Practical Know-How and Challenges from the Field of AI Sales Strategy

AI sales strategy is a key factor determining the growth of technology startups. Especially for early-stage AI startups, an effective sales strategy to convince the market is just as important as technical excellence.

In this article, we take an in-depth look at the challenges and solutions in actual AI sales fields, as well as differentiation strategies with ImpactiveAI's sales team leader, Director Min Byung-joon. Check out key insights into AI technology sales, from overcoming initial reference-building problems, securing customer trust, industry-specific customized approaches, to 'Deepflow Alpha,' an innovative internal AI modeling system.

‍

‍

The Beginning of AI Sales Strategy and the Role in AI Startups

What is your role as a sales team leader in an early AI startup?

I'm currently in charge of the sales team at ImpactiveAI. With extensive experience working at AI startups, I joined ImpactiveAI with the hope that I could contribute to building sales processes and teams for early-stage startups.

We still have a long way to go, but we're gradually achieving results. I've experienced various roles as a sales position in the AI business field. In terms of job functions, I've experienced not only sales but also pre-sales, project manager, and various other positions. I thought these experiences could help me serve as a business professional capable of multi-playing in an early startup.

Why did you decide to join an early-stage startup?

I enjoy challenging myself to improve things. Rather than working at a company with well-established structures, I joined with the expectation that I could create more efficient and active processes and results at an early-stage startup based on my experiences.

Of course, I had concerns when joining ImpactiveAI. Particularly, I wondered if domestic AI technology could be competitive in the global market. Since it's difficult to clearly understand the technical capabilities of a company from an outside perspective, I decided to join after resolving some of these questions during the interview process.

‍

‍

Practical Challenges and Approach Strategies in AI Sales

What is your strategy when explaining AI technology to customers?

In the last five years, customers' understanding of AI has significantly increased. Especially with the emergence of tools like ChatGPT, accessibility to AI has improved, so even decision-makers without technical backgrounds understand AI quite well.

Therefore, I've recently been striving for more technical logic and sincerity. When customers ask, "Is this also possible?" about something difficult to implement with current technology, I honestly explain the current technical limitations and discuss what vision we have for future improvements.

I think building trust through sincerity and logical communication is more important than unconditionally agreeing or accepting all customer requests.

‍

Are there differences in sales approaches by industry?

The basic approach is almost identical across all industries. The core problem our solution aims to solve is "resolving predictions based on data and AI that previously relied on experience or intuition." This broad framework is the same regardless of industry.

However, we approach the details according to industry-specific characteristics. Our prediction technology requires an understanding of the industry to which the prediction target belongs, and after understanding that industry, we go through the process of building an AI model. Accordingly, we adopt a strategy of expanding to companies that operate similar businesses within an industry where we've already built references. Since manufacturing OEM companies and B2C consumer goods companies have different data structures, patterns, and derived AI models for prediction targets, we secure as many similar references as possible and explain them in a way that each customer can easily understand based on these references.

‍

‍

What makes ImpactiveAI different?

ImpactiveAI's differentiation lies in our solid technological capability based on prediction accuracy and modeling competitiveness. Currently, we are focusing on prediction performance and algorithm precision, which are the essence of AI, rather than UI aspects.

While usability is important, what many customers are really curious about is how accurately AI can predict and have a substantial impact on their business.

From the initial implementation stage, we explain value centered on model reliability and predictive power, emphasizing that we've built an accurate and reliable AI solution, not just a visually appealing system.

Major concerns customers have when adopting new AI solutions are AI reliability and the possibility of job replacement. For predictive AI in particular, many customers tend to expect "100% accurate predictions." While accuracy is important, what's more important is "productivity effect improvement before and after AI adoption." If AI predicts with 80% accuracy what humans previously predicted with 50% accuracy, it can save 30% in costs even if it's not 100%. From this perspective, we emphasize that our solution is not about predicting the "correct answer" but a tool that enhances human "decision-making ability and responsiveness."

Deepflow reduces data analysis time and increases the amount of data that can be analyzed, helping to make better decisions. We approach it from the perspective that AI doesn't replace humans but makes them more capable.

‍

‍

Reference Building and Trust-Securing Strategies for AI Startups

What challenges did you face in securing your first customers?

The biggest challenge in securing initial customers was gaining trust in a situation where there weren't enough market-verified references yet.

Especially for AI solutions, from a customer's perspective, they need to comprehensively evaluate technical excellence, implementation effects, and company stability, making the verification barrier high for new startups. To address this, we focus on technical verification.

Whenever possible, we try to prove the effectiveness of our technology through POC (Proof of Concept). While providing free POCs can be burdensome for the company, we approach early startup POCs with the mindset that they are an important journey to find Product Market Fit, regardless of cost and resources.

‍

How do you solve the resource problems associated with POCs?

This is a truly important challenge. We're solving this problem with an internal system called 'Deepflow Alpha.' This system is optimized to automate the model experimentation process internally and build models with qualitatively and quantitatively excellent performance. According to our internal records, we've reduced the time for one AI modeling from three months to within 14 days using this system.

Deepflow Alpha works by running more than 200 model experiments in parallel, selecting the 10 best-performing ones, and then optimizing them for the client's data, as opposed to the previous method of experimenting with models one by one. The strength of this process is that we can provide technical verification materials to clients in a very short period, demonstrating 'feasibility.' Additionally, by minimizing human prejudice and optimizing based on data derived by AI, it's yielding excellent results in terms of performance. This has led to good results in comparative POCs.

‍

What is the ideal profile among potential customers?

Given the nature of time series prediction, companies with substantial time series data and high data frequency are ideal. For example, if there's only monthly data for a year, that's just 12 data points, but if it's daily, that becomes 365 points, which is much richer.

Ideally, B2C consumer goods companies with many SKUs (Stock Keeping Units) and little product variation are good candidates. However, since the actual business environment is much more complex and diverse, we continue to R&D to build customized models for clients in various industries. For example, we are currently preparing to commercialize advanced prediction models, such as models that better learn and predict irregular ordering patterns, or models that interpret sudden unexpected variables well.

The encouraging aspect is that regardless of industry, the need for tight inventory management is the same across clients, and the purpose of utilizing predictions is similar even if the prediction targets differ. Currently, we are focused on enhancing our solutions while concentrating on the common forms of service desired by clients from various industries.

‍

‍

Building a Specialized AI Sales Team and Future Strategy

What are your criteria for hiring talent for the sales team?

The type of talent we look for in the sales team is someone logical and flexible. I value flexibility in customer relationships as well.

I believe it's important not to unconditionally accept customer demands but to wisely resolve them with appropriate alternatives, and I'd like to work with someone who can lead this approach to provide optimal results for customers.

‍

What are your future challenges and goals?

At the company level, continuing to develop technological competitiveness as an AI company is important. We need to cultivate the capability to quickly solve customer problems based on our unique technological prowess.

Personally, my goal is to build sales capabilities and software competencies that allow us to be self-sustaining without additional investment after Series A. I think the harmony where sales effectively deliver software and developers create products with completeness is important.

In a situation where competitors continue to emerge, speed is the most important factor. It's crucial to quickly enter the market and establish the perception that "this company really does well." Just as GPT did in the language model market, the top 1 or 2 companies that establish themselves early in the software market rarely change. Therefore, one of the key strategies is to rapidly expand market share in the early stages.

‍

The Core of AI Sales Strategy: Balancing Trust Building and Technology Verification

Through the interview with ImpactiveAI's Director Min Byung-joon, we confirmed that AI sales strategy is a complex process that goes beyond simple product sales to include ensuring technical credibility, solving customer problems, and building long-term partnerships.

In particular, the problem of lacking references that early AI startups face must be overcome through honest technical communication and quick technology verification processes.

The innovative internal system 'Deepflow Alpha,' emphasized by Director Min, shows a new paradigm in AI sales. By reducing the modeling process from 3 months to 14 days, it has become possible to quickly demonstrate value to customers and build trust. This is a case showing how closely technology development and sales strategy should be connected.

Additionally, the approach of positioning AI solutions as a 'tool that enhances human decision-making ability and responsiveness' rather than a 'perfect prediction tool' is notable as a strategy that is realistic while appropriately managing customer expectations. This honest approach can help build long-term trust relationships.

In a situation where competition in the AI market is intensifying, the importance of the "speed race" emphasized by Director Min provides important implications for AI startups. Since companies that secure technical capability and credibility early are likely to dominate the market, the balance between effective sales strategy and technology development is more important than ever.

Ultimately, the success of AI sales depends on how well the triangular balance of technical capability, customer trust, and teamwork is achieved. By not just focusing on sales, but understanding customers' real problems and providing technological solutions to solve them, a foundation for long-term growth can be established.