Hello Reader,

It's 4pm on the last Friday of the month. You need to finish the revenue forecast by the end of the day. The only thing missing is the contract terms for the three new deals. And Kevin, the only person who knows where everything is, just went on vacation.

I know this because I have lived this. At my last company, we had two people maintaining two separate contract trackers. When either of them was out, we were stuck.

Month-end close got delayed. The board got angry. And we looked incompetent because we could not find basic data in our own contracts.

Here’s what I don’t like. You are spending probably 15+ hours per month on this, doing it manually. And you are still finding errors, and making adjustments all the time it feels like.

What if I told you AI can do this in 15 minutes?

Extract every payment term, every milestone, every renewal date from your contracts automatically. So when Kevin's away, you do not care.

Here's how...

The Bible Problem

In my last company, we had something called ‘The Bible’ The place where we stored contract details (you might have a different name for this).

What was worse. Every department had their own version of ‘The Bible’, and it drove us all mad.

We were all tracking the same contracts, but everybody had different fields, different update schedules, and different versions of the truth. Sales cared about commission. Legal cared about termination clauses. Finance cared about payment terms and milestone schedules.

So everyone just created their own tracker. And nobody talked to each other about it.

This was painful for me, especially, because I am a numbers guy, and I like to see everything in one place.

And this is the crazy part: Within Finance alone, we had TWO separate Bibles. One person maintained the sales and margin analysis tracker. Another person maintained the payment terms and milestones tracker.

Different people needed different views, but, instead of figuring out one system, we just created another Excel file.

Double the manual work.

Double the chance of errors.

Now, think about what happens when you need to close the books.

You're trying to validate deferred revenue, confirm billing schedules, and reconcile customer balances. But first, you need to track down which version of which Excel file has the right information. Then you need to hope it's actually up to date. Then you need to cross-reference it with the actual PDF because you don't trust fully the manual entry.

This is called manual contract administration, and it doesn't work.

It's inefficient. And it will be completely unnecessary in 2026.

And this isn't just an internal finance problem. WorldCC research shows that typical contract value leakage is 8.6% of contract value. This is not a rounding error. This is real money you are losing because nobody can find the payment terms, milestone triggers, or penalty clauses when they actually need them.

AI Contract Intelligence

Here's the better way: You teach AI to read your contracts once, and then it reads every contract forever.

Plus, you have a lot of options:

1. Use off-the-shelf tools like ChatGPT, Copilot, Gemini, etc.
2. Use a dedicated tool like Maxio that is specially designed for revenue
3. Build a custom automation using things like Document Intelligence in Azure

By the way, Maxio have released a new AI update that uses MCP (Model Context Protocol - more on this later) to execute tasks, like generating revenue schedules, detecting churn risk, and producing charts for board decks. You can read about this here.

Whatever option you choose, it works because: AI doesn't get tired, doesn't make typos, and doesn't go on vacation.

You give it a contract PDF, and it pulls out exactly what you need based on the instructions you give it. Think of it like that junior team member we always talk about, except this one can process 50 contracts in the time it takes you to drink your coffee.

But really important to understand here: This isn't about replacing humans. It's about eliminating the manual, repetitive parts so your team can focus on the analysis and decision-making. AI extracts the data, you validate it, and then you use it.

And the best part? Once you set this up, it scales. Whether you have 10 contracts or 10,000, the process is the same.

Your AI Contract Intelligence Roadmap

Okay, so how do you actually build this? Let me break it down into four steps you can start with this week.

Step 1: Audit Your Current Contracts

First, you need to understand what you're dealing with.

Get a sample of 10-20 contracts and list out every data point your team currently tracks. Payment terms, milestones, renewal dates, termination clauses, etc. Then ask yourself: Where does this data live right now? Excel? CRM? Email? Kevin's brain? Write it all down.

This gives you your baseline and shows you exactly what needs to be extracted.

Step 2: Define Your Contract Schema

Now decide what data points you actually need AI to extract. Don't just copy everything from your current tracker—there's probably a lot of rubbish data in there.

Ask yourself: "What information directly impacts revenue recognition, cash forecasting, or customer analysis?"

Create a simple table: Data Point | Where It Lives Today | Where It Should Live | Priority (High/Medium/Low).

Step 2.5: Create a Single Source of Truth for the Whole Company (Optional)

Here, you have an opportunity to showcase your business partnering skills.

Find out what other departments need. Add their high-value data points to your schema.

Really quick - Ask Sales, Legal, Operations etc: "What contract data do you track manually right now?"

Add those fields to your extraction.

By doing this, you become the contract data owner for the entire company. One database that Sales, Legal, Operations, and Finance all trust. You have no more conflicting versions of the truth.

And when you present this to leadership? You are saving not just Finance 30 hours per month. You are saving the company 100+ hours across all departments.

This is the business case that gets you budget and a seat at the strategic table.

Step 3: Test with Your Preferred LLM (Right Now)

Before you build anything fancy, validate that AI can actually extract what you need. Take one contract PDF, upload it to ChatGPT, Copilot, Gemini, or Claude, and use this prompt:

"Extract the following data points from this contract: [your list]. Return the results in a table format with columns: Data Point, Value, Page Reference."

Did it work? Great. Did it miss something? Refine your prompt and try again. Spend 30 minutes on this. If AI can do it in a chat interface, it can do it at scale.

And here's the key trick: Once you've got it working consistently, ask the AI:

Save that prompt. You're going to need it.

This is my best kept secret that still nobody is using yet ;)

Step 4: Build It Into Your Workflow

You've validated that AI can extract your contract data. This is good.

Now you need to decide how to integrate this into your actual process.

You've got two routes:

Path A: The Fastest Setup (No IT Required)

This is about creating a repeatable extraction process using AI tools you can set up yourself in 30 minutes.

Step-by-Step Setup:

Option 1: Custom GPT (ChatGPT Plus or Enterprise)

1. Go to ChatGPT → Click "Explore GPTs" → "Create a GPT"
2. Paste your system prompt from Step 3 into the instructions
3. Name it "Contract Data Extractor"
4. Done. Now anyone on your team can upload a contract PDF and get consistent extraction every time.

Option 2: Gemini Gem (Google Workspace)

1. Open Gemini → Click the star icon (Gems)
2. Create New Gem → Paste your system prompt
3. Name it "Contract Extractor"
4. Save. Now you have a dedicated AI agent for contract extraction.

Option 3: Copilot Agent (Microsoft 365 Users)

If you use Microsoft 365, this is your fastest automated setup.

1. Open Microsoft Copilot→ Create New Agent → Name it "Contract Data Extractor"
2. Under Knowledge Sources → Connect to SharePoint "Contracts" folder
3. Paste your system prompt from Step 3 into Instructions
4. Test: "Extract payment terms from Contract ABC.pdf"

Option 4: Claude with Projects + MCP

1. Create a Claude Project → Add your system prompt as project instructions
2. If you want Claude to access contracts directly from Google Drive, just enable the Google Workspace connector in your settings
3. Upload contracts or point to Drive folder
4. Claude extracts data following your prompt consistently

A note on MCP:

You don’t need to understand the complexities of MCP when using Google Drive, or any connector. You just need to know what you can connect. Full list of Claude connectors here.

This week, in the masterclass I was giving to the AI Finance Club members, one of the fractional CFO’s participating explained how after one training session, he connected Claude to all of his 25 years history of deliverables. Via the Claude MCP, he is using this now to make (and these are his words) “A client deliverable that would take a week in just a few hours.”

Quick video on Quickbooks & Claude Below for you as well here:

​

The Workflow:

1. New contract signed → PDF saved to your contract folder
2. Team member opens your Custom GPT/Gem/Copilot Agent/Claude Project
3. Uploads contract (or references it if using MCP)
4. AI extracts data into table format automatically
5. Team member validates output (check totals, dates, names)
6. Copy-paste validated data into your central Excel/Google Sheet tracker
7. Mark contract as "Processed" in your log

Time per contract: 15 minutes instead of 90 minutes

Where data lives: Your existing Excel or Google Sheets tracker (no new systems required)

Path B: The Industrialized Route (Full Automation, IT Partnership Required)

This is where you eliminate manual uploads entirely with workflow automation.

Step-by-Step Setup:

Microsoft Stack (Power Automate + Azure Document Intelligence)

1. IT sets up Azure Document Intelligence with your contract schema
2. Create Power Automate flow:
   • Trigger: When new file added to SharePoint "Contracts" folder
   • Action 1: Send PDF to Azure Document Intelligence
   • Action 2: Extract data using your schema
   • Action 3: Write extracted data to Excel Online (or your ERP API)
   • Action 4: Send Teams notification to Finance: "New contract processed, please validate"
   • Action 5: If validation approved → Lock row in Excel
3. Finance team member reviews extracted data in Excel, approves or flags exceptions
4. Done

Google Stack (Apps Script + Document AI)

1. IT sets up Google Document AI with your schema
2. Create Apps Script automation:
   • Trigger: When new file added to Google Drive "Contracts" folder
   • Action 1: Send PDF to Document AI
   • Action 2: Extract data using your schema
   • Action 3: Append to Google Sheets
   • Action 4: Send email notification: "New contract ready for validation"
3. Finance reviews in Sheets, marks "Validated" column as TRUE
4. Done

Alternative: Zapier, Make.com or n8n (Platform Agnostic)

1. Connect your cloud storage (Dropbox, OneDrive, etc.)
2. Trigger: New PDF in "Contracts" folder
3. Action: Send to ChatGPT API or Claude API with your system prompt
4. Parse response and write to your Excel/Airtable/Database
5. Send Slack/Email notification for validation
6. Done

The Automated Workflow:

1. Contract signed → PDF saved to designated folder
2. Automation triggers instantly
3. AI extracts all data points
4. Data flows directly into Excel/ERP
5. Finance gets notification: "Contract XYZ processed, validate here: [link]"
6. Finance reviews extracted data, approves/rejects
7. Approved data becomes locked source of truth

Time per contract: Zero ongoing manual work (after IT setup)

Setup time: 2-4 weeks with IT

What you show IT: "Here's my ChatGPT test showing 95% accuracy. We process 50 contracts/month at 45 minutes each = 37.5 hours/month saved. This pays for itself immediately."

What This Actually Looks Like in Practice

The company Honeywell integrated their contract, financial, and customer data systems. And now, they're projecting up to $50 million in working capital unlocked just from reducing cycle times. That's not ‘AI hype’, that's cash flow improvement you can take to the board.

PKF O'Connor Davies implemented AI for data extraction from PDFs and reported 30-40% time savings per person, per day. Think about what your team could do with an extra 3 hours every day.

Here’re some more use cases.

The Bottom Line

You remember when everyone decided that they were going to go paperless with all documents?

Now we need to get to the point where we are ‘humanless’ when it comes to recalling information like this.

For sure, a human needs to validate - But they don't need to waste hours they could be using becoming better business partners asking “what contract is this most up to date, and where is it?”

Because (and this is super important) contracts touch everything. Revenue recognition, cash forecasting, customer analysis, compliance. All of it starts with contract data. And if that data is stuck in PDFs or living in Kevin's Excel file, you are missing a big opportunity with your revenue process.

AI Contract Intelligence fixes this.

• Audit your current contracts
• Define your contract schema
• Bring in other departments (optional)
• Test extraction with your choice of LLM
• Choose your path (Simple Q&A or Full Automation)
• Build validation into the workflow

Your Move

So here's what I want you to do this week: Take 30 minutes and find five contracts. Upload them to AI and test the extraction. See if AI can pull the payment terms, milestone schedules, and renewal dates accurately.

If it works, you have just validated that you can eliminate manual contract administration.

Then decide: Do you want the quick win with an AI chatbot processing your files? Do you want to use a dedicated tool like Maxio, or are you ready to go full automation with IT?

You do not need a team of data scientists to make this work. You just need to be willing to stop doing things the hard way.

And remember. You didn't become a finance leader to search for PDF contracts. This is your chance to get that time back (for you and your team) and build a faster, more accurate revenue process.

​

Best,

Your AI Finance Expert,

Nicolas

​

​P.S. - What did you think of this newsletter? Did you find this useful and practical? Hit reply, and let me know (I read all replies)

P.P.S. - We already have 2,000 confirmed guests for the biggest AI finance event on Tuesday; the seats are limited, so hurry up!
​The link to reserve your ticket