A level-by-level walkthrough of what to learn, which tools to try, and how to keep leveling up your accounting team's AI fluency.
Level 1: AI Skeptic
You Google everything. You write every document manually. You think AI doesn’t apply to your specialized accounting work. Here’s the hard truth: your "uniqueness" is a bottleneck. AI can standardize the repeatable parts (reconciliations, commentary, data pulls) so you can focus on what truly requires your expertise.
What to Do Next
- Replace every Google search with an AI chatbot like Claude or ChatGPT
- Open a reconciliation in Excel and ask Claude in Excel to review it and suggest ways to streamline or templatize it
- Commit 30 minutes daily to experimenting with AI on real accounting tasks
- Make it a team sport: start a weekly meeting where everyone shares AI wins, failures, and learnings
Expected Impact: Save 2–5 hours weekly. Stop feeling left behind.
See It in Action: How AI Lets Accountants Leapfrog Years of Technical Learning
Modern accountants have a unique advantage: AI tools let you skip years of technical learning curves. Instead of mastering advanced Excel formulas like SUMIF, VLOOKUP, and INDEX MATCH or learning VBA from scratch, you can use AI to solve complex problems directly inside your workbooks with context-aware assistance.
- Skip the traditional learning curve for advanced Excel and VBA
- Get instant, context-aware help instead of hours on Stack Overflow
- Gain 3 to 5 years of efficiency improvement starting today
Recommended Tools
Claude and ChatGPT are the most accessible entry points for accountants. Both are secure (paid plans don't train on your data), handle accounting questions well, and can draft variance explanations, summarize financials, review journal entries, and answer research questions in plain English.
How to start: Sign up for Claude or ChatGPT Teams and use it as a research tool. For example, ask it to explain ASC 606 and what it means for your revenue recognition. Pro tip: tell it to explain it like you're five so it puts complex standards in plain English. Don't stop at one answer. Have a back-and-forth conversation, ask follow-up questions, and dig deeper until you actually understand it.
Watch out for: Always use the Teams or Pro versions of Claude and ChatGPT to ensure the model is not trained on your data. Be careful when inputting confidential client information. Neither tool knows your chart of accounts, your business context, or your prior period numbers unless you tell it, and the output is a starting draft that needs your professional judgment.
Claude in Excel brings AI directly into the tool accountants already live in. Instead of copying data between apps, you get context-aware assistance right inside your workbook. It understands your specific data, formulas, and structure.
How to start: Install the Claude for Excel add-in. Open a workbook you use regularly, like a reconciliation or depreciation schedule, and ask Claude to explain a complex formula, suggest improvements, or help build a new calculation based on your existing data.
Watch out for: Claude in Excel works best when your workbook is well-structured with clear headers and labels. Messy or unlabeled data will produce less useful results. Start with a clean, organized workbook to see the best output, but honestly, we've seen it do amazing things with unstructured data too.
Curious what this looks like applied across the close cycle? See how our outsourced accounting team uses AI-assisted reconciliations and variance commentary to close books faster for high-growth companies.
Level 2: AI Curious
You use AI occasionally: some flux commentary here, a research question there. You ask, it answers, but it's not part of your process yet. No templates, no shared prompts, no systems. You’re seeing glimpses of value but AI is still an afterthought in your day to day processes.
What to Do Next
- Use AI for 10 different accounting tasks this week (flux commentary, prepaid schedules, JE review, meeting notes, etc.)
- Enable AI notetakers on your next meeting and review the auto-generated summary and action items
- Build a shared prompt library with your team so your best prompts become everyone’s best prompts
- Track your time savings across every AI task and build the business case for going deeper
Expected Impact: Save 5–10 hours weekly. Start seeing real efficiency gains.
See It in Action: Modernize Your Existing Workbooks with AI
Don't build AI tools from scratch. Start with a workbook you already have, like a fixed asset roll forward, and ask AI to recommend improvements. In minutes, you can replace hard-coded values with dynamic formulas, add data validation to prevent input errors, auto-generate journal entries, and create a user guide for your team.
- Turn manual, hard-coded workbooks into dynamic, formula-driven tools
- Add data validation and auto-generated journal entries with a single prompt
- Instantly create documentation for knowledge transfer and team continuity
- Start with simple weekly team touchpoints to build the AI habit
Recommended Tools
AI notetakers in Microsoft Teams, Zoom, or tools like Abacor automatically capture meeting transcripts, action items, and summaries. No more scrambling to take notes during client calls or status meetings. You get a searchable record of every conversation with key takeaways pulled out for you, so you can fully focus on the meeting at hand.
How to start: Enable the built-in AI notetaker in Microsoft Teams or Zoom for your next client meeting. After the call, review the auto-generated summary and action items. You can also try dedicated tools like Abacor for more detailed transcription and search across all your meetings.
Watch out for: Always let participants know the meeting is being recorded and transcribed. Some clients may have policies against recording. Check with your team and clients before enabling AI notetakers on sensitive calls, but this is quickly becoming standard operations across most companies.
At this level, Claude and ChatGPT become your daily drivers. Use them for drafting flux commentary, researching accounting standards, building variance tables, writing multi-entity consolidation narratives, summarizing meeting notes, and reviewing close checklists where format consistency matters.
How to start: Start with your month-end close. Take your P&L variance analysis along with your GL detail, feed it into AI, specify the threshold (>$5K or >10%), and ask it to draft flux commentary formatted for your management report template. Build reusable prompt templates for recurring tasks like account reconciliations, journal entry reviews, and intercompany eliminations. The key is consistency: use AI on the same tasks every close cycle so you can refine your prompts over time.
Pro tip: Once you have a prompt that works well, turn it into a custom GPT or Claude Project so you don't have to re-prompt every time.
Watch out for: Don't just use AI for the easy stuff. Push into tasks you think are 'too specialized' for AI. You'll be surprised how well it handles accounting-specific work with good prompts. Try asking it to construct a prepaid amortization schedule from an invoice or check your existing debt amortization schedule for errors. The more context you give it about your business, the better the output.
At Level 1, you used Claude in Excel to review a single workbook. Now it's time to go deeper. Use it to overhaul your core close workbooks: replace hard-coded values with dynamic formulas, add data validation to prevent input errors, and build auto-generating journal entries like we showed in the fixed asset roll forward video above.
How to start: Pick your most error-prone or time-consuming close workbook. Open it with Claude in Excel and ask it to audit the formulas, flag any hard-coded values that should be dynamic, and suggest data validation rules. Then ask it to build a journal entry section that auto-populates based on your close month selection.
Watch out for: Always review the formulas Claude suggests before sending them off to your reviewer. Test with a copy of your workbook first, not the live version. Once you're confident, roll the improvements into your actual close process.
As AI becomes part of your day-to-day, the next bar is making sure outputs are technically sound. Our technical accounting specialists help teams build prompts and templates that hold up under audit scrutiny.
Level 3: AI Builder
You’ve noticed patterns. You’re doing the same prompts repeatedly. Time to automate. At this level, AI acts as the glue between your apps. Your accounting tools talk to AI. AI talks to other tools. You’re not in the middle anymore. The system runs itself.
What to Do Next
- Pick your most repetitive accounting task and build an AI workflow to automate it
- Turn your best prompts into Custom GPTs or Claude Projects so your team can use them without prompt engineering
- Connect your accounting tools (email, ERP, spreadsheets) to AI using tools like Zapier or n8n so data flows without you in the middle
- Build a knowledge base by uploading your SOPs, memos, and close checklists into tools like Lindy, Relevance AI, or Microsoft Copilot Studio so your team can ask an AI agent questions instead of waiting on the one person who knows how something works
- Document every workflow you build, share it in your weekly AI meeting, and iterate as a team
Expected Impact: Save 10–20 hours weekly. Close faster. Do more with the same team.
See It in Action: End-to-End Workflow Automation for Financial Chat Messages
This demo walks through a Level 3 workflow that automates complex tasks by processing financial chat messages using n8n. A webhook waits for a new message (e.g., from ClickUp), parses the content, sends it through an AI agent with prompts for technical subjects like ASC 606, converts the output to markdown, chunks it to handle message length limits, and replies automatically in the original chat. The real power is stitching workflows together to create end-to-end automated processes that standalone chat windows like ChatGPT simply can’t do.
- Webhook-triggered automation that listens for incoming chat messages
- AI agent processing with specialized prompts for technical accounting standards
- Automatic markdown conversion and chunking for message length limits
- End-to-end workflow stitching that goes beyond what standalone chatbots can do
Recommended Tools
You’ve been copy-pasting the same prompts every close. Custom GPTs and Claude Projects let you package your best prompts, instructions, and reference documents into a reusable tool that anyone on your team can use. No code required. Think of it as turning your prompt expertise into a shared team resource.
How to start: Take your most-used prompt (e.g., flux commentary, JE review, or memo drafting) and turn it into a Custom GPT or Claude Project. Upload your templates, specify the output format, and add instructions so it behaves consistently every time. Share it with your team and iterate based on their feedback.
Watch out for: These tools are only as good as the instructions you give them. Be specific about output format, tone, and what context the tool should ask for before generating a response. Review outputs regularly as your processes evolve and update the instructions accordingly.
These platforms connect your accounting tools (QBO, Xero, HubSpot, email) to AI without code. Build workflows where client emails get auto-summarized, reconciliation data gets pre-analyzed, or close checklist items get auto-populated. Zapier is the easiest to start with, Make.com offers more complex multi-step visual workflows, and n8n gives you full control with a self-hostable open-source option.
How to start: Pick a trigger (new email, new spreadsheet row, form submission) and build a workflow that sends the content to an AI chat for analysis, then routes the output to your team. Start with something simple like auto-summarizing client emails. Map out a manual process you do every close and build it as an automated scenario with AI steps in between.
Watch out for: Automation workflows can get complex fast. Start with one simple automation, get it working reliably, then build the next. Don’t try to automate your entire close process in week one. Each platform has a different learning curve, so pick the one that matches your team’s comfort level and scale from there.
These platforms let you build AI agents that handle multi-step tasks autonomously, like processing a batch of invoices, categorizing expenses, or generating standardized client reports. Lindy.ai is the easiest to get started with, Relevance AI lets you build teams of agents that work together on complex processes, and Microsoft Copilot Studio plugs directly into Teams, SharePoint, and the rest of the Microsoft ecosystem most accounting firms already use.
How to start: Identify a multi-step process that currently requires you to babysit it (e.g., receiving vendor invoices via email, extracting key data, categorizing, and logging into your system). Build an agent to handle the full flow. If your firm runs on Microsoft 365, start with Copilot Studio for the tightest integration with your existing tools.
Watch out for: AI agents can make mistakes that compound through a workflow. Build in checkpoints where a human reviews output before the next step executes, especially for anything client-facing or financial-statement-level work. Start with one agent on one process before building a multi-agent system.
Stitching workflows together across tools is where most teams stall. Our accounting transformation team designs and ships these automations end-to-end so the system actually runs without you in the middle.
Level 4: AI Transformer
You’re not just using AI tools. You’re building them. Custom micro-apps, custom tooling, AI-powered internal solutions. AI isn’t making you faster. It’s making you different.
What to Do Next
- Prototype using no-code tools (Lovable, Replit) and get a working version in days, not months
- Identify one manual process that could become a custom micro-app for your team using Cursor or Claude Code
- Test with your team, refine based on feedback, and iterate until it sticks
- Connect directly to your ERP or financial systems via API and build a custom dashboard that surfaces insights your team would otherwise spend hours compiling manually
Expected Impact: Build tools no one else has. Become the team everyone wants to work with.
See It in Action: Building a Custom Dashboard That Replaces PowerBI and Tableau
This demo walks through a custom-coded internal dashboard built as a more flexible alternative to traditional tools like PowerBI or Tableau. It integrates directly with API endpoints to consolidate data from financial software, time tracking systems, and CRM pipelines into one view. Using Claude Code, new data sources and features can be added in minutes, not weeks.
- Complete control over how data is displayed and organized, no vendor limitations
- Analyze profitability by project or company with live data from your ERP
- Interactive deal pipeline with direct links to external platforms
- Rapid iteration with Claude Code to add new data sources on the fly
Recommended Tools
These platforms let you build functional web apps by describing what you want. Perfect for prototyping internal dashboards, analysis tools, and team workflows without a development team.
How to start: Pick a spreadsheet your team uses heavily (a close tracker, client reporting template, etc.). Describe it as an app to Lovable or Replit. You’ll have a working prototype in hours that you can iterate on.
Watch out for: No-code tools are great for prototyping but may hit limits at scale. Plan for a migration path if the tool becomes mission-critical. Also consider data security and ensure any client data stays within your security perimeter.
Cursor and Claude Code let you build custom tools with AI assistance, even if you’re not a developer. Describe what you want in plain English and the AI writes the code. Build custom reporting tools, analysis dashboards, and internal workflow apps.
How to start: Think about a tool you wish existed for your accounting workflow. Open Cursor, describe it in natural language, and let the AI scaffold it. Start small with a single-purpose tool that does one thing well.
Watch out for: AI-generated code needs testing and review. Don’t deploy anything client-facing without thorough testing. Start with internal tools where mistakes have lower stakes, then graduate to client-facing applications.
Direct API Connections
For maximum flexibility, connect directly to AI APIs (OpenAI, Anthropic) from your own systems. Build custom analysis pipelines, automated reporting, and AI-powered internal tools that give your team a proprietary edge.
How to start: If you have technical resources (or use Cursor/Claude Code), start by building a simple API integration that takes data from your ERP and runs it through an AI model for analysis. Begin with a non-critical workflow to learn the patterns.
Watch out for: API costs can add up with high-volume usage. Monitor your spend. Also, you’re responsible for data security when using APIs directly, so ensure you understand the provider’s data handling policies and your own compliance requirements.
Building custom AI tools for your accounting team is the highest-leverage move you can make. If you want a partner on it, request a demo and see how Zeroed-In ships custom dashboards and micro-apps for finance teams. For more stories and frameworks, browse our Zeroed-Insights blog.
