Kazakh Founders Built an AI Presentation Tool and Raised $4 Million in the US

Alibek Dostiyarov and Yersultan Sapar studied together in the US before going on to build careers in different fields. In 2024, the two Kazakh founders launched Perceptis, an AI-powered service for creating business presentations. Today, the platform has around 800 paying corporate clients and 15,000 users.

For the joint Digital Business and Astana Hub project “100 Startup Stories from Central Eurasia,” Alibek Dostiyarov shared how Perceptis grew from a tool designed for US consulting firms into a global product, why the company hires most of its team in Kazakhstan, and how the startup manages to compete with some of the biggest AI services on the market.

“We wanted to solve one problem for consultants, but ended up opening the product to everyone”

– Alibek, how did you come up with the idea for Perceptis?

– Yersultan and I studied together at UC Berkeley. After graduation, we went in pretty different directions. He went into AI at Apple, while I joined McKinsey and built my career in consulting. At some point, we started talking about a product that could help people put together presentations and website mockups. We built the first prototype in just two weeks and applied to Y Combinator. We didn’t get in, but we made it all the way to the final interview with Garry Tan, YC’s CEO. We applied again later and once again got invited to the final interview. By that point, though, we had already accepted an offer to join The House Fund’s AI Accelerator, a program for AI startups. Taken together, those experiences made us feel we were onto something and heading in the right direction.

But pretty quickly, we realized the product was trying to do too much. So we narrowed the focus to one specific use case: automating proposal decks for consulting firms. A consultant would enter the key details, and Perceptis would use AI to turn them into a finished presentation for a prospective client.

– How did you get your first customers?

– I’d find a managing partner at a consulting firm on LinkedIn, send them a connection request, and write something like: “Hi, I’m Alibek. I used to work at McKinsey and was involved in selling $40 million worth of projects. Would be great to jump on a call. I’d love to tell you what we’re working on.” By the time we got on the call, many of them had already looked at my profile and would ask, “What’s Perceptis?” That gave me an opening to explain the product and convince them to buy a subscription.

Our first client paid us around $3,000 over two months, and the new projects they won delivered a return 32 times greater than what they had spent. Perceptis was turning proposals around so quickly that the consultant would deliberately wait before sending them out. He told us: “Guys, I get the presentations from you and then sit on them for a day, just so the clients don’t think I’m sending them some unchecked nonsense.”

Back then, even ChatGPT couldn’t do anything like that, so people genuinely thought it was magic.

– But then a lot of AI models learned how to create presentations. Did Perceptis evolve too?

– Yes, and it really came from the users. Once they saw what the product could do, they wanted more from it. We started hearing things like, “We’re winning projects so quickly now that we need something to help us get the work done.” That’s what pushed us to broaden the product and move from sales proposals to presentations on pretty much any topic.

For a while, though, Perceptis was still very much a B2B product. We onboarded clients manually, set up their accounts, and helped them upload their data.

Then, in January 2026, we opened the platform up to everyone. Now anyone can sign up, choose a paid plan, and start using the service straight away.

One of the main reasons was that we were curious to see how useful Perceptis could be outside the industries we had originally focused on. And it turned out there was a lot of demand. We started getting plenty of corporate users from engineering, tech, finance, and pharma companies, as well as research organizations.

Today, Perceptis is a powerful engine for creating sophisticated presentations on pretty much any topic.

– If someone opens Perceptis for the first time, what does the process look like from the initial prompt to a finished PowerPoint?

– There’s just one box. You type in your prompt, hit send, and the presentation comes back.

If you’re making a presentation on a general topic, you don’t need to upload anything. Perceptis pulls from public sources. But if you want the deck to be based on your company’s internal data, you can upload existing reports, financial models, or other Excel files.

The system takes about three minutes to think things through and then gives you a slide-by-slide outline. At that point, you can say what works for you and what you’d like to change.

Then you hit “Confirm,” and about five minutes later, you get the finished presentation as a PowerPoint file.

– How much does a subscription cost?

– There’s a free plan that gives you 10 slides a month. Paid plans start at $29 a month. Pro is $129, while Business is $390. Larger companies can go with the Enterprise plan, which is priced individually.

“We grew our revenue 4x in just five months”

– How did opening up the platform to everyone affect your revenue?

– Unfortunately, I can’t share the actual numbers. We only disclose those to our investors. But since opening Perceptis up to everyone, we’ve seen revenue grow very quickly. It doubled in the first two months, then doubled again over the next three.

Today, we have 15,000 registered active accounts on the platform and around 800 paying companies.

– What helped you attract users?

– Partly through advertising. But we’ve also had some pretty unusual viral moments. At one point, we noticed a huge number of orders coming from Japan, and we had absolutely no idea why. We weren’t running any ads there. We couldn’t see individual users’ data, but we could see the topics they’re asking about. That’s how we realized people were using Perceptis to prepare corporate documents for senior executives.

Later, we found a few Japanese influencers on X who had discovered the tool on their own and started talking about it.

– How did you fund the company, both at the start and later on?

– At first, we funded everything ourselves. Our first engineering hire, for example, was a student at Nazarbayev University. He was incredibly talented and way more experienced than you’d expect for his age. In the beginning, we paid him out of our own pockets.

But we knew from the start that we were building a pretty complex product, and bootstrapping it all the way would be almost impossible. So we started looking for outside funding from day one. Altogether, Perceptis has raised $4 million.

We raised our first outside investment in spring 2024, when The House Fund and Silkroad Innovation Hub came on board. By August, we already knew we’d soon be going out for a full seed round, so before that we brought in a few more friendly investors. They included MOST Ventures and several angel investors.

The next step was a $3.1 million seed round, led by Streamlined Ventures.

More recently, we closed a small bridge round backed by Kazakhstan’s Astana Hub Ventures and the Silkroad Angels investor club.

– Why did you decide to raise a bridge round before Series A?

– We’re planning to raise a full Series A by the end of the year at a higher valuation. Before that, we wanted to bring in a few funds and angel investors we already had a good relationship with, so they could get in a little earlier.

As for Astana Hub Ventures, we think they’re one of the most exciting venture players out there right now. What really stands out is how active they are. They’re already in talks with some very promising US startups, including founders that hardly anyone has heard of yet, but who are already on the radar of investors from Kazakhstan.

– So even though you’re building the business in the US, Perceptis still has pretty strong ties to Kazakhstan. What does that look like beyond investment?

– First and foremost, through the team. For a long time, there were only three of us. Yersultan is basically a whole army of engineers on his own, so early on we didn’t need a big development team. Now we’re up to 15 people, and 12 of them are based in Kazakhstan. We’re making the most of the opportunity to hire back home, where there’s a really strong pool of talent.

It’s tough to hire local engineers in Silicon Valley right now. Meta, OpenAI, and Anthropic are offering compensation packages that are very hard for a startup to compete with.

Once we raise our Series A, we want to roughly double the size of the team, and we’ll be hiring in Kazakhstan. We’re also planning to register a company there and apply to join Astana Hub.

“If you’re happy with what Claude can do, then you’re not our customer”

– Perceptis isn’t the only tool for creating presentations. What gives you an edge over the competition?

– We win because we’re the best. And I’m not just saying that. Between May and August this year, several independent rankings of presentation tools came out. In August, for example, Perceptis took the top spot in the enterprise category of Design Arena’s crowdsourced benchmark, ahead of Claude Opus, Gamma, Manus, and other tools.

Another independent benchmark came from Caliper Lab. They put us head-to-head with Claude for PowerPoint on the same tasks. Perceptis scored 86 points, compared with 83 for Claude.

– What do you think helped you get there?

– We built a hybrid engine that’s really all about quality. Depending on the task, it mixes different AI models with our own algorithms to get the best result.

Behind the scenes, we’re using the same models everyone else can access: Claude, Gemini, GPT, and others. The difference is what happens after you send something into our pipeline. We break the request down into a whole series of smaller tasks and route each one to whatever is best suited for it. Some parts are handled by simple rule-based algorithms, because there’s no point using AI where you don’t need it. If something calls for business reasoning, we might send it to Claude. If it needs strong visuals, that could go to GPT. And so on.

– Claude, ChatGPT, and other major AI tools can already create presentations as part of a regular subscription. So why pay separately for Perceptis?

– That’s exactly what we tell customers who ask us the same thing: look at the benchmarks and ask yourself what you actually need. Do you want the best possible presentation, or are you happy with what Claude gives you? If it’s the latter, then you’re probably just not our customer. But if you need something more sophisticated and really high quality, that’s where Perceptis makes sense.

– What’s next for the startup?

– For now, we just want to keep growing. We’re not thinking about an exit or starting another company. We can see that this product is working really well, and we want to keep building it.

The next step is to make presentations interactive, so they become documents you can “talk” to. You’ll be able to ask them questions. For example, you could click on a chart and ask, “How does this sales bar break down by region?” Or, “What happens if we change this assumption?” and get an answer straight away.

Right now, independent benchmarks put us ahead on quality. And we want to keep that lead, because our competitors aren’t standing still either.