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

In recent conversations with founders, one topic keeps coming up: how the he** to position the product without sounding like everyone else. I’ve previously featured frameworks from Anthony Pierri and an AI GTM context system from Matteo Tittarelli. Today I’m bringing you a real-life story from Melissa Kwan, co-founder of eWebinar.

Melissa has spent the past 5 months repositioning her entire company with Claude. It’s working: 50% more time spent on the website, closing her first enterprise deal, and far fewer objections. Keep reading to go behind-the-scenes and learn how to build your AI brain for positioning. (Follow along as Melissa builds in public through her newsletter and podcast.)

I thought I had a good business. Profitable, bootstrapped, around 700 customers, no investors to answer to. What I didn’t have was a good business that was growing into a great one.

Companies who use eWebinar love it. It’s on-demand video software with chat that lets you deliver demos, onboarding, and training to your entire audience without anyone having to be there live. (For six years, we called it automated webinar software, which is a much smaller category.)

The product delivers real value, which made the situation confusing. If this many people loved it, why was it so hard to find more people like them? Why did nothing we tried seem to move the needle?

We were in the most dangerous place a company can be: good enough to not question it. Not bad enough to raise alarms to stop and dig.

For a while, I wondered if this was just our ceiling. Maybe this was as big as we could get. I asked myself whether I’d be happy running a small lifestyle business, and when the answer turned out to be absolutely not, I went looking for the problem.

I got back on sales calls and went to a conference and talked to prospects and customers in person. I asked the uncomfortable questions, “Why aren’t you moving forward? Why haven’t you expanded?” I listened between the lines, and after a couple of months the pattern was obvious.

Most people didn’t fully understand what we did or what problem we solved.

We had a positioning problem. Specifically, we didn’t have product-messaging fit.

People who heard our pitch or landed on our homepage couldn’t say what we did in a few seconds, or why we existed. When that’s the case, nothing downstream works. Not SEO, not sales calls, not ads. Every conversation starts with an explanation, except for the small percentage that’s problem aware (I dissected the diagnosis in detail here).

So I decided to reposition the company, and go find true product-market fit.

I thought repositioning was just rewriting our homepage. I was naive.

How you describe your product and the problem you solve is your entire business. If you change those things, you have to change everything they touch. Homepage, every subpage, onboarding and nurture and churn emails, blogs, every directory and app store listing, explainer video, demo script, in-product copy, and call scripts.

All of your content has to match because positioning is retraining, and retraining only works if every touchpoint says the same thing. The customer who reads the new framing, signs up, gets the old framing in their inbox, is at risk of going back to the mental model you’re trying to get away from.

That's a lot of content for a company like us with no marketing department and no budget.

In the past I would have hired an agency or a messaging expert, which costs a fortune in money and time. So I decided to build an AI brain instead. Messaging fit is the founder’s job. Nobody else is going to figure out what your company is.

Related: How to build an AI GTM system

So what is this brain, and what does it do? (P.S. I use Claude.)

A clone of me and the company, one place where everything we know gets downloaded so it can help put the pieces together and generate content that sounds like me.

In practice, it’s an AI project that holds everything about my company and its customers:

It’s the single source of truth behind every piece of content we write.

The quality of the AI brain is the quality of what you feed it. Here’s what I put into my project. Use some creativity as to what to put into yours. This is the fun part.

One note on this. I reread Obviously Awesome and took my own notes. I decided which parts mattered for us. I asked AI to help me apply a method I had already chosen. It did not choose the method.

Don’t outsource your own thinking. The direction has to be yours.

Related: The definitive product positioning framework (with Anthony Pierri)

After uploading the first inputs, I had Claude create two documents it could work from, so it wasn’t rereading everything from scratch every time. Faster, cheaper, and more consistent.

What came back was a sentiment analysis across years of evidence that I could not have assembled any other way. It was magical to be able to read this in an organized way for the first time.

From then on, every piece of content could point back to a decision we’d already made, instead of starting over.

After everything was uploaded, I started a thread for it to write its own project instructions.

This is an example of what I used:

I kept that thread open. Every time I added a new document, I uploaded it there and asked it to update the prompt. Then I replaced the old one.

If you’d rather keep everything local, you can run this out of a folder on your computer using Claude Cowork, with your files, a memory folder, and an instructions file the AI edits directly as you go. Make it a synced folder in Google Drive or Dropbox and your team can get to it too. The upside is portability. Your brain lives in files you own, so switching AI platforms later is relatively painless.

With all of that in place, I was ready to write the first page of the website. The homepage is your storefront and the foundation everything else gets taken from, so I did a fair bit of research on pages I liked and what I wanted to borrow from them.

Then I told the brain what we were trying to achieve. Educate prospects on a pain they don’t know they have yet, then introduce the solution, then the benefits. Because we’re blue ocean, most of our prospects don’t fully understand the problem they have, which is why we need that order. Yours might be different. You have to decide what matters for you, based on the frameworks you trust.

Knowing the goal was education first, it proposed a structure for the page and I agreed with it. Hero, status quo, problem, solution, solutions by role, features, comparison, FAQ, CTA. I told it I wanted customer quotes woven throughout to support the story, instead of a testimonial wall.

Then it wrote the whole page based on everything I’d given it, and I went back and forth with it on every section until I was happy.

The most important part was the hero title and subtitle. That’s how prospects decide what category bucket to put you in. Nothing was ever set in stone, and everything changed over the following months as new ideas came. Because it had all the context, every suggestion it made was an informed one. Every conversation was like talking to a personal senior consultant who remembered every detail about my business since day one.

Once the homepage was 90% of the way there and I felt good about the language, I stopped. I didn’t want to write another piece of content until I knew we were heading in the right direction. Everything up to this point was still a hypothesis.

I sent my COO to a conference to test the messaging on real prospects.

I had the brain write him a messaging and conversation guide to study beforehand. Old messaging versus new, and why. What to say, what not to say, what to try. How to answer “what do you do?” How to handle the objections and what to listen for.

What we learned there was how much tiny changes in language change the way someone reacts to your entire offering.

When someone asked what we did, we used to say “automated webinar software.” Only a narrow group of people, usually marketers, knew what that meant. Most would follow up with “what does that mean?”, and we’d spend the next ten minutes explaining, only for them to say “we don’t do webinars, this isn’t for us.”

At the conference, we tested “on-demand video software with chat.” People understood it right away.

With real feedback behind the positioning, I felt confident about the direction and ran the same process on everything else. Pages written for roles like sales, marketing, and customer success. Pages for industries like SaaS and proptech.

Same method every time. Where pages covered similar ground (like FAQs), I asked it to paraphrase so no two read the same.

After each page was finalized, it went into the AI project and the prompt got updated, so the brain always had the latest version of us.

While I was working through the web page rewrites, I shared the project with my team and they took on updating everything else. Customer emails, blog posts, anything still carrying the old positioning. They didn’t need me to brief them, because the brain already knew the rules. That’s when it stopped being my project and became how the company writes.

Once all the content was finalized, I wanted an unbiased product-market fit expert to review our messaging and tell me whether it landed for someone reading it for the first time. We operate in a silo, so what seems obvious to us isn’t obvious to someone else.

I found Carole Boutet on LinkedIn, who helps B2B tech and SaaS companies sharpen positioning and messaging. She went through the homepage and the core pages and told me where things were clear, where they weren’t, and how it read to her.

Only someone with zero context can tell you what isn’t working. Case in point: the tagline we’d worked on for weeks, the one we thought was awesome, didn’t fully land for her. We ended up swapping it for something from our demo script. The language existed, we just had it in different places.

This final step is not to be skipped.

It wrote our new demo script and explainer video script. Those two tasks would’ve taken a couple weeks to do, and it only took a few hours.

It writes new pages in minutes, usually very close to where I need it to be.

It fills in directory listings and edits app store profiles in seconds. It pulls quotes on demand. It writes replies to prospects in my voice when an objection comes up.

It’s an ongoing voice of reason. When I want to add something to a page that’s already done, I ask whether it belongs there. It tells me yes or no, and why.

For every new project I start, it becomes the source of truth for how we talk about the company. What to say, what not to say, and the customer evidence behind it.

For example, when I set up an SEO project for our agency, I asked the AI brain to write a language and positioning guide they could check their content against. I did the same for the newsletter project our marketing coordinator uses to write our emails, and for a growth advisor project I use to think through strategy.

The biggest win so far is that time on our website went up 50% right away. Here’s what we ended up with.

Here’s the thing… before this, I wasn’t confident putting effort behind any marketing activity, because most people didn’t fully understand what we did. I’d even avoid sharing our site with top prospects without a conversation first, because I knew a long explanation had to follow. Something was fundamentally off, and we held back on growth because we knew we’d be burning time and money on a message that wasn’t landing.

The project took five months in total, including a redesign, since the design had to accommodate all the new content. But we started using the new language on sales calls and in outreach emails long before the site was ready, and the difference showed up immediately.

Far fewer objections, if any, about what we do. The “why does this matter?” question has disappeared. Before, we were selling the exhaustion of doing the same thing over and over, and not everyone finds their work exhausting, so it was always debatable. Now we’re selling reach, and going live limits reach, which limits growth. That part isn’t debatable. Conversations got easier as it’s now about how it fits, not why it should fit.

We closed our first enterprise deal at $40k/year, against an average LTV of $2,650. That was huge.

Now that the new messaging is everywhere, we’re finally confident enough to put ourselves in front of more people.

Without product-messaging fit, you’ll never have product-market fit. You can have the best product in the world, but if people don’t get it immediately, they won’t be compelled to try it.

This project would not have been possible before AI. With a consultant, it’d be a year and a budget I didn’t have, and the output would have been worse, because it would have been a one-time deliverable instead of a system. Once the brain exists, it gets better every month, and you decide what else you point it at.

The thinking was mine, that part didn’t get outsourced. But the execution went from impossible to a few months of focused work, with iterations as needed ever since.

Want to build your own brain? Here’s the checklist to follow:

Collect these

Create these

Have the AI create these

Then, as you go

One last thing. Try pasting this article into your own AI project and ask it to walk you through building one. It’ll tell you what it needs.