Claude Needs a Source of Truth (And So Do You)

Before you connect AI to anything, decide where each type of information live, and never let it live in two places. Alane Boyd and Micah Johnson tackle the biggest mistake teams make when implementing AI tools like Claude.

If you've ever wondered why your AI gives inconsistent answers or pulls outdated information, this episode reveals the culprit. The hosts get real about why downloading files and re-uploading them to AI projects creates chaos.

In this episode, you'll learn:

• How to establish a single source of truth that works for both humans and AI
• Why duplicate files kill AI accuracy before you even realize it's happening
• The knowledge layer approach to connecting AI with your existing systems
• Real examples of what goes wrong when information lives in multiple places
• How to properly connect Claude to SharePoint, Google Drive, and other storage without creating copies
• Why standardization isn't optional when AI tools are part of your workflow

If you're ready to stop making AI guess where to find information and start giving it access to the right source, this episode shows you exactly how.

Tools/Platforms Mentioned

• Claude
• SharePoint
• Google Drive
• Dropbox
• Box
• PM Tools: ClickUp, Asana, Monday.com
• Grain
• Microsoft 365
• Ignite
• Rippling
• Pipedrive
• Obsidian
• Trainual

Learn More about our AI Leadership Workshop

Winning businesses aren’t just working harder, they understand how to use AI strategically. Sign up for the AI Leadership Workshop, a 3-hour expert-led mastermind designed to give leadership teams clarity, alignment, and a practical path to stay competitive.

Book a call to customize your workshop today!

This episode is brought to you by Biggest Goal.

Every quarter your team spends evaluating AI is a quarter your competitors spend shipping. Most leaders feel the pressure but get stuck between ignoring AI and getting it wrong. More tools and more demos won't fix it. What actually works is hands-on training for the people doing the work.

Our workshops align your leadership team on a shared AI roadmap. Our 4-week cohort get your teams building real AI agents, and automation workflows. And our ongoing groups keep leaders and builders sharp long after the cohort wraps.

Upcoming Sessions:

• Claude Cowork Masterclass (Beta)
• Claude Deep Dive Series

Connect with Us:

• TikTok
• LinkedIn
• YouTube
• Community

Disclosure: Some links are affiliate links, meaning we may earn a commission at no extra cost to you. Thanks for supporting the podcast!

For more information, visit our website at biggestgoal.ai.

Transcript

Micah Johnson (00:01)
Before you connect anything to AI, decide where each type of information actually lives and never let it live in two places. You need a single source of truth for both humans and AI to work from.

Alane Boyd (00:17)
You know, Micah, what we love to do on this podcast is we bring up topics that we're talking about already every day with people, whether it's just casual conversations or potential clients. And so this one that we're going to be talking about today with a single source of truth is something that both you and I have been talking about in conversations so much lately. And it's not something

new. It's something that's been around for a long time. That was a big focus of our previous company GoFanbase that we ran together. But now with AI, I'm seeing this become more of an issue because people want to save documents in their SharePoint or their Dropbox, but then upload them to individual projects in something like Cloud Cowork.

Micah Johnson (01:00)
I mean, I think anytime a topic like this enters our quote unquote casual conversations, it's more than like a big issue. This is like I'm seeing an immediate issue. Like, pay attention to this now because you could write this off before. You could go out and say, All right, well, whatever. We can double check stuff and we can look in multiple folders and Sally

in marketing, she'll find it for us. Not a big deal. When we're now implementing tools like Cowork, when we have Skills that need a place to go to find stuff, this becomes an urgent problem. This is a big deal. If you don't have a source of truth for AI to look for things, you're forcing it to make a judgment call. If you've listened to any of our previous episodes, you know, don't make AI make judgment calls. And so, yeah.

I don't know. I don't even know how to make this more of a bigger deal.

Alane Boyd (01:57)
Other than us talk about that, it's a big deal. But yeah. So let's just take a second and and we talk about a single source of truth as like our language. Like we understand what we're talking about. But for somebody that maybe hasn't really thought through what a single source of truth is in their department or their organization, what that means is it's a single place that any team member can go to get the most updated version and information.

Micah Johnson (01:59)
Or that we do a whole episode on it.

Alane Boyd (02:24)
That could be Google Drive, that could be something like a Trainual where it's got an SOP that you've got or a How-To. So it's the place that people know to go to to get an answer.

Micah Johnson (02:35)
I like what you said just there, Alane, because we do need to expose the fact that a lot of times sources of truth, you think about SharePoint or Google Drive or you think about a database. But the reality is now that tools like Cowork can connect to your apps, the source of truth could be Monday.com. It could be ClickUp. It could be Grain. It could be name any platform you have it connected to.

Alane Boyd (03:01)
Mm-hmm. Well, and I guess when we think about a single source of truth too, like even when I was just saying, you know, my explanation of it is that it's the single source of truth for that type of information. Because we have some documents in Dropbox and we have some documents in Drive, but our team needs to know what goes where, and that's what we do communicate. But from HR, it could be Rippling.

Micah Johnson (03:15)
Mm-hmm.

Alane Boyd (03:27)
You know, but you know where to go for that single source of truth. And the nice thing about Cowork that we're gonna get into is that it can connect with that source of truth and then you can operate from it.

Micah Johnson (03:38)
Yeah. So to clarify and put it plainly, it is a standardized single location where you get the source of whatever information you need.

Alane Boyd (03:49)
That's not confusing, but yes, that's true. So with what happens that I'm seeing in my conversations, and Micah, I want you to add your perspective on this. What I'm seeing is, and I'm just gonna use just a general explanation. What I'm seeing. Companies use SharePoint as their file storage. Fantastic. You wanna save your files in there, you've got some type of

Micah Johnson (03:51)
Dang it. I tried. I tried. We can edit that part out.

Alane Boyd (04:14)
Folder hierarchy, you've got some internal documents, you've got external documents for clients and things that you're saving. Fantastic. That is your single source of truth for that information. Your team is trained on it. You understand where those things are. But what I'm seeing happen is, well, maybe they want to interact with those files and ask it questions. So instead of connecting Cowork to share file and asking

Cowork to go and look up those files and ask those questions. They're downloading those files, uploading them to an individual project or an individual task so that they can work from that.

Micah Johnson (04:49)
Mm mm. I already

don't want to give my opinion on this. It's disgusting.

Alane Boyd (04:53)
So

I'm just like freaking out and I don't hold it in. I'm like, you're creating a nightmare. Whether you're a big company or a small company, you now have two places to update information. And I promise you, humans are terrible at remembering. We have an entire episode on humans are bad at consistency because they are, and they're not going to remember when a new document is updated.

Micah Johnson (05:01)
Mm-hmm.

Alane Boyd (05:20)
Or when a document is updated to update it in the other place.

Micah Johnson (05:24)
Seriously. And then you forget you put it in another place. So then you put it in another place. Now you have three different sources that it could pull from. But getting back to the original, and it all comes down to standardization. One, I hate that idea, Alane. Don't download it and then upload it into a project. Just don't do that. Connect Cowork to SharePoint

where it exists so that you can access the original documents. I think a thing to point out in that is you can have even duplication inside of that. Think about naming conventions or versioning. Well we got version one, version two. If you forget to name it correctly or you don't have naming conventions for your files inside your source of truth, if it's a folder cloud storage, for example, well then how are you going to know which one is the right one without opening it up and reading it? So there's

even nuances on sources of truth that it is worth spending the time to say this is how we're going to do it. And don't let people dilly dally and be well, I've always done it this way, or I name tough noodles. This is how we're naming it moving forward.

Alane Boyd (06:34)
It it it makes a huge difference. Micah, I've got a related, unrelated example with this because we love to blame the AI when it gets it wrong. But the truth is a human did it wrong in the first place, and I am the wrong human in this instance. So I know I'm gonna give an example. I was so pissed this week because I was writing a to help me with some sales stuff.

Micah Johnson (06:52)
⁓ okay. My ears just perked up.

Alane Boyd (07:01)
And to review the pipeline in Pipedrive. And I told you, this shit just is not working. It pulled a deal, it pulled a deal from eight years ago. It was literally our very first client at this company together. And I'm like, this is just garbage. And you're like, Alane, you didn't tell it which pipeline to look at. So it guessed, and it was the

first pipeline that we used that is still in there, but we don't look at it because we just look at this other one. You didn't remember that that pipeline existed. Cowork looked at that and said, this looks great. And I'm like, gosh, you're right. The human is wrong. I did not tell it that. And this absolutely translates to anything that we're given it, and especially in our file storage, because it has so much content and history in it. And

the AI is just going, well, I'm just gonna guess, just like an intern at your company would if you don't give it enough information or if you're disorganized in your system.

Micah Johnson (08:05)
Yeah. And I mean, one of the things that I was going to say is how do you create that source of truth? Well, it's creating Skills and it's leveraging automation and it's doing all the stuff that we talk about on this podcast to help create those standards and enforce those standards so that it's not a human saving the file, naming the file, using the naming convention, remembering where to put it, et cetera. Automation and AI can do that now. So you create the Skills, you create the automation, you put that all in.

When you're creating those things, you still have to go through and think through all this to say, am I creating it in the right place? Am I giving it enough context? Am I testing it to get all this done? I yeah, excellent point, Alane.

Alane Boyd (08:46)
So let's talk through some ways that you know a company can use a single source of the truth with something like Cowork so that they're not downloading and uploading files into projects. So yeah.

Micah Johnson (08:58)
Yeah, first of

all, if you're if you're listening to this and you're downloading anything and then uploading it and you're using Cowork, you're using Cowork wrong. Go back to I don't even know what episode it was, but it's The Future of Work: Claude Cowork. And then listen to every episode forward from there. And I promise you it'll change your outlook on this because you should not be doing it that way anymore.

Alane Boyd (09:24)
No, you shouldn't. and I'm just laughing, Micah, because you are so passionate about this. And I understand why, but I hear it. I know that's that's why we wanted it to be a topic because it is something that we're talking about multiple, multiple times a day. And and the root of it, Micah, that you just said is that people are using Cowork like chat. They're not using Cowork like their coworker, and it's evident in hearing things that

Micah Johnson (09:29)
I know. I can't help it. I spend my day talking about this now.

Mm-hmm.

Alane Boyd (09:52)
like this that are coming up. Like there's still this huge disconnect in how they're using things because this is the first time that we've been able to operate like this. So it's okay if you're in that boat, but join us and and get into the the new the new times. So so the fancy into a fast yacht.

Micah Johnson (10:08)
The fancier boat. Get out of the rowboat and get into the speed boat.

Okay, super fast. Love it. Yep.

Alane Boyd (10:17)
All right. So there's two main things outside of deciding on what your single source of truth is for that information. When you're using Cowork, is there's Connectors for a lot of these file storage. So we've got Drive, Box, SharePoint through the Microsoft 365 Connector. If you're using Ignite, there's even one with it.

And we use Dropbox, but we have a workaround for it through the other way would be through the local file storage that syncs with your cloud storage.

Micah Johnson (10:51)
Yeah, so somehow connect Cowork up to the source. Don't move files from the source into Claude

Alane Boyd (10:51)
Or an MCP server.

Ooh, I love that, Micah. So simple.

Micah Johnson (11:04)
Finally. I really I

struck out earlier in this episode, so I was aiming for at least a base hit.

Alane Boyd (11:11)
You did. So yeah the idea would be you're connecting it to the where your file storage is already happening, where it can at a minimum read those files and answer questions, which I mean, Micah, that is what I'm seeing a lot of the times that companies need is a way to reference those files to get the answer that they need. Now that's not always. Sometimes it needs to write new files or whatever it might be, but at a minimum, the most common use cases.

I don't want to have to dig through 40 folders to find what I need and then read through the document to get the answer. They're really just trying to shortcut that to get the answer to a question.

Micah Johnson (11:50)
Well, and what we've been referring to this a lot lately, this is the knowledge layer. So connect it up to the platforms that have the source data and the knowledge, which becomes context for the task that you're asking Claude to do.

Alane Boyd (12:03)
And then the other way that we talked about is through the syncing on your local drive. So for us, we're using Dropbox, we have some of the folder syncing to our local computer, and then we're using Obsidian to communicate with those files. We have an entire episode on this. So if you're getting lost in what we're talking about, you can listen to it. But even other platforms, other file storage has a local sync. So this is an easy way

if you don't have a direct connector or if you want to be able to write files as well and save files through Cowork to something like Dropbox, then it will do that.

Micah Johnson (12:40)
Yeah, I love it. That is a perfect place to start and to start practicing the give it the source of truth.

Alane Boyd (12:49)
So for anybody listening, I think obviously the takeaways are a single source of truth, but if you don't have one, picking one, one that has cloud storage, most appealing, and then start operating from that single source of truth. Come up with your files, naming conventions, how you're gonna save things, train your team and then connect it up with Cowork.

Micah Johnson (13:12)
Yep. And remember, we have two different places that need to re consume this now. It is the humans and the AI.

Alane Boyd (13:23)
And if you're wondering how to get this set up, we have a free Cowork Masterclass in our Community. It's your.biggestgoal.ai that walks you through, I think Micah, how many lessons are in that?

Micah Johnson (13:36)
We're at 26 or 27 now.

Alane Boyd (13:38)
Okay, there's 26 or 27 lessons that we have written to help you get set up in Cowork for real, not using it like chat. It is completely free right now. So go sign up at the Community, your.biggestgoal.ai, and sign up for that free course. And as a bonus, you get Micah's daily AI News Brief. So you can start sounding like you're an AI expert in any room.