A few years ago, I inherited a contract folder from someone who had left the company three months earlier. There were 312 agreements in it. The newest one was a vendor renewal that had auto-extended for another year, at a 7% price bump, twelve days before I sat down at the desk. Nobody had flagged it. Nobody had read it. Nobody even knew it existed until the invoice came in.

When I told my boss, she asked the question every manager asks in that moment: “How do we make sure this never happens again?”

I have heard a lot of answers to that question over the years. Lately, the answer is almost always some version of “we should put AI on it.” AI clause extraction. AI redline review. AI risk scoring. AI summaries of every PDF in the drive.

I am not against any of that. Some of it is genuinely useful. But if you are sitting on an inherited mess and someone tells you the first move is to deploy an AI review tool, they have skipped about four steps. The contract that auto-renewed at 7% did not need a language model. It needed an email on a Tuesday in March that said “Heads up, this thing renews in 60 days and Marcus owns it.”

That is what I want to talk about. If you are building contract management automation from scratch, or trying to fix something that is broken, you start with reminders, intake routing, ownership, and renewal alerts. AI review comes later, and it works better when the boring stuff is already in place.

Why teams skip the boring layer

The boring layer is not exciting to demo. Nobody schedules a leadership meeting to look at a calendar reminder. The vendors selling into legal and procurement know this, so the pitch deck opens with the AI demo, not with the part where you assign an owner to every contract.

There is also a real psychological thing going on. If you have inherited 300 contracts in a shared drive and you do not know what is in half of them, the idea of a tool that “reads them all for you in minutes” feels like a rescue. I get it. I have wanted that rescue. But the AI summary of a contract you do not own, with a renewal date nobody is tracking, just gives you a more articulate version of the same problem.

The other reason people skip this layer is that it sounds like it should already exist. Of course you have renewal reminders. Of course someone owns each vendor. Of course intake routes to the right reviewer. Except in most companies I have worked with, none of that is true. The Director of Procurement thinks Legal owns it. Legal thinks the business unit owns it. The business unit owner left in 2022 and was never replaced in the spreadsheet.

So before we talk about anything fancy, here is the order I would actually do this in.

Step 1: Get a list of every active contract in one place

This sounds obvious. It is not. At most companies under about 500 people, there is no single list. There is a folder. There is somebody’s spreadsheet. There is a procurement system that only tracks vendors over a certain dollar amount. There is a sales tools subscription that only Marketing knows about.

You do not need a CLM platform to make a list. You need a spreadsheet with these columns:

• Counterparty name
• Contract type (NDA, MSA, vendor SaaS, customer order form, lease, etc.)
• Effective date
• Renewal or end date
• Notice period for non-renewal
• Internal owner (a real human, not a department)
• Where the signed PDF lives
• Annual value, if you know it

That is it. Eight columns. If you can fill that in for the contracts that actually matter, you are already ahead of most teams I have worked with. Notice I am not asking you to do this for every NDA from 2014. Start with the contracts that cost money, generate money, or carry real obligations. Everything else can wait.

Step 2: Decide who owns each one

Ownership is the single most useful piece of data in contract management, and it is the one most companies get wrong. A vendor contract without a clear internal owner is a contract that will auto-renew, get breached, or get forgotten.

The rule I use: every contract has exactly one owner, that owner is a named person, and that person knows they own it.

That last part matters. I have seen spreadsheets where the “owner” column is filled in based on whoever signed the agreement four years ago. Half of those people no longer work there. The other half have no idea their name is in a cell somewhere.

When you assign an owner, send them an email. Tell them what the contract is, when it renews, what the notice period is, and what you expect them to do about it. If they push back and say it should be someone else, fine, update the cell. The point is to get to a state where every contract has a human who would feel bad if it auto-renewed without warning.

Step 3: Set up renewal reminders before anything else

This is the part I would do first if you only had one afternoon.

For every contract with a renewal or end date, set a calendar reminder. Two of them, actually:

• One at 90 days before the renewal or notice deadline, whichever is earlier.
• One at 30 days before, as a backstop.

If you have a contract with a 60-day notice period and a January 1 renewal, your first reminder fires around early October. That gives the owner time to actually do something: renegotiate, cancel, switch vendors, or just confirm they want to keep it.

You can do this in Outlook or Google Calendar. You can do it in a shared spreadsheet with conditional formatting. You can do it with a free Zapier or Make scenario that emails the owner when a date approaches. None of this requires a CLM platform. None of it requires AI. It requires somebody to sit down for half a day and set up reminders.

I once worked with a small ops team that ran their entire renewal program off a Google Sheet and a Tuesday morning email script. They handled about 180 active contracts that way for two years before they moved to a real system. They never missed a renewal in that time. They were not impressive. They were just consistent.

Ironclad has a decent write-up on why reminder software matters, and I will tell you the punchline is the same as mine: missed renewals are not a tooling problem, they are a “nobody got told in time” problem.

Step 4: Build a real intake process

The other thing that breaks contract management long before AI can help is bad intake. Bad intake looks like this: a salesperson Slacks the General Counsel directly with a Word doc and the subject line “quick one.” A vendor sends an MSA to whoever they have an email address for. Marketing signs a click-through SaaS agreement at 11pm because the trial expired.

Once any of that happens, the contract enters your environment without a record, without an owner, and without a review. You cannot automate something you do not know exists.

A workable intake process for most teams looks like:

• One front door. A shared inbox, a form, or a Slack channel where every new contract request lands.
• A small triage rule set. NDAs go to a template, vendor agreements under $X go to a standard review path, anything custom or over $X goes to a named reviewer.
• A note in the request that captures the business owner, the counterparty, the rough value, and the deadline.

The form does not need to be sophisticated. Google Forms works. A simple Jira project works. I have seen a single Notion page work fine for a team of 40. The point is that nothing enters the contract pipeline without someone seeing it and routing it.

If you build this layer, you will be amazed at how much “AI review” stops feeling urgent. Most of the review pain people complain about is not the review itself, it is the fact that contracts arrive in random channels at random times from random people, and nobody can tell which ones are important.

Step 5: Then, and only then, think about AI

Once you have a list, owners, reminders, and intake, you have an actual system. At that point, AI tools start being genuinely useful, because they have something coherent to work on.

AI is good at things like:

• Pulling key dates and clauses out of a stack of signed PDFs you already track.
• Flagging unusual language in an incoming draft you have already routed.
• Summarizing a 40-page MSA so the business owner you already assigned can actually read it.

Notice the pattern. AI works on top of the boring layer. It does not replace it. If you run AI clause extraction on a folder where 30% of the documents are duplicates, expired, or unsigned drafts, you get a beautiful report about garbage. The model is not wrong. The inputs were never sorted out.

The teams I have seen get real value from AI review are the teams that already knew which contracts mattered, who owned them, and when they renewed. The AI made those people faster. It did not make a broken process work.

What to do this week

If this article hit close to home, here is what I would do in the next five business days, in order:

1. Pull a list of every contract you can find. Do not try to be exhaustive. Hit the top 50 by spend or risk.
2. For each one, write down the renewal date and the notice period. If you cannot find them, write “unknown” and flag it.
3. Assign an owner. A real human. Tell them.
4. Set 90-day and 30-day reminders for every renewal you do know.
5. Pick a single intake channel for new contracts and tell people to use it.

That is the contract management automation that actually pays off. It is unglamorous, it is mostly free, and it will save you from the kind of inherited 7% auto-renewal that started this article.

The AI part can wait until next quarter. The reminder for the renewal in 60 days cannot.

---

I’m Dave, and I write about contract management the way it actually works. No jargon, no sales pitch, just what I’ve learned from 15+ years of doing this job. If this was useful, stick around.