An Interview With Chad Silverstein
From my personal experience, AI is still mostly and broadly used by software engineers and similar technical positions. The adoption rate of everyone else is very low. This is because these verticals are not very keen to adopt AI due to — as you mentioned — fears, misconceptions, and the lack of tools being built for their specific use cases.
In today’s tech-driven world, artificial intelligence has become a key enabler of business success. But the question remains — how can businesses effectively harness AI to address their unique challenges while staying true to ethical principles? To explore this topic further, we are interviewing Boris Lapouga.
Mr. Lapouga is the Co-Founder & CTO of Primetime, a human-AI collaborative platform that leverages multiple AI agents to help humans solve complex tasks. In a 2-decade career spanning from software engineer to CTO, he has transformed healthcare for 20 million patients in Ukraine by co-founding Helsi and led technology innovation at several successful startups. At his recent venture WorkHQ, he pioneered an AI recruiter agent that fully automates talent sourcing for staffing agencies.
Thank you so much for joining us in this interview series. Before we dive into our discussion, our readers would love to “get to know you” a bit better. Can you share with us the backstory about what brought you to your specific career path in AI?
Greetings Chad and the readers. Thank you for having me today. In my early 30s, I got a leadership role and was relocated to Ukraine, where I worked for a big corporation before co-founding a healthcare startup — Helsi, which became a national sensation, and today serves more than 20 million patients. This was a pivotal moment in my career and an important point to the conclusion of my backstory, because I realized what it means to bring real value to people.
I worked and lived in other countries since then — Spain, Germany, and back to Ukraine, seeking to create value for people through technology. I left Ukraine in 2022 when the war broke out and I was displaced, bringing me to the United States.
When joining my previous company WorkHQ, we had a mission of helping recruiters find great candidates and save time rummaging through many irrelevant profiles or CVs by leveraging big data and AI.
We’ve been building internal ML models that would classify job titles, seniorities, industries, and infer various skills and professional experience on pieces of information we had about a certain person, crafting beautiful profiles.
We soon realized that while our platform is great as is, we could use AI to fully automate this complex and nondeterministic workflow — sourcing people for a job. We built an AI Recruiter that did just that. I felt again that I’m bringing something very valuable to people, meeting real customers and hearing all that feedback from them.
That’s when I understood the true value of AI in today’s world and co-founded Primetime with my ex-colleague and a friend Zach Lupei. Sharing the same vision and ambitions, we decided to focus on a much more horizontal problem, utilizing AI to its fullest potential, and helping people work more efficiently with multiple AI agents at the same time, boosting their productivity. I believe AI is a human’s exoskeleton and not a replacement.
Can you share the most interesting story that happened to you since you started working with artificial intelligence?
At WorkHQ, we faced a challenge of inconsistent job titles in our database of over 200 million profiles. Let’s say a recruiter is searching for “Senior Software Engineer”, but people title themselves as “Senior Software Developer”, “Senior Developer”, “Sr. SWE”, etc. You can’t expect an LLM to predict all possible variations.
After manually curating 10,000 normalized titles — I must say, your custom ML solution is as good as your training data — and data labelling is the hardest step.
Experimenting with various classifiers like XGBoost, Random Forest, SVM with varying results, I decided to build an ensemble model that uses all 3 together. I didn’t have enough experience in stacking ML models before, only using them separately.
“Can AI build AI?” I ask myself. I set out to test it myself. Using Cursor and a lot of prompting from different models such as GPT, Anthropic, and Gemini, I was able to construct a working ensemble, AI can build AI with human supervision, and if you know what to expect as the output, or course correct if it’s hallucinating.
I entered the training and validation phase. To check if your new set parameters were a good choice, I had to train the model and test it. It takes a lot of time if you do it even on the most powerful Macbook. So I started training it on my gaming PC, with a 4090 GPU which resulted in very fast training.
Unfortunately, the results were a lot worse, though there are many reasons as to why, one of them as I’ve learned is that I didn’t know back then about lower-precision arithmetic parameters. So our best model at that time was trained on a CPU and it does a great job even today.
The moral of the story — even if you have limited to no experience in AI/ML space, you can start with simple problem solving through the innovative tools at your disposal today, but you must read, understand and double-check what it is you are feeding to your models and how it’s going to impact your training. It’s never too late to get into AI/ML, you don’t have to build the next ChatGPT, build simple tools first!
You are a successful leader in the AI space. Which three character traits do you think were most instrumental to your success? Can you please share a story or example for each?
- Surround yourself with curious-minded people: I try to hire people who are not just “group-think individuals”. I look for a unique superpower in each candidate — whether it’s expertise in data analysis, deep understanding of complex problem solving, or a knack for generating great hypotheses and ideas.
This approach as a leader allowed me to create teams that were fueled by their own goals and ambitions that would align with the business. They would dive deep into the most challenging topics and come back with solutions or sometimes failures, but with invaluable learnings.
If you want to succeed as a leader, especially in a space like AI, which is driven by hard work, dedication, and curiosity, hire people that are smarter than you or people that will challenge your ideas. - Make a deal with your inner (“innie”) impostor: I read an NIH study indicating approximately 80% of people suffer from impostor syndrome, which I also see among my colleagues, friends, and family. I feel that in myself. This is our biggest inner doubt voice, our biggest procrastination driver that paralyzes us.
Engaging in internal dialogue is one of the strongest psychological feats that can help you regulate your cognitive behaviour.
I overcome self-doubt by engaging in an internal dialogue. I ask myself what my fears truly are, assess the risks, and then decide to act regardless. This practice has helped me transform hesitation into informed risk-taking. - Listen to your team, don’t talk: This is primarily addressed to other leaders. When we take on a mantle to lead people, whether it was verbally said to us or not, we commit to help our people to thrive, make them feel safe, heard, understood, and respected. Telling people what to do, bending them to your will, and expecting success are not attainable. Your success is the value your team generates. You must create a compounding effect — enable the people closest to you, delegate and expect them to do the same with their direct reports, eventually, when it comes down to the individual contributors, it is only then your organization starts getting real returns.
I create an environment where team members feel safe to speak up. By asking open-ended questions and demonstrating that mistakes are learning opportunities, I build trust and encourage honest feedback..
Let’s jump to the primary focus of our interview. Can you share a specific example of how you or your organization used AI to solve a major business challenge? What was the problem, and how did AI help address it?
When building our AI Recruiter agent designed to fully automate talent sourcing, we faced the challenge of accurately labeling candidates’ responses and routing them to the correct stage of our pipeline. We manually categorized the responses into groups: “positive”, “negative”, “inquiries”, and “do not contact”. We employed the LangChain framework to create a sub-agent that analyzes candidate responses and determines sentiment based on extensive synthetic data generated for each category. As a result, we automated email responses: we send a thank-you note for negative response, provide the candidate with a calendar link to book an interview if it was positive, try to answer the questions they had from information provided by recruiters, or mark candidates who do not wish to be contacted anymore and remove that profile system-wide. After speaking personally with our customers, I learned how much time they save on initial outreach to candidates while ensuring a personalized candidate experience.
What are some of the common misconceptions you’ve encountered about using AI in business? How do you address those misconceptions?
The biggest misconception is that many businesses believe they can replace humans with AI. Some do it for engagement on their posts, others exist in a sphere of disinformation.
What many seem to not understand or unwilling to, is that the current state of AI, particularly Large Language Models (LLMs) are essentially sophisticated pattern recognition tools, that recursively try to find the best next word when generating an answer to you, and many times the responses are as good as your request.
If you missed a crucial point, it will not be able to respond to that properly. LLMs are also only able to respond to what is in their training data, meaning past human knowledge.
If you had to visualize LLM knowledge corpus, you’d see a “swiss cheese”, where empty holes mean the areas where it lacks certain knowledge and will hallucinate (provide wrong information) in the most convincing way.
Some companies like OpenAI, Perplexity, and Grok are using real-time search and data analysis, allowing the models to benefit from current knowledge as well, but if you don’t explicitly guide it where to look for that data, it will come up with search criteria that will not yield good results either. After all, they all still use old-school search engines like Google to pull up the information, and very often that information reflects the author’s bias or subjective perspective.
At Primetime we take a different approach, we explicitly say that we do not intend to replace humans, but rather allow them to access information and knowledge faster, addressing the “swiss cheese” problem through various bleeding edge solutions such as knowledge graphs (where your internal communications and documents are constructed like a graph with rich relations between them) and vector databases (that pull data based on your query and whether there’s anything in the storage close enough to it semantically).
The recent rise of Agentic AI tries to address these issues differently — Agents are being trained or configured to solve very specific vertical problems, minimizing the risk of hallucination and providing better results, but we still need humans to supervise those decisions or outputs to make sure they make sense.
To apply AI effectively in your organization and stay away from common misconceptions, review the tools you employ and have your employees verify the output before turning LLM or Agent response into a grounded fact or a decision in your organization.
In your opinion, what is the most significant way AI can make a positive impact on businesses today?
Over the last decade the market was oversaturated with SaaS platforms, just like we have many entertainment streaming services (Like Netflix, Hulu, Max, AppleTV) right now to choose from and pay for each one separately, which not only leads to higher expenses but also makes it almost impossible to share data you generate and store in those SaaS tools.
If you have for example an HR system, support ticketing system, CRM, analytics database and other crucial business tools that do not have a native or easy support to transfer data between each other, you may find yourself in a high friction environment copy & pasting a lot of information, which is prone to mistakes.
With the right tools and human supervision, AI can eliminate the need for too many SaaS tools in your toolbelt and instead streamline a specific problem solving within one system or help you build data transfer between those tools.
AI can also supercharge your employees to be more productive. Automatic workflows, doing research, producing working documents faster, reviewing documents, extracting data, or even answering your customers’ questions from your knowledge base.
This doesn’t mean you need less people, this means the same amount of people can produce more valuable work and accelerate your business.
Ok, let’s dive deeper. Based on your experience and research, can you please share “5 Ways AI Can Solve Complex Business Problems”? These can be strategies, insights, or tools that companies can use to make the most of AI in addressing their challenges. If possible, please share examples or stories for each.
- Cursor — As software engineering remains the hardest job of any online business, the cost and time to market associated with it is the biggest concern of any company, especially a startup that needs to move fast. Cursor is an integrated development environment that reads your code and allows your engineers to produce new artifacts faster, while still allowing them to supervise the outcomes. When my team adopted this tool, this allowed them to look into some old code (usually called “legacy code”) if we needed to fix something and nobody knew how to do it. It also allowed interdisciplinary engagement, for example, a UI engineer would all of a sudden be able to address some server-side (backend) issue without the need to learn this craft for years.
- Perplexity — it’s a great new way to search for information, which can entirely replace the need for Googling. With the recent addition of “DeepResearch” (same can be found in ChatGPT Pro or Grok 3), you can now analyze many data sources and get a well-crafted research report that you can share with your colleagues or discuss with AI some of the points from that report to learn more about it. It allowed me to do a lot of research about immigration, technological solutions, law, and financial markets. As you can see, the application of this tool is very horizontal, every person of any walk can benefit from it instantaneously.
- If you run a legal company, there are many legal AI solutions, among the ones that made it to the news — HerculesAI, Luminance, etc. Humans train for years to be attentive to details, but even the best of us can miss something. These tools are custom tailored to specific use cases and help you avoid mistakes by reading the contracts or generating them, creating addendums, keeping track of your court dates, cross researching the law, and helping you argue a case. This is just an example of a vertical, highly specialized AI solution that can be found today, but it’s available for nearly all departments you could think of in any company.
- v0 — Unlike Cursor, this tool by Vercel specializes in creating UI code from designs. If you are prototyping a new product or you need to create something fast, this tool has saved us a lot of time. We’d just feed it our new designs; with a little back and forth, it would give us an interactive UI within a few minutes. We were able to feel what the user experience would be like before either disposing of this solution or adopting it into our main product. And all of this can be done without engineers, just during your marketing-product brainstorming sessions.
- To conclude with a fifth point — I explicitly named 4 different verticals to showcase how horizontally spread AI solutions are. The strategy to adopt AI in your company is to assess through each head of department what they can benefit from if something could’ve been automated. Research available tools on the market, ask your employees what tools they are familiar with, choose the right tool for your mission, and test it. Not everything that the tool promises to do, is gonna do it well. Before committing to a year-long contract, you have to ensure you are getting the promised value or the tool is just not ready yet to serve you. Stay critical, but open up your mind to allow all these amazing tools into your organization if you truly benefit from them. If your data is sensitive, make sure to research where your data is being stored, how, and what control over it you have. Last thing you want is losing your ISO certificates or data leakage in general.
How can smaller businesses or startups, with limited budgets, begin to integrate AI into their operations effectively?
I mostly work in small businesses or startups. I am building my own AI startup right now. Large Language Models become a commodity, they plateau in their “intelligence” progress, which makes them very affordable to small businesses.
Very sophisticated enterprise level AI solutions usually leave SMBs behind, creating new markets to be picked up by rising competitive solutions. If ChatGPT Pro decides to lock their DeepResearch function behind a $200 a month subscription, then Perplexity or Grok will give pretty much an on-par solution for just $20–25 a month.
If you are looking to engineer your agents — the cost of tokens today is very low and you don’t have to “fine-tune” and train your models like the big players do. There are cheaper ways to enhance any frontier model with context-infused knowledge, such as already mentioned — by building knowledge graphs or using vector storages.
AI and access to it is so affordable today, that even my CEO is producing decent code and allowing our startup to move faster.
What advice would you give to business leaders who are hesitant to adopt AI because of fear, misconceptions, or lack of understanding?
Don’t transform your entire company to some ambiguous AI solutions. Start small, on a team level or individual levels. Take your best performing team in any department and ask them to research and propose a few AI tools that can help them do their job even better. Test locally, draft conclusions and move to the next department.
It’s important not to test on your lowest performing team, because adopting AI can send a wrong signal to them, and they won’t be as engaged trying it out and proving it can improve their performance.
If you fear that your data is going to be used for training purposes, there are very affordable AI models to be deployed in your server infrastructure, such as AWS Bedrock. You can keep your data inside, while still benefitting from AI transformation.
In your opinion, how will AI continue to shape the business world over the next 5–10 years? Are there any trends or emerging innovations you’re particularly excited about?
I’d be lying if I said I knew what will happen in the next 5–10 years. The rate of AI evolution and adoption is so rapid, people can barely manage to keep up with the news.
One emerging trend is, of course, Agentic AI. I believe this will remain relevant for years to come and will only become better and more efficient. Workflow automation, UI-less (no User Interfaces) task execution, document composition or summarization and many more will shape how companies work. I even know immigration law firms that use AI to draft cases for their customers based on submitted evidence.
The other emerging pattern is knowledge centralization. For far too long, every company’s knowledge has been scattered across multiple tools that are not interconnected. There are multiple players on the market right now that try to bring knowledge into one place, allowing employees to easily access it and perform their duties more efficiently.
How do you think the use of AI to solve business problems influences relationships with customers, employees, and the broader community?
From my personal experience, AI is still mostly and broadly used by software engineers and similar technical positions. The adoption rate of everyone else is very low. This is because these verticals are not very keen to adopt AI due to — as you mentioned — fears, misconceptions, and the lack of tools being built for their specific use cases.
I’ve built AI tools for customers who would otherwise not have access to AI, and won’t be able to build it for themselves. The more companies will start serving the non-technical public, the better the AI adoption will be.
This approach will of course positively influence the relationship with customers that benefit from your solutions.
Employees will collaborate better, as everyone will be empowered by the same tool, potentially exposing redundancy in some jobs, while creating new positions at the same time.
You are a person of great influence. If you could start a movement that would bring the most amount of good to the most amount of people through AI, what would that be? You never know what your idea can trigger. 🙂
I don’t want this to sound promotional, so I’ll use it without a name — this is exactly the reason I currently co-founded my new startup. I’d like to create a movement that democratizes AI. ChatGPT and similar solutions are not enough, they are merely a demo of the AI power. We need to use these frontier models to serve all kinds of customer bases, not only technical people. We need to put an end to complex user experiences, endless forms and workflows that are confusing and make AI more natural to us — a digital co-worker that helps us do stuff.
How can our readers further follow you online?
I mentor people through https://adplist.org/mentors/boris-lapouga
Sometimes I write articles on Medium https://medium.com/@boris.lapouga
Or post just have fun on LinkedIn https://www.linkedin.com/in/blapouga
This was great. Thank you so much for the time you spent sharing with us.
About the Interviewer: Chad Silverstein is a seasoned entrepreneur with 25+ years of experience as a Founder and CEO. While attending Ohio State University, he launched his first company, Choice Recovery, Inc., a nationally recognized healthcare collection agency — twice ranked the #1 workplace in Ohio. In 2013, he founded [re]start, helping thousands of people find meaningful career opportunities. After selling both companies, Chad shifted his focus to his true passion — leadership. Today, he coaches founders and CEOs at Built to Lead, advises Authority Magazine’s Thought Leader Incubator.
Boris Lapouga of Primetime On How Artificial Intelligence Can Solve Business Problems was originally published in Authority Magazine on Medium, where people are continuing the conversation by highlighting and responding to this story.