👋 Hi, it’s Kyle Poyar and welcome to Growth Unhinged, my weekly newsletter exploring the hidden playbooks behind the fastest-growing startups.

Last year I invited Alex Shartsis to share how he’s playing with vibecoding to build GTM leverage. This become of the most popular newsletters of 2025. Now Alex is hooked on something new: creating his own personal AI CRM. He built it with Claude in 20 minutes for only $20 per month, and it’s already more powerful than a traditional CRM. Alex is the co-founder of Skyp, the AI-native outbound platform. He also advises companies on GTM and writes a newsletter.

Today’s edition is free. Subscribe or upgrade to support Growth Unhinged.

CRMs were one of the earliest use cases for software in business, and while they've evolved through the SaaS revolution, Web 2.0, and more modern web technologies, they are still painfully difficult to use. CRMs are mostly just databases, infrequently updated.

What’s more, typically only sales and marketing really use the CRM (if anyone does). Yet so much context exists that would benefit anyone in the organization, much of it locked up in communication tools like email.

As a bootstrapped founder and experienced CRM user/admin, I am all too familiar with the downsides of setting up a major CRM – even a modern one. But I still need to figure out who to follow up with, understand my sales cycle, and report to the rest of my team about what’s going on.

So I built out Claude to do all of this for me, and I'll tell you how to do the same thing. It’ll take you about 20 minutes (even with a hiccup or two along the way) and cost just $20 a month, all while delivering 90%+ of the value of a CRM. I'll also give you a few CRM vendors to check out that are truly AI first – in case you’d like to go that route.

The toughest challenge in this project was determining the starting point — specifically, identifying which problems I wanted a personal CRM to solve.

I settled on follow-up management because that's where I experienced the greatest context gaps and spent the most time.

I wanted to replicate the process by which I thought through follow up as a human – go back through meetings, emails, and meeting notes to figure out where things stand. Then, come up with a list of follow up tasks. Many of those would be one-to-one emails, which I would draft and send. There would also be a lot of large-scale follow up to be done – which I also would want one-to-one emails for, but by the dozens.

This led to two objectives:

I wanted to do these jobs while drawing on all of the data I already have from my emails, calendar, meeting notes, and customer list. Flexibility was important, too; sometimes I’m managing relationships around marketing (like working with Kyle on this piece) and other times I’m talking to sales leaders about software. These are very different motions, and I like keeping them separate.

Importantly, this meant I didn’t need a bunch of features you’d usually expect in a CRM (and that would’ve made it unwieldy). Things I felt were dead ends:

Briefly, why Claude? This is not an arbitrary choice. Anthropic (Claude's parent company) invented and popularized the MCP server standard. MCP stands for model context protocol.

I don't love jargon but this concept is important. MCP is a standardized protocol that lets AI models interact with external data sources and tools. Think of it as a universal adapter. Instead of building custom integrations for every service (Gmail, Stripe, Grain, etc.), MCP provides a common language that LLMs understand natively. The AI can discover what data is available and how to access it, without you writing any integration code.

This means your LLM always accesses the freshest data without you uploading anything or paying to store it twice, and without you building connectors. This is fundamentally different than "training" an LLM on your data. That is a one time event (also worth learning, but beyond the scope of this post) and there are limits to how much data you can include. Your entire Gmail archive will almost certainly exceed those limits. Instead this is giving an LLM access to fresh data, every time it needs it, without having to build extra infrastructure to do it. Find vendors that support MCP, and it just works.

To connect to your personal data sources securely, MCP currently requires a local application. The MCP servers run on your machine to broker access to Gmail, Calendar, and other services. Claude Desktop supports this today. ChatGPT does not. All of this will likely work in ChatGPT once it supports MCP, but I prefer solutions that work today, not in the future. An alternative is Cursor, which supports MCP servers, but Claude is more accessible for most people so these examples are in Claude. Anything I explain here can probably be easily implemented in Cursor in a similar way (if an MCP server connects with one LLM, it should work with any LLM).

The setup is incredibly simple, and doesn't even require vibecoding knowledge:

If you have the time, it's worth connecting your other data sources. The more you give your personal CRM access to, the more it can do for you – whether that’s simply collecting knowledge and answering questions for you, or actually making changes in your systems. Some additional systems to consider connecting:

Some of the connections are unstable, or will reset. So you may have to come back here to reconnect them. Claude will typically try to connect first, especially if you explicitly tell it “go check Grain meeting recordings” – but if it then comes back with something to the effect of "I can't access X", it won’t stop! It may just make stuff up in order to fulfil your original request. You have to stop it yourself.

While Claude is better about this than ChatGPT in that it won't invent data to finish its job, it will still do a lot of useless work that is destined to fail – costing you tokens. On your first request involving an MCP connection, always keep an eye on it and make sure it did actually connect. If it didn’t, cancel it and go troubleshoot the connection.

Many MCP servers are rate limited, just like any API. Claude usually figures this out for itself as part of the MCP protocol. What that doesn’t solve is too much context. If your dataset is very large (perhaps you have 100,000 customers in Stripe), you may exceed the “context window”. A higher-tier plan with Claude may solve this, or you can discuss with Claude how to filter the dataset to keep it within the context window.

Claude builds scripts and writes files to temporary folders that you can’t directly access as a user. If you want to reuse them, and access them directly, switch to “Claude Code” mode. This gives you visibility and access to the files and can make them more reusable.

You can also tell Claude to write the scripts to a specific folder instead of its temporary folder. This is less about cost savings and more about time efficiency. Therefore it only matters if you are using this a lot, or making the same request a lot. Even the $20 plan has enough credits where you won’t run into credit limits.

All of this is somewhat advanced. I’d start with plain Claude, as described below, and then when you see the value and have a firmer understanding of what you want done – maybe consider switching to Claude Code then.

Now that you have these things connected, you can do the job to be done. Here's the prompt I used to set up my list for end-of-year follow ups:

You can obviously go deeper, for example adding Give me people I met with in the last 30 days, or emailed but didn't meet with. Basically having Claude as your CRM is like having a very smart, hardworking analyst who has access to all of your data.

When you get your answer, it will be as a markdown or CSV within Claude. You can review it, and iterate. Much faster and easier than either using the reporting features of most CRMs, or doing it by hand in Excel. Like an analyst, expect it to miss occasionally. Also, if you have a lot of data, expect it to start on a smaller subset on its own volition (maybe the last week or month). Double check its work.

In my workflow, I take the CSV as an output, review it (sometimes) and upload it to Skyp to send email campaigns. You could do that – or use any email outreach tool. The advantage of a tool like Skyp is that it uses AI to write the emails in their entirety so can pull whatever knowledge you include in the CSV from your email exchanges or meeting notes. You can also do this in a tool like Lemlist or Instantly, but you’ll have to include custom variables in your templates and map those variables to your CSV data. Doable but extra work: I prefer AI native workflows.

Another use case is analyzing the week coming up. My cofounder and I meet every Monday to discuss what happened last week and what’s coming up this week. We do this as separate meetings for both sales and product. I have Claude put together a summary prior to the meeting:

Doing this during the holiday break led to some interesting behavior – but a quick discussion with Claude sorted it out, so “last week” really became the week prior to Christmas, and “next week” became the week of January 5. This kind of flexibility would be hard to impossible with a traditional CRM.

This task was mainly figure out who to follow up with, but really it's about closing new customers. You are not done yet until you have gone all the way through that process and closed some customers (or gotten to "no").

You can simply go back after a week, and cut/paste in the same prompt you used the prior week – or, just ask Claude what is happening. While unstructured, this can be a great way of catching opportunities, stuck deals or follow ups that may have slipped. In my case, replies or meeting bookings will land in email and/or calendar, meeting transcripts in Grain, and new customers in Stripe. So Claude has full context: the entire sales cycle, at your fingertips. No dashboards required.

I am routinely blown away by things that I missed that Claude picks up – customers who booked a meeting, but no-showed. Or emailed me back, but I missed it and forgot to reply. While it’s tempting to beat myself up about it, it’s important to realize that everyone is human. The best humans leverage AI to appear more perfect. The winning combo is human and machine working together.

There are pieces of this process where dedicated tools still win over generalized AI like Claude. Adding context about the prospects might be something you have Claude do when generating the list, but the infrastructure to do enrichment at scale is not trivial. A lot of sales people (myself included) know that the hardest part of selling is coming up with the list, and following up. This tool helps solve both of those problems extremely well for a single salesperson.

Claude doesn't run off and do a ton of stuff that you didn't tell it to do, like (ahem) some other GPTs. But you can tell it to go off and do extra credit if you want it to.

On the list building front, you can tell Claude to crawl the web or do other enrichment tasks. Some enrichment databases are also available by MCP, so you could connect Claude to them and have it leverage that data to add additional context to your list before you add it to your email software. Be careful here – it's very easy to end up having your emails look like AI slop if you over-personalize.

If you are doing manual follow up, you can have it draft emails to all of those people and cut-and-paste them into Gmail by hand. I do not recommend adding Claude-written emails to the sheet and automating via Google scripts; while possible this is a surefire way to get your domain put in the SPAM penalty box. It is nearly impossible to get out of it, and can take months at a minimum.

You can generate powerful and very useful reports from this data set, without writing code. For example, try something like this:

This resulted in a clear table of each customer, when we first spoke with them and how long it took between that time and when they became a paying customer. We were further able to dig in on characteristics and things they said from the meeting notes to build a much more actionable ICP for future outreach and ad campaigns.

This example only solves for a small fraction of the functionality of a full blown CRM. I would argue it covers the most important parts, but that really depends on your role.

In case it doesn’t go without saying – the usefulness of this system extends well beyond salespeople into nearly every other aspect of the company. Bringing all of the context you have as a human together in an AI like Claude is incredibly useful.

This is not a true “CRM replacement” in the sense that it isn’t wired deeply into everything. If you have a larger sales team, there may be ways to aggregate their communications and calendars together and give Claude access – but I am not sure what they are, and they are unlikely to be easy.

On the other hand it is capable of doing a lot of things that a CRM might, such as forecast sales, analyze sales cycles, spot recurring customer issues, identify your ICP – things that are valuable not only to sales but also success, marketing, finance, accounting, and leadership.

To overcome single player limitations, someone with the proper access could build a Google Sheet and share that with those who need it via Google Drive. Then Claude would have access to what it needs. For example, if Stripe access isn’t realistic, a Google Sheet of customers that either is manually updated from time to time or automatically updates from Stripe would ensure each seller had accurate customer information. Without some complicated, multi-month CRM integration.

You could absolutely build something more comprehensive and multiplayer. I’d start small, with the single player version, so you can first learn how people engage. If you want to build out something more complex, or have built it already, I’d love to hear about it.

What I learned in building a vibecoded CRM and then setting up this Claude desktop configuration is that the hardest part is knowing what jobs to be done you want to do, and thinking through the steps of how you do them.

To me this is the value of paying for a CRM: smart founders and product managers have spent lots of time both with people like you and future versions of you – addressing the needs you didn’t realize you had now, and don’t realize you will have one, two, or five years down the road. The two truly AI native CRMs I have spent time with are Day AI and Lightfield. There are others out there that market themselves as AI, but be careful here: an application that was built more than two to three years ago almost by definition cannot be AI native.

CRMs serve many critical roles in business, but the AI era highlights the gap between promise and capability. Here I’ve laid out a simple 20 minute implementation within Claude that can deliver on 90%+ of the value of a CRM to a solo seller, account exec or marketer – at around $20 per month. Could you do more? Sure. Does it do everything? Absolutely not.

The only downside of the MCP-based approach is that you have to wait for Claude to reach out to the MCP and gather data each time. If you’re used to snappy queries written by your SQL analyst, that might disappoint. But if instead you change your expectations to Claude almost replacing your SQL analyst… then it’s pretty quick. A few minutes is better than waiting days until he or she gets around to your data request.

The hardest part isn't the technology; it's knowing which jobs you actually need done. That's why I still see value in purpose-built tools for specific workflows (email outreach, contract management, pipeline reporting). But for the core job of "who do I need to talk to and why?" — Claude and MCP gets you 80% of the value for $20/month instead of hundreds or thousands.

Additional resources: