for the accounting and finance industries. Firms that haven’t decided on an AI provider yet should be choosing Claude.

Gallup’s Q4 2025 workplace survey found that 26% of U.S. employees now use AI at work at least a few times a week, up three points from Q3, with daily use rising from 10% to 12%. Overall AI adoption, people using AI at least a few times a year, held flat, and nearly half of workers (49%) still say they never use AI in their role. Organizational integration is also stalling: only 38% report their company has implemented AI tools, 41% say it has not, and 21% aren’t sure.

Use is heavily skewed toward knowledge industries. Technology reports 77% as AI users (daily, weekly, monthly, or annually), while finance and higher education follow closely at 64% and 63%, respectively. Professional services reports 62% as AI users, including 36% frequent and 16% daily users. Finance and professional services saw some of the largest quarter-over-quarter gains.

Why it’s important for us:

The AI user base is growing, but it’s an important reminder that still only about 26% of U.S. employees use AI on even a weekly basis. Sometimes the social media algorithms and the friends and colleagues we’re around every day inundate us in the space and make it feel more widespread than it actually is.

Finance and professional services AI users have grown very steadily over the last two years. Approximately 19% of finance employees report using AI on a daily basis, much higher than the 12% for all U.S. employees.

This highlights the value the finance industry is finding in AI. The jump in daily users shows they’re finding use cases. I suspect client communication is the most common across the board, followed closely by research.

At the same time that the AI user base is growing, AI governance is lagging. An earlier KPMG study reports that around 44% of U.S. workers are using AI tools in unauthorized ways, and about 46% have uploaded sensitive information to public AI platforms. It’s clearly time for companies to expand their focus on AI governance and roll out AI policies, if they haven’t already.

PUT IT TO WORK

With Claude in Excel rolling out to Pro plan users, it has instantly become much more accessible to millions of people. It’s quite possibly the most powerful tool ever released in Excel.

Here’s a demo of it using a GL to create a fully reconciled income statement and executive summary:

Claude in Excel Demo - Watch Video

WEEKLY RANDOM

Anthropic just announced that Claude can run interactive apps directly in the chat window. Asana, Slack, Box, Canva, Figma, and more. ChatGPT has been doing something similar with its integrations.

The broader trend: AI providers want to be the place where you actually do work, not just where you ask questions and then leave. The more they can keep you inside the model, the better for them. It’s probably better for users too. No need to switch between apps as often and copy and paste.

It’ll be interesting to see where this goes. It reminds me a bit of the early App Store. There were limited options, and people weren’t sure what to do with it. Over time, it built a platform that enabled new business categories. We could be in the same early phase here.

I’m honestly not sure where this is headed long-term. But businesses should pay attention and figure out how to build things that live inside AI models. Being “AI-embedded” might be the new “mobile-friendly”.