How we’re using AI internally at Knak and what we’ve learned so far

When we launched KnakAI late last year, we were aiming to give our customers a better way to produce their campaigns with the power of AI.

But we also know that AI can help us unlock greater efficiencies internally at Knak as well. In this post, I want to talk about what we’re doing to leverage the benefits of AI inside our own company.

It’s not always evident. AI is so new, and it’s changing so fast, that it’s not necessarily obvious what we should be doing or how we should go about it.

Here’s what we’ve learned so far.

Get everyone on board

I like to say that at Knak, AI also means ‘all-in.’ We embrace AI to the fullest. And that is why we want to use it in our internal processes as much as in our client-facing services.

But that requires an internal cultural shift.

I have discovered that individual employees have differing approaches to AI.

Some are so AI-forward, they got excited without any prompting. These are the ones who jumped right into Codex or Claude Code and started creating things.

Then there are employees who are AI-curious, but unsure where to start or what to do. They want guidance but are eager to jump in with the right help.

There are others who are what I’d call AI-neutral. They’ll use Chat GPT like they would Google, but aren’t naturally inclined to go much further.

There’s a final category – people who are AI-resistant. I like to think that we don’t have anyone in that category at Knak!

If we want to leverage AI internally, we need to get everyone on board – not just the AI-forward people, but also the curious and the neutral. Using AI has got to be part of how everyone works.

We’re pushing that cultural shift through education and reassurance.

To get people knowledgeable about AI and what it can do, we’ve now instituted mandatory lunch-and-learns on the topic. We want everyone to know that if you work at Knak, you need to be using AI in your job.

We also have AI office hours, where anyone in the company can show up and ask questions of our AI experts.

Another thing that helps get people onboard is being able to reassure them that their jobs are not on the line.

It’s a legitimate fear.

When people ask, I tell them that because we’re a fast-growing company, we’re not planning to reduce the size of the team; we are simply more deliberate in our decisions to add to it. But I add that we need to become more efficient, and be able to do more with the same number of employees. AI can help us do that.

My feeling is that AI is not going away. So as a leader what I’m doing is preparing everybody on our team for the future of work, whether it’s at Knak or somewhere else. AI is here to stay.

Structure your internal AI team for success

Because the possibilities with AI are so great, you have to figure out what you want to do with AI internally.

We got the ball rolling by holding an AI hackathon, where people submitted ideas for what we could create using AI and then tried their hand at bringing those ideas to life in a day and a half.

While there was no shortage of ideas, and while the results were cool, they weren’t always useful or scalable, for a variety of reasons.

We realized we needed a more structured approach.

That’s when I turned to Hai Nghiem, co-founder of an AI consulting company called AGI Ventures. At his suggestion, we created an internal AI ‘dream team’ that was structured for success.

The team, which reports directly to me, has four main tasks:

1. build and maintain internal AI applications that scale;
2. educate our employees on how to use AI and how it can be helpful;
3. evaluate third-party AI products to make sure we buy the right technology; and
4. provide thought leadership while pushing the company on important AI initiatives.

The team includes three full-time employees plus AI champions from every department. (We have ~120 employees, for anyone wondering about the ratio of people we are dedicating to this initiative.)

In their role, our AI champions experiment and share knowledge amongst themselves.

They also encourage their individual departments to use AI.

As for the engineers, one acts as a forward-deployed engineer (FDE): part engineer, part solutions consultant. He is paired with two go-to-market engineers (GTMs) whose job includes, for example, finding AI solutions to tedious or repetitive work.

The FDE is responsible for the AI applications that we build – writing the code (with the help of AI) and making sure everything is consistent and of good quality. His focus is on building scalable and reliable applications.

One of the GTM engineers leading the team has been building AI applications for companies for two years now – a lifetime in the AI world. He has more of a business lens and focuses on what the application does.The other GTM engineer actually came from our sales enablement team and was focused on building internal AI applications to enable the sales and BDR teams. This person brings a business lens and a deep curiosity of AI tooling.

The AI dDream tTeam assesses all the ideas that have come in from the different departments, identifies which ones they want to pursue, and works with the business team to build the applications.

This is all very new for us; we put the team together only a few weeks ago and it hasn’t yet brought a project to fruition.

The first project it is working on is in marketing. We want to have the best AI marketing production process in the world, and we’re determined to build it here at Knak!

By the way, I believe that anyone thinking of creating an internal AI team needs to be careful about perceptions. It would be wrong for employees to believe that because the company has a dedicated AI team, they don’t need to use AI or be curious about it. The AI team is there to bring to life AI concepts that make employees more efficient.

And employees will need to be trained to use anything it creates.

Structure your data to support your AI efforts

Because our dream team is still so new, it doesn’t have a lot of history to draw on. We’re still learning and experimenting.

But our first big learning is that you have to structure your data to support your efforts.

That means bringing data from different systems together so that the AI and the agents can look at the data and assess the context well enough to be able to give our employees the information they need or do tasks for them.

It’s a cliché, but it does all start with the data.

Already, we have identified some gaps and we’re working to close them.

I am convinced it will be worth it.

One of the biggest advantages of AI is that an agent can understand all the data across all your different systems and make sense of it. Before AI, you’d need an army of analysts to do that kind of thing.

Restructuring our data for AI should bring big efficiencies!

In addition to data, one of the challenges I foresee is finding a place for all our new AI tools to live. Will an image generator be a ChatGPT plug-in or something else? How will people know where to find it? Will other things be set up on special servers that not everyone has access to? These are all problems that need to be solved. We’ll learn more as we go along.

Keep your end goals in mind

We have two very different reasons for wanting to boost the use of AI internally.

First, we want to make our company more efficient. Our stated goal is to make Knak 20% more efficient by the end of this year through the use of AI.

Our second goal is to keep Knak at the forefront of AI.

With regards to efficiency, we aren’t yet certain what metrics we will use to measure our progress; that should become clearer in the next few months. Right now, we are leaning towards CARR per employee.

One issue, though, has already started to emerge. We are beginning to wonder whether it might be better to buy an existing AI product rather than create our own.

We’re already having good success with Fin AI, a customer support agent that’s an off-the-shelf AI tool. And our customers, by the way, are having great success with KnakAI, an off-the-shelf (for them) AI tool for marketing.

Even though we are a software company, we never built any internal software. We always bought it off the shelf. The only software we built was the software we sold, which seemed to me to be the best use of our resources.

Does AI change that? That’s something I’m struggling with right now.

We’ll see what our dream team comes up with.

I’ve already told the team that we may end up circling back to our original premise: Buy internal tools, build anything we sell. At this point I just don’t know. Our AI team should be able to guide us.

As to staying on the forefront of AI, we believe that AI is going to fundamentally change the way people work. So in addition to increasing our own efficiency, we want to experiment, to learn what does work and what doesn’t, so that we can stay on top of things.

Do I feel like I’m always running a bit behind when it comes to AI? Probably; but I know a lot of CEOs in my position feel that way.

I also believe that when you are in the game early you learn a lot, and that will translate to benefits for the company.

There are plenty of innovative solutions that don’t exist yet. I hope that Knak’s team will help develop some of them.

I’ve been energized by the whole process of looking for internal efficiencies, and I’m very excited to see where it all leads! It feels good to me, as CEO, to know that we have people who are working every day to help the company keep its edge.

That’s beneficial to both Knak’s employees, and its customers.