Welcome to Carl’s Newsletter.

In today’s issue:

• An interview with BuildBetter’s CEO and Co-founder Spencer Shulem about building products for product managers (yes it’s meta), and the intracacies of building with LLMs

• News from across the product-verse: More layoffs, and Apple’s nefarious application of Dark Patterns against Epic

Not subscribed? Sign up here.

Today’s post is sponsored by Alex Rechevskiy. He’s an ex-Google Group PM turned top-tier product content creator and career coach.

This month, he launched the Product Career Accelerator. It’s an all-inclusive monthly program designed to help you land the right next product role, with at least three live sessions every week, including:

• Resume review

• Interview & negotiation prep

• Application & Follow-up Methodology

Alex’s clients have received offers from Google, Facebook, and Amazon. Others have negotiated $50k - $100k+ salary increases during the offer process.

Every spot in the January cohort sold out almost immediately, but he’s enrolling students for February now!

Lessons from BuildBetter.ai on Building AI Products Better

Building products for product managers is hard. There’s no shortage of problems to solve for us (😢), but there are big barriers:

1. PMs are extremely busy

2. PM problems are hard to solve

3. Integrating new tools takes time

AI has a huge opportunity to help with some of the deeper problems of product management – writing great documents, improving cross-functional communication, making information more accessible.

Lots of companies are building in this space, but one has stood above the rest – BuildBetter.ai.

Today’s interview is with the CEO and co-founder Spencer Shulem about building AI products and products for product managers.

Yes, this gets a little bit meta.

Carl: What's the origin story for BuildBetter?

Spencer: My background was in product, I was Head of Product and consulted for companies around building product, and around 2020 it changed with COVID and remote work. We saw three trends:

1. Companies are seeing significant revenue growth from focusing on increasing retention, usage, and product expansion than ever before (often over new-sale/marketing growth).

2. AI will get good enough to solve these problems for the first time in product teams.

3. Companies are (finally) giving product teams not just responsibility but now authority over their own budget for tooling (and they’re remote).

We knew there was something for product people to build, but we didn’t know what. So we did what any PM would, we kept talking to users, and found this new “Product Ops” role emerge, so we took a bet, and started to iterate in that direction.

I’ve probably tried a dozen or so “AI for PM” tools, how is yours different?

Anyone who's used any AI tool has experienced this: “Oh, crap, I actually can't do what I thought I wanted to do with this thing! It doesn't work as well as the demos at all.”

So when we started building our product with AI at the forefront, we said “it needs to solve a hair on fire problem, it needs to be problems that product managers face every single day, and it must tangible value that is usable today.”

How do you define the difference between product management and product operations?

Product management is like strategy, prioritization, planning, stakeholder management. It’s kind of like the glue between organizations.

Product Ops is everything you need in order to do those things. In some ways, it changes based on your organization. For a B2C product, you might be spending more time on data. For B2B, it might be more process management, like organizing and managing the processes within your company and enforcing specific structure.

Either way, imagine a user research call happens. How is that accessible to the product managers? How do we get information out from sales onboarding calls? What new feedback system or process do we need to kind of have?

It was best explained to me by a Product Operations leadership as, “the success of my role is the success of the product team, not the product.” Where the inverse is true for Product Managers.

So BuildBetter focuses on product operations. What does that look like?

We're not here to do roadmapping, we're not here to do strategy, we're not here to do planning.

A lot of tools within product management today are trying to do that, like trying to automate your roadmap or trying to automate your strategy, or trying to automate your planning, trying to do product management. And our stance is that product managers like doing product management, but they don't like doing operational work.

So we're not going to have an opinion on how you run product within your organization. We are going to have an opinion about how you do operational work within your organization.

What was the first product ops problem you decided to tackle?

We really focused on call recording. If you talk to product managers, you find that they don't really have time to deal with and learn new tools. They're constantly in calls back to back to back.

As the saying goes, “be where your customer is.”

Our whole focus was: How do we help them go from these calls that they're constantly in, whether it's a customer call or team call or product call or a stand up or whatever it might be, to turning that into something actionable and insightful?

There is a ton of actionable insights from user research calls:

• What questions were asked? Were they bias? 

• What issues are the customers having?

• Can you find all the times customers have complained about this?

And meta-information product managers might want from every call. Things like:

• Was this call efficient?

• Did I actually do a good job running this call and what can I do better next time?

• Who interrupted the call the most?

• Who derailed this call?

These are really specific questions that PMs want to improve our their process, but they struggle. But it's available to us now, and we can just ask these questions.

Can you explain more what you mean by actionable and insightful?

If you give ChatGPT a brunch of information and say, “write me a PRD,” it will write a fine PRD or give you an outline for one, but it's actually not usable. You couldn't actually send that to someone. It looks like a lot of words. It feels like it's something usable, but if you actually read through it, you're like, I'm going to have to delete 80% of this and start all over.

Our goal is to create outputs that are immediately usable.

For example, if I wanted to generate a user persona based on calls.

I literally pick a call with a customer, and have it generate a user Persona based off calls with them. And these are, copy and pastable user personas. They have specific quotes from that customer. They have a demographic based off of their history.

These are real tangible assets that I can give to my team and say, “Hey, you weren't on this call, but here's all the questions I asked. Here's all the answers, and here's a breakdown of how that's going to impact you or how it's going to be used."

From there, what did you do to help product managers get more insights from their calls?

A feature called CustomContext. It's a multi page document that companies fill out, which basically embeds everything you do within your organization into your BuildBetter AI.

So let's you say you want to write a PRD about this specific feature at your company. You don’t have to say "Oh, and by the way, this is what our company does." 

CustomContext paired with call data helps BuildBetter know what call data to retrieve, even if you don't ask it to.

It's like talking to a Product Ops person at your company. It has that information on hand, ready to answer any question you have. It’s magical.

So it knows when to follow up with questions?

Early on, we really focused on the “Question Machine.”

A huge focus for us was training our AI, "If you don't have the answer, ask the questions that you need to understand to get the answer."

And so if you say, write me a PRD about XYZ thing with no context, it will say, "Can you provide me with more context about what you want to write or how you want to write this, or any calls or data that you want me to use for this?" It's not just going to say, okay, here's a PRD.

It sounds like you’re fixing many of the weaknesses of ChatGPT for product work. How did you decide to solve these problems and iterate over time?

When we first used our product and we got it out there, we didn't use it. It wasn't that valuable.

We kept wanting to use it, and we have every incentive to use it, but it was just like, this doesn't provide anything for us, we don't have any reason to use it. We see this a lot with AI tools now. So we literally just kept using our own product and adding and changing the features until we wanted to use it.

And we saw that when our usage climbed, our customers usage climbed.

Now we have companies that are 20-2000 people or so, and their PMs are using us at least 15 times a day. So there's all these use cases that we start seeing companies start using our product for that we never thought of, so we start, and then we start building out custom  workflows for things like “Smart” documents or “Auto-updating” project briefs.

But I think the breakthrough for us was asking the question: “Do we want to use this? Is this something that we can use five times a day?”

What's the long-term goal for BuildBetter?

Our goal is to run the entire operational workload for product teams. Right now that looks like assisting with Product Ops, but there's so many different Ops related tasks within product teams. We see design organizations adopting our tool, we see user research departments adopting our tool, onboarding teams, sales teams, support teams adopting our tool.

Our focus is expanding our integration capacity with different tools that aren't just your call related data like Slack, which is coming very soon so we'll see a lot of that type of expansion.

The other aspect to it is expanding our use cases because we look at product as the glue between every department within the organization. And what that means is the adoption of BuildBetter has to be clear for every department within the organization. 

Can people try BuildBetter today?

Yes. No waitlist, no demo, no hop on a phone with me to get access. You can sign up for it right now and actually use it as a Product Manager, or as a Product Operations person, and it will make a significant impact on your life. We're not selling a pipe dream, something that will maybe work in five years. You can sign up right now and within your next call, get value with our product.

(Trials are granted 15 hours of recordings and uploading).

Takeaways:

1. Solve an urgent, hair-on-fire problem your target users face daily. Don't try to automate strategic work they enjoy.

2. Integrate tightly with existing workflows and data sources to become a seamless part of users' flow.

3. Dogfood intensely and only build features you would use multiple times daily. Usage indicates value.

4. Iterate rapidly based on dogfooding learnings to create tangible value as soon as possible.

5. Building products based on LLMs requires solving the most common problems with them: that they answer questions before they have enough information, that they don’t provide actionable outputs, that they “overprocess” information instead of providing direct quotes.

Again, here’s the link: BuildBetter.ai. Spencer’s graciously offered the promocode GOATPM for 15% off the first 3 months if you want to see all this for yourself.

From Across the Product-Verse

T’was another exciting week in the world of product.

1. January Tech Layoffs: The January layoffs are coming right on schedule, with the likes of Google, Amazon, and Discord letting go of some staff. Fortuntaely, they aren’t as bad as last year, and hiring is up. Here’s a cool resource I found with a catalog of every main tech layoff, along with articles that of analysis. These are all good, if sad, case studies on product strategy and current business trends.

   

2. Bad Apple: Last month, Fortnite creator Epic won a massive dispute against Apple over the “Apple Tax” Apple charges for every in-app transaction. While Apple is technically complying with the ruling, they’ve done so in a way that effectively negates the ruling. It’s a an absolute masterclass in the application of dark patterns – product design used against users.

That’s All For Today

Last things:

1. Sponsors: If you or your company is interested in sponsoring this newsletter, I’ve posted sponsorship information on this snazzy new sponsorship page.

2. Upcoming AI course: I am getting close to finishing my first-ever paid course “AI for Product Managers” which goes in-depth with tactical ways PMs can leverage AI tools to save hundreds of hours at work. If that sounds interesting to you, drop your email in this 1-question form and I’ll send you updates, sneak previews, and a discount code when the course is launched.

Here’s where else you can find me:

1. Follow me on X, where I post the most content, including lots of memes.

2. Follow me on LinkedIn, where I post longer-form content.

3. Follow me on Instagram if you just want the memes.

And don’t forget that as a subscriber, you get free access to:

1. My massive database of resources

2. A huge list of ChatGPT megaprompts to help you with product work

3. A 60-day guided tour of the best product content ever written