How We Built an AI & Data Platform: A Step-by-Step Journey

In this case study, we’ll dive into:

• The problem

• Product Discovery process

• Shaping the value proposition

• MVP Development process

• Strategic growth framework

• Release & Go to market

The problem - How do we simplify the complex and lengthy process of building data & AI transformations?

Our clients, a data engineer and data scientist, had a deep knowledge of the complex processes involved in building data pipelines for large companies. They envisioned simplifying this complexity through a low-code approach to data transformations, integrating various AI models to optimize for simplicity, speed, and cost-efficiency. Our challenge was twofold: to assess the core areas the client was addressing in the ML, AI, and Data landscape, and to synthesize these insights into a cohesive value proposition and product scope. This approach aimed to lead to a successful launch and an ongoing Product-Market fit validation.

The Product Discovery process

At Sudolabs, we believe in conducting multiple rounds of product discovery workshops complemented by our own market and user research. While the client is always the primary domain expert, our discovery process gathers comprehensive insights from various sources. This approach helps us fill knowledge gaps and enrich the client's perspective, ultimately leading to a more refined value proposition and feature scope.

Target audience

First, we had to understand who we're solving for. For early-stage products, it's best to start with a narrowly defined audience. In Fleak's case, it was data engineers. Especially in the B2B segment, it's crucial to understand both the user and buyer perspectives and gather granular data on:

• Firmographics (geography, industry, company sizes)

• Buyer Personas (teams, roles—economic buyer, technical user)

• Technographics (tech stack, programming languages, etc.)

Together with the client, we conducted a series of user interviews to assess core challenges and opportunities. Reddit threads in specific niche communities proved to be great sources of insight, helping us understand what users value—and don't value. They're also an entertaining read at times.

Our ultimate goal is to find the Ideal Customer Profile (ICP) - the buyer and user that converts best for our product offering.

User needs & Use cases

When shaping a B2B product, it's crucial to understand it from two perspectives:

• Buyer's business use case - The specific problems and outcomes the potential buyer's company needs to address in their industry directly impact their bottom line. It's best to start with vertical use cases (e.g., focusing on specific industries, markets, or similar problems) rather than horizontal ones (broad and spread across different market segments). This approach allows you to build features that effectively solve the buyer's problem already in the MVP phase.

• Technical user's challenges - The day-to-day work-life problems faced by the potential users of your product.

In our case, the client's research and our own led us to understand the technical user's challenges first, while the buyer use cases emerged in later stages. Rapid iteration in the face of uncertainty, adapting to each new insight or user feedback, defined our approach.

---

Don’t just deliver features. Understand the buyer’s use case and deliver features that solve that use case well.

---

Market Analysis

We took a structured approach to understanding the market segments in which Fleak would compete. This is crucial as it directly informs the product's future positioning. Our analysis included:

• Industry trends: Regulatory changes and emerging technologies that could impact competitors and the broader market.

• Search trends & website analytics: Insights into trending and emerging keywords.

• External factors: Data influencing the market and our product's potential.

• Time to market: Assessing why now is the right moment to address this challenge.

• Dependencies & potential roadblocks: Identifying obstacles such as legal considerations that could arise.

Competition analysis

The final stage of our market analysis involved examining key players in the industry. Our competitive research revealed their core strengths, weaknesses, unique value propositions, unfair advantages, features, pricing strategies, and other critical details. These insights played a key role in shaping our product positioning strategy.

From reinvention to simplification - Shaping the value proposition

Our value proposition evolved over the early months of the project as we deepened our understanding of the problem and the market. Initially, we aimed to center the product around building data pipelines for real-time streaming, leveraging the growing Apache Flink technology and community. However, this concept proved too novel for the audience, as the immediate need was found elsewhere.

Bringing on an experienced CMO and his marketing team helped refine our discovery process and value proposition. This led us to tap into a familiar mental model—workflow-building tools—positioning Fleak as an AI and Data workflow orchestrator with a powerful serverless computation engine, able to process up to 10,000 incoming events per second in real-time. To seamlessly integrate into data engineers' workflows and remain flexible across different company data stacks, the client made a technical decision to make the transformations available via API, allowing anyone to easily build and integrate their custom APIs, ensuring adaptability and ease of use.

Early Concepts

The early product proposition iteration was aimed at data teams and their cooperation.

Agile development

As the value proposition continued to evolve on the go-to-market front, the team stayed busy with ongoing discovery and delivery phases—iterating, designing, and developing the product from the ground up. We adopted an agile development approach with time-based milestones, leading to a private alpha release followed by a public beta.

Tech stack and product responsibilities

Although Sudolabs typically handles end-to-end product development, in this case, the client was a highly skilled backend developer with their own team of engineers. This allowed our team to take over responsibility for designing and developing the front-end interactions, managing the overall UX, overseeing product management, and orchestrating analytics and reporting."

The front-end tech stack consisted mainly of React as a JavaScript frontend framework, along with the Shadcn UI library.

For analytics, we chose Amplitude, as its deliberate philosophy of understanding your core metrics within acquisition, retention, and monetization first makes it very aligned with how we think great products should be built.

Strategic growth framework for early stages

The three key drivers of growth in almost any digital business are acquisition, retention, and monetization. For our MVP release, it was critical to understand the scope of work and how it aligned with these core growth pillars.

Growing an engaged userbase is what matters

In the early stages of product development, we focused on understanding our core activation journey and metrics—tracking everything from the first sign-up to that key “AHA” moment users realize the product's value. Early product engagement is crucial; it’s the best indicator of Product-Market Fit, and with limited time and resources, it became our primary north star metric.

Monetization is also critical information for B2B buyers before they consider the product, so we clearly outlined our pricing and monetization strategy. However, we didn’t allocate much engineering effort for the initial release to develop this functionality fully. If our product isn't gaining traction (i.e., a growing engaged user base), there's little point in using our limited resources on advanced features—especially when manual workarounds can suffice for the time being.

---

The best indicator of Product-Market Fit for early-stage startups is your user engagement.

---

Acquiring and engaging potential buyers

Measuring your channels' acquisition and conversion rates is the third piece of the product puzzle. Here, the goal is to identify the channel that consistently converts your ideal customer. Depending on your business type, this is where you’ll spend your time on sales and marketing activities, experimenting with different channels, messaging, and target sub-segments to pinpoint the ideal customer profile (ICP) that reliably converts through a specific channel. At this early stage, before achieving product-market fit, this was the primary focus for our client.

Go to market

Private Alpha Release

The purpose of the Alpha release was to develop an MVP with sufficient testable functionality for both moderated and unmoderated testing, which would guide our roadmap for the public launch. It’s best practice to avoid a public launch initially and release in closed stages to prioritize user feedback on value assessment over bug fixes and technical improvements.

Public Beta Release & Product Hunt Launch

Within just a few days of launching Fleak, we gained traction from the Product Hunt community, ranking as the #3 Product of the Day and ultimately becoming the #1 SaaS Product of the Week. While the Product Hunt community has unique characteristics, it can provide valuable early exposure that may lead to initial leads. However, the journey to validate product-market fit has only just begun.

On the journey to validating product market fit

The product release is a significant milestone, marking the point where the client has a working solution to share with the world. At this stage, it’s crucial to strike a balance between reaching out to potential customers, gathering their feedback alongside usage data, and conducting ongoing usability testing to optimize for adoption. This will help us transition from assumption-based development to a data-driven and user-informed approach.

Check out Fleak.ai to build your own AI workflow, and let us know if we’ve hit the mark in solving the efficiency problem for data engineers!