How can artificial intelligence become a powerful tool to support sales in small and medium businesses? Danylo Liakhovetskyi, software engineer and founder of the startup Naratta, shares the story behind the creation of his product, attracting initial investments, and entering the U.S. market. He also discusses how AI is already reshaping the sales landscape, promising faster and simpler processes for companies and their customers.
Can you tell us about Naratta and the problems it solves? What inspired you to create this tool?
The idea for Naratta came to me while working as a software developer with my co-founder, Ilya Zubkov. We noticed a significant inefficiency in how time was used during work calls. In 2022, we decided to create a tool to address this issue.
The first version of our product was a meeting agenda feature for Google Meet, where participants could view the meeting structure and the remaining time for each topic. However, after conducting over 300 interviews with professionals—CEOs, product managers, and developers—it became clear that this approach did not effectively solve the problem.
We then moved to the second version, where I developed an AI assistant to moderate meetings. This assistant would alert participants when they deviated from the topic or exceeded their allotted speaking time. Unfortunately, this version also fell short of expectations, as the primary target users—managers—did not see the problem as critical enough to adopt the solution.
For the third version of Naratta, I introduced real-time meeting analysis technology, focusing specifically on sales calls. The tool is designed to assist sales managers, particularly those who are less experienced, in improving their skills. It provides live suggestions during conversations with clients, offers phrasing improvements, and warns them when the discussion veers off course. This not only increases the likelihood of closing deals but also accelerates the professional development of salespeople.
Even before launching this iteration, we received our first prepayments, which underscores the market's interest in our solution.
What kind of recommendations does the bot provide, and how can it outperform a human in this area?
When a salesperson lacks experience in client communication, it’s easy for them to panic, deviate from the script, or mismanage time. Meetings are typically very short, and maintaining the client’s interest is crucial. The bot helps salespeople stay focused and effective during calls. It analyzes conversations in real time and provides helpful suggestions. For instance, if a salesperson strays from the script or speaks for too long, the bot offers guidance on how to correct the situation. It might suggest specific arguments or rephrase sentences to keep the client engaged. This approach mirrors the support of an experienced mentor but without requiring human resources.
While developing the bot, my goal wasn’t to replace humans but to create a powerful support tool that boosts the confidence and productivity of salespeople.
How well does the bot handle conversation nuances and monitor tones to ensure accurate recommendations?
At this stage, our bot focuses on analyzing the content of the conversation—specifically, what the salesperson and client are saying. This enables the bot to provide highly accurate recommendations, with tests showing that 90–95% of suggestions are relevant and genuinely helpful. However, it currently doesn’t account for tone. From my analysis of available technologies, I found that no tools on the market are precise enough for this task yet. I plan to begin exploring tone-based solutions in 2025.
We are advancing step by step, testing hypotheses as we go. For now, the emphasis on text analysis is justified—it’s sufficient for generating effective recommendations, especially when the salesperson provides context and goals for the call in advance. In the future, we plan to develop our own system that incorporates voice characteristics, emotions, and potentially facial expressions to make recommendations even more precise.
Tell us more about how the service works from a technical perspective.
When developing the service, I chose to create a browser extension for Google Meet, as it’s widely used by small businesses. The extension also integrates with Google Calendar, allowing salespeople to input call context, including the meeting’s goal, client details (such as a LinkedIn profile), and a meeting structure. This information helps the bot better understand the conversation and provide accurate recommendations.
During a call, the salesperson sees a small window with tips and notes. The bot uses Google Captions to convert speech into text, which is then sent to the OpenAI API. Through our custom prompt-engineering approach, the bot generates relevant suggestions. If the conversation is on track, the bot stays silent. However, if deviations occur, it alerts the salesperson with helpful recommendations.
The system currently works in English and is set to expand to support over 20 languages. It generates tips approximately every 10 seconds, if necessary. In the future, we plan to adapt the solution for other platforms, such as Zoom and Microsoft Teams.
The project has already secured initial funding and completed several product iterations. How long did this process take, and how has the product evolved during this time?
The project began in July 2022. We launched a trial version (MVP) within two to three months and spent the next six months refining it by fixing bugs and adding features. In the summer of 2023, we rolled out a major update, introducing note-taking during meetings and one-click follow-ups. At the same time, we began developing an AI facilitator, which launched in December 2023.
However, by March 2024, we realized that the facilitator model was not scalable, prompting us to pivot. By summer, we shifted focus to leveraging AI for sales, which became the foundation for the current product version.
The development process faced challenges, such as gaining access to Google Captions and overcoming other technical barriers, which delayed progress. By fall 2024, the product was 95% complete, and the final version officially launched in December.
Initial funding came from personal investments by me and my co-founder, Ilya. We also received two grants from the Ukrainian startup program CIG R&D Lab in 2023 and 2024, totaling $12,000. Additionally, participating in accelerator programs helped us connect with over 20 potential investors from the U.S., Ukraine, and Europe.
Naratta has already entered the U.S. market and secured its first sales. How did you attract your initial clients, and what feedback have you received?
At the outset, we had six signed Letters of Intent from companies, including CaratX, Unicorn Universe, and Gradual. These companies were interested in using our product, though their commitment often depended on further development, such as Zoom integration or new features. Sales began with a previous version of the product, but it struggled to retain customers, which motivated us to create a new iteration.
To attract our first users, we invested heavily in marketing efforts, including outreach on LinkedIn and participation in major conferences like Web Summit 2023 in Portugal and Cannes Lions in France. Networking at industry-specific events also allowed us to build relationships with potential clients.
The new version of our product has received positive feedback from companies familiar with its earlier iteration. Some of these companies placed preorders, expressing interest in the updated concept. We now have organizations—primarily from the U.S., but also from Ukraine and Europe—ready to serve as initial testers or buyers after the release.
Our product stands out due to its focus on small and medium businesses (SMBs). This market has a limited number of sales optimization solutions tailored to smaller companies, unlike the corporate-level options available to large enterprises.
What major challenges did the Naratta team face during the product launch, and how did you overcome them?
The main challenges we faced fell into two categories: technological complexities and a lack of experience in launching a business.
As CTO, I encountered numerous technical issues, particularly with integrating our product into Google Meet and Google Calendar. These platforms frequently update their designs and features, requiring us to adapt quickly. For example, a change in the Google Calendar interface could render our system nonfunctional until we updated it. Additionally, publishing our browser extension in the Chrome Store was a lengthy process. We were rejected six times due to permission-related requirements, and only after multiple revisions did we secure approval.
On the business side, both my co-founder and I are software engineers with no prior experience launching a startup at this scale. We faced numerous unfamiliar tasks, including defining our business model, understanding client needs, conducting interviews, building marketing strategies, and managing a team. Even fundamental activities, such as registering the company or working with freelancers, were new to us.
Despite these hurdles, we struck a balance between developing an MVP and ensuring product quality. Through experimentation and refinement, we gradually built up our knowledge and processes, enabling us to tackle these challenges and bring the product to its current stage.
What goals have you set for Naratta in the coming year, and how do you see AI transforming the sales industry in the future?
Our primary goal for the end of 2025 is to implement our product in at least 50 small and medium-sized businesses, primarily in the U.S. and Europe. While we don’t have specific financial targets, our focus is on expanding our client base and enhancing functionality.
As for sales transformation, existing AI technologies have immense potential that remains underutilized. AI can significantly improve the sales process by making it more efficient. For example, our product provides real-time AI mentoring for salespeople, helping to boost company revenues and enhance customer experiences.
In the near term, we plan to integrate with platforms like Zoom and Microsoft Teams. Additionally, as mentioned earlier, we aim to develop more advanced models for analyzing text and tonal nuances in conversations. While more complex features like video analysis are part of our long-term vision, it’s already clear that AI in sales will streamline communication, making it faster and more effective for both sellers and buyers.
