What happens when Tom goes on holiday?
Why does documenting processes so often fail, and what can we do about it? A handy insight into how to make your next documentation project actually succeed.

You've had a new colleague for a few months now. She's settled in and capable. But when a customer question came in this morning, she asked Tom anyway. You know the one — the colleague who's been here for six years. He just knows the answers off the top of his head, after all.
Tom's answer was in the knowledge base too. But the product gets a lot of updates, so who's to say that answer was still correct? It seems far more sensible to just ask before you send the answer off to the customer. It only takes thirty seconds, and that way the answer is guaranteed to be right.
Tom answers. And the customer is helped. Everyone happy, right?
No, not quite. In the name of efficiency, you just missed a big opportunity. Because "just asking" works fine for everyone ... except for Tom. Because he's become a walking knowledge base, and the company is quietly leaning on him.
No doubt it was noted down somewhere that the knowledge base 'needs a quick update'. But in practice that doesn't happen, or rarely. That's because we're approaching it with the wrong mindset.
We talk about writing documentation as if it's a task with an end, something you "just need to get done at some point". You block out a couple of days for it and then it'll be finished.
But the moment you finish writing, that documentation already starts to age. Your product gets an update, or a process changes, or new exceptions crop up. Each of those things gnaws a little piece off your documentation, and not once do you get a notification saying "watch out, article 14 is no longer correct."
Keeping documentation up to date is a constant challenge. And really, it's in nobody's job description. Everyone on your team has their own job, their own to-do list to get through. Which is why documentation is always "something for later".
The result of that isn't a dramatic crash, but a small problem that gets worse step by step. Article by article, month by month, until you have a knowledge base that nobody really trusts anymore. Just asking is simply faster.
If this sounds familiar, you've no doubt already tried to appoint an owner. Someone with ownership of the documentation. That's a logical thing to do, but that person also has their own job. And for them too, the documentation will eventually slide down the priority list. You really have to free up a portion of that person's time for it. And the more often your products or processes change, the more time you have to set aside for it.
But that doesn't feel right, it meets resistance. So you schedule a periodic review. Just go through the entire knowledge base again every quarter. And that's certainly not a bad idea, but in a review you only look at what's already there. You don't look at what's missing. That one question that comes in every week, for example, but isn't documented anywhere yet.
You won't spot that during a review, because you can't review what isn't there yet.
Sarrai tracks every question the AI couldn't answer, every answer from a colleague that deviated from the documentation. And Sarrai AI simply lists it for you, complete with a suggestion for how to close that gap.
With other knowledge bases that information disappears completely; with Sarrai it's part of the strength of the system.
Getting the right answers to your customer is the easy part. You do that through Tom. The valuable part, the part that's normally lost, is making Tom's answer immediately reusable.
With Sarrai, every article that's missing, every absent nuance or exception is tracked and gets a solution proposed for it. Maintaining your knowledge base isn't a task you have to think up and schedule yourself, but something that presents itself.
That's exactly what Sarrai does with the approval inbox. Your customers' questions are answered based on your knowledge base, and whenever the AI doesn't know, it's handed off to a human. And that answer automatically becomes a suggested improvement.
That suggestion appears in your approval inbox and you approve it, edit it or throw it away. Where documenting used to mean nothing but 'write, write, write', it now just means 'review and approve'. Every shortcoming produces a ready-made suggestion. You no longer have to guess where the mistakes are. Sarrai AI is your digital colleague that writes along with you.
The difference is enormous. Because writing a new article is a real chore. But reviewing a suggestion is thirty seconds of work. Instead of scheduling a review of your documentation every quarter, you now block out half an hour every Monday to clear out your approval inbox. Just as fast as little mistakes creep into your documentation, the solutions present themselves. Sarrai is the knowledge base that maintains itself.
Nobody will be more grateful to you than Tom himself. The problem wasn't that a lot of knowledge sat with him. That's actually a richness, a colleague to be cherished. The problem was that the knowledge sat only with Tom. The next time Tom is interrupted because someone has a question, the answer will also automatically appear in your documentation a moment later. And next time Tom can simply keep working, instead of being interrupted.
A knowledge base that maintains itself makes the knowledge of your best people available to everyone on your team. Even when someone is away on holiday. Your best people are relieved of the load, so they can spend their time on the things that genuinely call for their expertise.
Start a free trial, upload your documents and watch the articles get written automatically. Or request a demo and we'll look together at whether Sarrai is a fit for you!