Rovo.
How big companies use AI:
I recently wrote how AI made me go from solo to having 8 employees.
Yes, AI made me hire more people, not less.
And people reacted violently:
“Ruben, you don’t work in a big organization.”
They are right. I don’t have 100 colleagues.
So I had to check what was going on. Benchmarking the best ones.
I keep receiving messages from you about “How to best use Copilot?”
So I took a look at Microsoft Teams and Copilot. And I’m clueless.
Other LLMs (ChatGPT, Claude, Grok, Gemini…) are just much better at writing, searching, thinking. And maybe that’s ok?
Because they are looking for different things.
People need collaboration, team workspace, making sure to have the entire context of your company without doing any sketchy things (like uploading your entire client’s legal doc into an LLM).
They want to partner with this many humans (yes, that’s how I call your colleagues, I know) on an AI workspace to:
Use AI collectively.
Navigate the team workspace better.
Let’s take the #1 reason people work together. Brainstorming ideas.
Well, we already have a first (big) problem.
AI is terrible at being creative:
1. AI is terrible at being creative.
When I think of “AI team workspace”, I think of brainstorming.
But AI is terrible at being creative.
And to some of you, it’s obvious. Because you experienced it too: you ask for 10 ideas, and none of them actually match your unique use case.
I made a deep dive into this paper, “Divergent-Convergent Thinking in Large Language Models for Creative Problem Generation” on X*.
*X is the platform where I’m a bit more technical than Linkedin.
Here’s the kick summary. Instead of asking:
"Give me creative ideas for [problem] with these requirements."
You split it into two steps:
Step 1 (Divergent):
“Brainstorm 10 wildly different ideas about [topic]. Ignore all requirements. Push for unusual, surprising concepts.”
Step 2 (Convergent):
“Select idea [X] and develop it to meet these constraints: [requirements]. If it doesn’t work, try another idea.”
You’re getting closer to doing better brainstorming work.
The right prompt triggers the proper response.
But now comes the second problem you keep running into: you have to:
(1) manually import the right context from your team workspace to ChatGPT
(2) and then copy and paste the relevant part into your team workspace,
(3) all while you are NOT supposed to copy and paste anything into ChatGPT from your company.
I pity you, really.
I honestly had no idea how a big corporation worked in 2026.
But I then met Atlassian’s team (the company behind Rovo) at a dinner in SF.
I told them about my experience with Jira and Confluence back at Trade Republic in Berlin. A young 24-year-old, completely clueless about how behave at work:
Now, back to that SF dinner, Atlassian told me their tools evolved.
There is this new AI. So they gave me access to it, on a workspace.
They paid for this partnership with one simple idea: I’m testing what it’s like to use Rovo at a big corporation and sharing my honest review with you.
So is Rovo any good, or just a “toothpaste AI”?
archive of past newsletters: https://docs.google.com/document/d/1pWuMCBVQo1zKcgKltX_BZxAr31KgxmOlp3Vzvmc5Hxc/edit?usp=sharing
2. Toothpaste AI or useful AI?
I am always skeptical about AI in big companies.
Because it always feels like this meme:
So when Rovo’s team asked me to try, I was skeptical too.
But here’s my first vibecheck (it’s like 4 minutes or so):
I started with the common culprit. When I was a (very young) social media manager, I always felt dumb asking, “where is that doc?”.
I lost so much time and trust, just not asking where the document I needed was.
That’s my first “aha moment” of this Rovo.
You can… ask it, in a very big knowledge base.
Think ChatGPT, but for your internal knowledge sitting inside Jira & Confluence (that’s like 300,000 organizations say AI).
What surprised me:
✦ It finds the right page, ticket, or decision fast.
✦ It answers the question, then shows the exact info.
✦ Turns “I think we agreed on” → “here’s the source.”
If your team has Confluence + Jira (and years of data), this is the difference between guessing & knowing. An AI that is useful from day 1.
No adoption needed.
And I think this is the key: the speed of adoption.
3. Speed Is All You Need.
If I were managing a 100-employee company, that’s the only metric that matters.
How fast my company adopts AI for a useful impact.
Not a shiny AI tool doing cool gimmicks for a week.
But an AI you will actually use, being useful, saving a couple of hours a week.
A quick do’s and don’t.
DO’S…
☑ Master the task before asking an AI to support you.
☑ Know both the limits and possibilities of AI (before even trying).
☑ Company’s culture is to play as much as possible, quickly, to determine whether or not it’s actually doable.
DONT’S…
✖ Use AI because it’s cooler than not using it.
✖ Keep repeating the same task over and over again, without considering AI.
✖ Ask for “where is that doc?” without checking on the company’s AI inside their knowledge base.
I see two challenges, really:
1 - Having a culture that supports playing, trying & experimenting with AI.
2 - Having the right infrastructure (like ROVO with Confluence/Jira) paired with the right directories of guidelines. For example, how can your employee ask an AI where a doc is if the doc does not exist? You can no longer afford not to keep track of everything on one massive company-sized database.
The context window of these (AI) tools keeps getting greater. It means AI is more and more capable of consuming and understanding your entire company’s knowledge without making any mistakes.
And this trend won’t stop.
I remember GPT-1 had a 500 token limit. That’s like 3-4 paragraphs of English.
Now GPT-5.2 has a 400,000 token limit. So 800x more in a couple of years. That’s like a 2-3 Harry Potter books.
So you can bet tools like ROVO, owning your entire team workspace, with the complete knowledge database, are becoming increasingly vital.
You can’t skip building on it.
If you struggle on the “How” to implement it at your company, leave a comment and I will personally DM you. I might be able to help my most active subscribers.
— Humanly yours, Ruben
PS: I received TONS of positive reviews on my article to go from 49 to 10,000 followers on Linkedin in 17 days. Read it if you’re into growing your Linkedin.
Documentation mess just happens when smart people move fast for years.
You can't out-process entropy.
The companies waiting around to "organize everything properly first" are getting lapped by teams who just accepted the mess and plugged in something like Rovo to make sense of it anyway.
Perfect systems are for the ideal world. Searchable mess is actually achievable.
I have been using Jira for over 20 years, and we added Confluence about 6 years ago. Unfortunately, I have found Rovo to be both annoying and unhelpful. Rovo's prompts on the pages frequently interrupt the flow of my work on Jira and Confluence pages, getting in the way of what I need to get done, and disrupting my concentration rather than adding value.
Getting support from Atlassian has long been a weak point, with the company relying heavily on the Atlassian Community Forums, where users bitch, share complaints, and attempt to help one another troubleshoot Atlassian errors, bugs, and changes that were not thoroughly thought out before they were implemented.
Rovo does nothing to address the support shortcoming. Instead, Rovo feels like yet another attempt by Atlassian to shirk their responsibility to properly support paying customers.