There is a version of AI adoption that most finance leaders are stuck in right now. You open a chat window, paste in some data, get an output, clean it up, move on. Useful, but you are still the one doing the connecting work. The AI is capable, but you are the operator.

I have been in that mode for a while, too. Then I started using Claude Cowork for tasks I used to handle manually or through regular chat with a lot of friction. And it felt different: I was not operating the tool; I was reviewing the work.

This edition is about what happens when AI stops being something you use and becomes something you delegate to.

In the free section, I walk through what Cowork actually is, five finance use cases that map directly to recurring work your team already does, and a short note on governance before you deploy it.

In the paid section, subscribers get the full tips and tricks for getting the most out of Cowork fast, plus a prompt library artifact with five ready-to-use prompts that have all the best practices built in.

Cowork is an agentic mode built into the Claude desktop app. You give it access to a folder on your computer, connect it to the tools you use, describe the outcome you want, and it plans and executes the work. It reads files, creates outputs, and delivers them directly to your file system. It checks in before anything consequential happens. You come back to finished work.

It is available now on Mac and Windows to all paid Claude plans (Pro at $20/month, Max at $100-200/month, Team, and Enterprise). It is still in research preview, which means it is improving fast, and some rough edges remain.

The best analogy for how to work with it: a new analyst. The first time you delegate something, you walk them through it step by step, check every output, and correct as you go. Over time, as you see how they work and what their output looks like, you extend more trust. Cowork works the same way. You start with full visibility into every step. You approve before anything consequential happens. You build confidence from there.

A few limitations worth knowing upfront.

None of these are dealbreakers for the use cases in this edition, but they are worth knowing before you build a workflow that depends on continuity or collaboration.

The pattern that makes Cowork useful: go to this folder, look at these files, extract or synthesize what is there, and produce an output that looks like this. That pattern maps directly onto some of the most time-consuming recurring tasks in a finance function.

Point Cowork at your vendor contracts folder. Ask it to rename files to a consistent convention (vendor name, contract type, expiry date), flag contracts expiring within 90 days, and produce a summary CSV. What used to take an afternoon of manual sorting takes minutes, with a clean output you can actually use.

Give Cowork access to a folder of receipts and PDFs. Ask it to extract vendor, date, amount, and category from each file and output a formatted spreadsheet ready for GL coding. It handles mixed formats and produces something your team can work with immediately rather than something you have to clean first.

Point Cowork at your close folder with the trial balance, variance notes, and last month's report. Ask it to compare current period against prior, pull the relevant numbers, and draft a variance narrative in the same format and structure as last month. It does not replace judgment. It eliminates the setup time before judgment kicks in.

Give Cowork access to three folders: your KPI data file, your CFO notes from the month, and the prior quarter's board deck. Ask it to draft the CFO commentary section using the same structure and tone as before. You still write the final version. You are editing a competent first draft instead of starting from a blank page.

Point Cowork at your standard contract template and a file with new client details. Ask it to populate the template, flag any fields that need manual input or legal review, and save the draft to your pending contracts folder. Straightforward for standard agreements, and it surfaces the exceptions rather than burying them.

Cowork stores conversation history locally on your device, not on Anthropic's servers. That matters for data handling, but it also means enterprise audit logs do not currently capture Cowork activity.

Before you point it at sensitive folders or deploy it across a team, establish which data classifications are appropriate, update your AI use policy to cover agentic desktop tools, and make sure your team understands that the review-and-approve step is the control point. Claude will show you its plan before executing anything consequential. That step is not a formality; take time to properly review and correct if needed.

We have covered what Cowork is and where it fits in finance workflows. The use cases are not hypothetical. They map to work that already exists in folders on your computer right now.

The difference between a frustrating first experience and a useful one comes down to setup and prompting. In the subscriber section, we get into exactly that: how to configure Cowork so it already knows your role and standards before you give it a single task, and how to write prompts that produce a usable first pass rather than something you have to redo. Plus, the prompt library is ready to adapt and use this week.

Cowork is still early. Some tasks will surprise you with how well they run. Others will need more steering than you expected. That is fine. The point is not to hand everything over at once. It is to start building a working relationship with a tool that can genuinely take things off your plate, and to do it with the same discipline you would apply to onboarding anyone new.

If you try any of these use cases this week, I would love to hear how it goes. Hit reply and let me know what worked and what did not.

 See you next Tuesday.

This newsletter is for you, and we want to make it as valuable as possible. Please reply to this email with your questions, comments, or topics you'd like to see covered in future issues. Your input shapes our content!

Want to dive deeper into balanced AI adoption for your finance team? Or do you want to hire an AI-powered CFO? Book a consultation!

Did you find this newsletter helpful? Forward it to a colleague who might benefit!

Until next Tuesday, keep balancing!

Anna Tiomina
AI-Powered CFO