Claude Skills are the Most Underrated AI Feature

Forget the full AI agent. Alane Boyd and Micah Johnson are making the case that Claude Cowork Skills are the most powerful, most overlooked feature in AI right now. One slash command. Twenty-five steps handled. Consistently, every time.

If your business relies on humans doing the same thing the right way every single time, you already know how that goes. The inconsistency, the retraining, the quiet shortcuts. It's not a people problem. It's a systems problem. Claude Skills are the fix.

In this episode:

• What a Skill actually is: part SOP, part system, part micro app
• How to standardize your team's work without retraining anyone
• Why adding a deal to your CRM is secretly a 25-step process, and how one command handles all of it
• How Skills connect to your tools and run automatically in the background
• How to build your first Skill the right way: start simple, iterate, don't wait for perfect
• How Alane's workshop analysis Skill works exactly like she does, so she doesn't have to be the one doing it

If your team is doing the same thing five different ways and you're ready to fix that, start here.

Upcoming Education Sessions:

• Free Claude Cowork Webinar
• Free Claude Cowork Masterclass
• Claude Deep Dive Series

🛠️ Tools & Platforms Mentioned

• Claude Cowork
• Grain 
• ClickUp
• Pipedrive
• Copper
• n8n
• Python
• MCP (Model Context Protocol)
• Anthropic

Learn More about our AI Agent + RAG Cohort:

Not sure how to get started with AI? Join our 4-week AI Agent Development Cohort, where you’ll build real AI agents and automation. No technical background required.

The July 20th cohort is filling fast. Save your spot and sign up 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
• 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

Alane Boyd (00:03)
Do you always need to build a full AI agent? Not necessarily with Claude Cowork Skills.

Alane Boyd (00:09)
What are Skills and how can you start using them? Listen to this episode to find out how and why they're possibly the most underrated feature of AI.

Micah (00:23)
All right, Alane, I've got a question for you at the top of this episode. It's a would you rather?

Alane Boyd (00:28)
okay.

Micah (00:29)
Would you rather train somebody to do 10 steps and hope that they do it the same way, even with a written SOP, and create the same quality across every single one of those same steps? Or would you rather just train somebody to type forward slash and then the name of the set of steps that they're supposed to be doing and have that happen automatically? It's a trick question.

Alane Boyd (00:53)
Well, the f

I feel like w this is a drinking game right now. Yeah. So the first one was so long I tuned out for part of it, to be honest. No. Yeah, exactly. So definitely the second one, man. And I know where you're going with this. And it is such a lifesaver and really

Micah (00:57)
It always this whole this whole podcast is a drinking game. Cheers. This is water.

So did the person you trained.

Alane Boyd (01:18)
The way that you just described it makes it so obvious, of course you'd want to go with option two.

Micah (01:24)
Yeah, and that's what I want to talk about today is Skills for Claude Cowork in particular.

Alane Boyd (01:33)
I love using Skills and I think that was one of the things when we first started using Cowork that it felt so scary to start doing and you're going like Alane, just tell it that you want to create a Skill and tell it what you want it to do and it's gonna build it for you. And I'm like

Micah (01:49)
Well, I'll tell you

the number one question we get about creating Skills is how do we create a Skill?

Alane Boyd (01:55)
Yeah, it is. And it really was that easy. And the great thing with Cowork, if any of you are using it that are listening, is that if it gets it wrong, you just say, hey, let's rework this. I want to fix this part of the Skill that you created. And it is such an easy process to work from. Micah, let's dive in a little bit deeper on what the heck we're talking about. Yeah.

Micah (02:15)
Yeah, what is a Skill?

All right. So at the very, very basic level, a Skill is a SOP or a Standard Operating Procedure, a set of instructions for AI. It's exactly the question that I asked at the beginning. If you have a 10-step process, would you rather train a human to do that? Or would you rather have AI do what it can do out of those 10 steps? And

A Skill is essentially writing that down once with the help of AI to perfect what those say 10 steps are. And then the next time you need to do those 10 steps, even as a person, right, you just ask AI to use that Skill. And it loads that in, runs through the 10 steps, it creates consistency, it creates standardization, it does it the same way every time, and it follows its instructions.

Alane Boyd (03:11)
so on a basic level, it's so similar to how we talk about AI agents. Like this is basically a done for you AI agent that Cowork is doing. And it's so much more than when I think of an SOP, I think of it as a concrete way of doing something. And it's for like one linear process.

Micah (03:19)
Yeah. Yeah.

Mm-hmm.

Alane Boyd (03:32)
What I love

about a Skill is that it's taking an input and can have multiple outputs associated and with other software platforms. So it is really automating a process with the instructions to go alongside of it.

Micah (03:47)
absolutely right. What you're talking about is embedding the logic that we used to have to program into this. Like we couldn't, we could program quote unquote program logic into an SOP, right? It was a bunch of if this you know value is this, and I'm not even talking about programming languages, I'm talking about writing in an SOP.

Alane Boyd (03:52)
Mm-hmm. Mm-hmm.

Micah (04:08)
We've heard the terms decision trees. Remember, we used to have to map all these possibilities out for people to follow so they knew exactly what to do when this happened or if this happened. But what you're saying you can build that into a Skill and AI can navigate that.

Alane Boyd (04:25)
Yes. And it's so incredible because it's all done with slash commands and Cowork. So it's slash and then the name of that Skill. It's that simple. And that's all you have to do with the Skill once it's created. So say it's a call debrief Skill, you just do slash call debrief and then maybe the Grain link so that it knows which call you want to do. And that's it. Then it creates the email draft.

It creates the action items in your project management tool like ClickUp. It'll create a summary of next steps and action items for that client. It does all of those things with just a few words.

Micah (05:06)
I don't even give it the link usually. I just say it's in Grain. The call that I had with blank,

Alane Boyd (05:11)
You can go either way. I started doing it with the Grain link, instead of telling it 'cause I couldn't remember the name of the meeting.

Micah (05:19)
That works too. That works too. And I I think that's one of the things that we should really call out in this, even with a Skill defined, the execution is standardized, but the inputs can be variable, which means if you give it a Grain link, sweet. If you say, Hey, you're connected with Grain, go look for this meeting that I had yesterday. Sweet. Like it can do all of that and it doesn't return an error. Like, oop.

Alane Boyd (05:20)
Sometimes I get creative.

Mm-hmm.

Mm-hmm.

Micah (05:48)
Wrong input, like I don't know what to do from here. And same thing with the outputs. And while it's executing, it's making all of those like slight adjustments based on each step on what it has.

Alane Boyd (05:52)
Yeah.

Yeah, I will say with the Grain link, and just I just want to mention this for listeners. The reason that my Cowork can see that Grain link is because I have it connected. So if I don't, it'll say, Hey, I can't see this. Can you paste the transcript? So, Micah's talking about it from a Connector standpoint because it is connected to Grain. But in a way, I am too, because even if I paste the link, it still needs to be a Connector in there. So I just wanted to put that nuance in.

Micah (06:13)
Good point.

Alane Boyd (06:31)
It can see things because we're also utilizing those custom Connectors.

Micah (06:35)
Yeah, so I started with Skills, Alane saying that they start as SOPs. What we're already talking about is a system. And so a Skill is actually like a miniature system because we could take this call debrief Skill that we're talking about and have AI do parts of what we want, get the transcript, analyze it, summarize it, do this. And halfway through that, it can come back to us and say, hey, does this look right?

Alane Boyd (06:41)
Mm-hmm.

Mm-hmm.

Mm-hmm.

Micah (07:04)
We can chat with it and give it feedback and then it can carry on with the rest of its thing. That's a system.

Alane Boyd (07:11)
Mm-hmm.

Micah (07:12)
And there's a lot of different ways that you can leverage that. But if we even go one step further, I would argue that a Skill is actually powerful enough to be a micro app.

Alane Boyd (07:25)
I love that you kind of came up with this idea because if they hadn't been created in Cowork, they would have had to have been an AI agent. And we have called those mini Saas for the last two years since we've been creating agents, because that is absolutely what you're creating, even though it's super easy now to create it in Cowork when you're just telling Cowork to create the Skill. It is still a micro app. You are plugging

together multiple software platforms that you're using with different outputs in each one of those. That's essentially an app.

Micah (08:01)
Yeah. So let me let me give you a really quick example of that I did earlier today. I needed a way to an API token. This is something that I'm gonna do over and over again. Now I could go and write a Python script But I work in Cowork all the time. So I thought to myself,

why don't I make a Skill for this? And there's one caveat that I really want to point out here, which is that when you're running an AI Skill, and I was confused about this when we first started with Skills, I thought everything went through the LLM. and everything was using tokens and usage, But that's not how it works.

You could actually have a Skill that you call and say, Hey Claude, make me a new API token. Awesome. It's going to start with AI, it's going to look at its Skill instructions. It's going to realize it has a Python script embedded into this. So it's going to go from AI to Python, which is a programming language. It's going to run the Python script to create the

encrypted API token, and then it's going to go back to the AI to say, is this what the user is looking for? Yes or no? And then finally reply back to me. And then I can chat with it and have it like that. That's like a super, super small Skill that does the same thing over and over. But it's not going to, Claude, and Claude isn't making up a string of characters. It's actually using.

an encryption library in Python using Python script and then sending the results of that within the Skill back to AI and then sending the output back to me as the user, which makes it an interface.

Alane Boyd (09:50)
Hmm, that's super interesting. You know, one of the things that I keep thinking about through all of this too, Micah, is the necessity for standardizing. Like that one's such a big topic. You know, it's kind of a common thread through our podcast. But, you know, thinking back even when we started doing Skills on the team level for our organization is we don't want

a Micah way, an Alane way, you know, an individual way of doing things for the same stuff that we're doing for call debriefs. We want a Biggest Goal way of doing that. And I notice this as a conversation starting to happen more, because people are noticing, like, we don't have a standard way. If I'm going to create an organizational Skill, then I need a standard way of doing something.

Micah (10:22)
Yeah.

Yeah.

It forces that thought process.

Alane Boyd (10:39)
It really does. and it can be hard for some of the people on the team because they're used to not working that way. They're used to being the lone ranger getting to make every decision on how they work. But when you're looking at, hey, we're going to use AI and we want to have a standard way, a company way of how we do things. This is the perfect opportunity to say we are going to create a Skill for this thing. And it is going to be our company way of doing it. That way it's reusable.

That's the whole benefit of having Skills is that it's a reusable asset that your team can use.

Micah (11:12)
So I want to give an example of this, Alane. this is gonna sound like a ridiculous fake example, but it's absolutely not. It's creating a Skill to add deals to a CRM. And here's why it sounds fake. Like, why the hell would I build a Skill just to add a deal to a CRM? That's so easy. But

Alane Boyd (11:24)
Mm.

Micah (11:35)
It's exactly what you're talking about with standardization, with templating, with naming conventions, with structure. All of that goes into creating a deal in a CRM. You've got to have do we want to require certain fields? Well, then we can build that into the CRM and customize that, or we train people to do it. I mean, think about that 10-step question that I asked in the beginning. I bet you there's 25 steps just to add a deal to a CRM if you're gonna do it a

Alane Boyd (11:52)
Mm-hmm.

Micah (12:02)
Biggest Goal way or a your company way. Because you're gonna want to name the deal in a specific way. You're gonna want to leave notes in a specific structure and format. You're gonna want to include certain topics in those notes. You're gonna want to extract the data from the prospects that you were talking to, the contacts, the people, the company. You're gonna interrelate all of that. You might need to make multiple deals, one for each product or one for each service. Maybe that's a standard way to do it. Training on

all of that, and we've all been there, right? it's let's just say it. I'm gonna say the word Alane, impossible.

Alane Boyd (12:32)
Yeah.

Okay.

So you know what I'm thinking through this is I'm gonna throw myself under the bus because I am the worst at keeping the CRM updated with deals. That is not time that I want to spend. And so it just doesn't get done. So thinking about every salesperson on the team, how many of other Alane's are there on the team? There's a lot of us in sales that just do not want to take that time.

Micah (13:04)
I would argue that there is zero percent of salespeople who want to take the time to do this exactly the right way every single time, following a twenty-five-step SOP.

Alane Boyd (13:15)
Yeah. And if it's getting done, they probably have an assistant to do it for them. So you're paying a person to do that. And then the other thing is that I was thinking through is kind of going back to what we were talking about just a second ago, is that if you waited for every individual salesperson on the team to create an individual Skill, they are not going to be done the same way. They're not all of them are going to do it.

Micah (13:20)
Two people.

Alane Boyd (13:38)
So standardizing this is the way that we keep our CRM updated and having a Skill for that creates it it's so easy after that.

Micah (13:46)
Here's the other thing. We've worked with hundreds of and have had, I don't know, probably close to a thousand calls with companies Alane of all different sizes of all different types. Never once has anybody gone, you know what I'd like to do? Standardize adding deals to my CRM. They want to invent these giant frickin' solutions that solve these big problems. But again,

Alane Boyd (14:04)
Yeah.

Micah (14:11)
How many salespeople do you have? And even if it's only you, I can also guarantee that I've never entered a deal in the same way exactly over and over again. And I certainly don't want to do it. It's not the work that anybody wants to do. So is it better spending all this time devising this giant process and this workflow? Or in an afternoon, can you define how you want to add a deal?

Alane Boyd (14:22)
Mm-hmm.

Micah (14:40)
To your CRM and communicate with Claude and have it help you write this Skill. And tomorrow, all you have to do is go to your team and go: when you add a deal to the CRM, go to Cowork, type forward slash add deal, and then point it to the customer name or an email address. And it will connect to your email. It will connect to the CRM to see what else has been in there in the past, past notes. It will connect to your call transcript.

software, your AI note taker, and pull out all the information. It'll look at the signature files and get the company. It will search the web and do discovery for you while you're putting this all in. And on top of that, it's going to fill in all the required fields. It's going to put it in with naming conventions. It's going to put it in in the standard way. I could keep going on, but I think everybody probably gets the point. That is such a simple thing in a business when you think about it.

But it is a massive, massive gain if you just made a Skill that does that. And do get me started or don't get me started, Alane, on. I'm already started on editing. Do we create a Skill for editing? Hell yeah, we do. And then what if we wanted to improve that or change? Obviously, we change processes all the time as a business.

Alane Boyd (15:47)
I think we already got you started.

Micah (16:03)
So what happens if we want to change the process of adding a deal to a CRM? Do we retrain all 10 of our salespeople, all 100, all one of our salespeople? Hell no. We just go to Claude and say, hey, can you update the Skill to do this? And we don't even have to tell anybody that we changed it.

Alane Boyd (16:19)
Yeah.

Yeah. I mean it's good if you do. I think that that you know

Micah (16:26)
Maybe. They don't

nobody's gonna go to the CRM at that point to look.

Alane Boyd (16:30)
Yes. I wanna just drop this little tidbit in is that you could have a scheduled task that runs a Skill.

Micah (16:37)
Mm-hmm.

Alane Boyd (16:38)
So you could have it check your email for things that you missed in the sales world. I know we're making this very sales oriented, but I think the reason that I've been talking about the sales example too is that this has been in the press a lot lately. Anthropic just did a couple of weeks ago a whole webinar on how they're using Skills in their sales process.

It's all in the backend stuff in the call prep, in the call summary, in your daily brief of what sales call. It's all in that backend part that is the time consuming part that and the humans get to do the selling. And JP Morgan, you know, Micah, your AI News Brief this week was all talking about that. And this is a common thread that we use internally ourselves, and that we're seeing these much larger companies also.

Micah (17:09)
The research, yeah.

Alane Boyd (17:28)
And we're all saying the same thing. Let humans be humans, where you want to be person to person and chatting and discussing things. Like I always laugh because every week I at least get one email saying, is this really Alane or is this an AI agent? And I'm like, no, this is really me because that's the human part, is the human connection piece. A human doesn't have to do the call prep stuff. That can be provided to me so I can be more valuable on our time together

Micah (17:55)
Yeah, I couldn't agree more.

Alane Boyd (17:57)
I was gonna transition into some of the things that you can leverage in Skills like templates and having a consistent output. What is your boilerplate template output that you want?

Micah (18:08)
Yeah, you can get as detailed as you want. You can give it a document to duplicate. You can give it placeholders. You can even get to the point where a Skill could call an n8n workflow to do things on other platforms that Claude or MCP servers can't do right now. So template-wise, man, you just start wherever you want to start really with the standards.

Alane Boyd (18:38)
So a couple of things Micah to wrap this up that I wanted to mention is, when you're creating Skills and you feel unsure, one of the things is that we have a free Claude Cowork Masterclass that's available. And it's in our community called your.biggestgoal.ai. And it's completely for free. We have

how to create Skills, how to maintain your Skills built into that. And then we mentioned your AI News Brief that you send out every single day. That's also in the community. So you can go to your.biggestgoal.ai and get access to both of these.

Micah (19:11)
Yeah, absolutely. Absolutely. So

We've got a Skill that can be an SOP. It can be a system. It can be a micro app. It can run its own scripting languages like Python. it can leverage templates, you can give it examples, it can leverage layouts, it can leverage assets like images, it can even produce all the actual outputs that you're trying to produce. So, like you were saying a second ago, Alane, you get to be the human, the director, the judgment that says, like, what if

What if we tweak this or what if we change this? That all is amazing. I will say there is a skill to making Skills.

Alane Boyd (19:53)
Well, we have to be the expert, right? If we just let AI, then where are our expertise? What makes us special as a company, as a service provider, whatever it is? So absolutely like the reason my workshop analysis Skill works so well is because I told it what I wanted the output to be, how I wanted it, and gave it the data to work from. So that Skill works exactly like I would.

Micah (19:55)
Yes.

Alane Boyd (20:18)
But I don't have to be the person doing it anymore.

Micah (20:20)
Well and I wanna ask this too, Alane, 'cause I I don't know. Did it get it right the first time or did you have to work through a few things?

Alane Boyd (20:28)
I had to work through a few things. and this is what I love about Cowork too, is it prompted me to give an example that I could work from for this? Well, absolutely, because I'd done it as a human multiple times. So I provided it the spreadsheet and the final output of one that I really liked that I had done. And I said, Hey, this is what I'm looking for. It produced a couple of examples for me

for the Skill and said, Hey, what do you like about these? And I made two changes. I said, I really want this to be a chart here. And I really like the finding the selection of people here. And so it went back and did that and I did maybe like one other tweak and I was done.

Micah (21:06)
Yeah. And each time I'm sure, well, I'm not sure, but I'm guessing you tested it and you looked at the output that it produced, and then you were like, okay, that works, or not this. So even how you communicate, how you're asking for it, like next time you do a Skill, you'd probably give that example up front, knowing that you already did it once and that you needed to give it the second time because it was asking. So, you know.

Alane Boyd (21:26)
Mm-hmm.

Micah (21:33)
You're developing your own skillset in how to build Skills with Claude.

Alane Boyd (21:36)
Mm-hmm.

And it's how we talk to our clients about this is that you have to get your feet wet. You got to get in there and just do it. Like there's nothing to be scared of. just because you do a Skill not a hundred percent right, like you're not hurting anything. Just work with it. And I guarantee you it's gonna save you so much time if you just get in there and get your hands dirty. Get your hands dirty, get your feet wet.

Micah (21:58)
Yeah, am I

I I was wondering you paused and I was really proud of you for getting the saying correct. I was wondering what was gonna come out.

Alane Boyd (22:04)
I had to think about it.

No, I didn't know.

Micah (22:10)
All right. So my tip on this too is also to start simple. You don't have to build the whole thing. And again, even take something that sounds so simple. Adding a deal to Pipedrive, adding a deal to Copper, adding a deal to any CRM, right? Like that sounds like one thing.

And we generalize that in our brains, but when we try to create a system or an SOP or a micro app around that, it gets really complicated really fast. So you don't have to start with the absolute perfect end result Skill, even if it sounds simple. Don't beat yourself up about it, because I know a lot of people do. You just gotta start with a Skill that says, hey, I'm gonna review email and I'm gonna get the information out of the signature file and I'm gonna create the records.

Alane Boyd (22:58)
Mm-hmm.

Micah (22:59)
That is f'in it. Just start there.

That's a success. You've already saved 20 minutes of going to finding all the contact information, finding the titles, finding the addresses, creating the records, associating the records. That's 20 minutes. You've already saved. So from that point, just get that working and then do a small little step. Well, now I want it to check the transcripts. Awesome.

That's an upgrade, that's an improvement to the Skill. Don't add all of that right out of the gates.

Alane Boyd (23:31)
So Mike, I want to try something. So we've been getting comments on our podcast, which we absolutely love. Thank you to those that leave us comments, especially the positive ones. Is if you have created a Skill, tell us about your Skill. Drop it in the comments wherever you listen to our podcast.