Introduction

When easybits first approached me, they had something rare: a powerful, modular AI infrastructure that could create incredibly smart virtual assistants. But there was a catch - only developers could realistically use it.

They wanted to change that. The goal was ambitious: create a web app where anyone - whether a founder, marketer, HR lead, or AI enthusiast - could build, train, and deploy their own branded chatbot across platforms like Telegram, WhatsApp, Facebook, or their website.

The catch? It had to feel simple and intuitive - without dumbing down the power under the hood.

Context: a powerful platform hidden behind technical barriers

The easybits team had developed an advanced backend platform capable of orchestrating AI assistants using their own infrastructure and LLMs. But using it involved multiple steps, external tools, and technical knowledge most users didn’t have.

The idea was to make this power available to more people in a business sustainable way. They wanted users to:

• Create their own AI-powered assistants;

• Upload and train them with custom data;

• Define brand tone and voice;

• Deploy bots on any channel, all from a simple web interface.

This wasn’t just about a better UX; it was about unlocking the business model. If users couldn’t use it, they couldn’t pay for it.

The Challenge: making powerful AI feel effortless

Creating AI chatbots is inherently complex: prompt chaining, embeddings, vector databases, multi-model orchestration. Most people don’t need to know any of that.

Our job was to strip the process down to its core and build it back up as something approachable.

Core questions that guided the design:

• How might we help users build a chatbot in as few steps as possible?

• How might we reduce the need to leave the app to use third-party tools?

• How might we give users confidence their chatbot is working and improving?

• How might we showcase Easybits’ backend power without overwhelming users?

Goals & Strategy: balance simplicity with power

Help anyone build a chatbot, regardless of technical skill

Create an experience where the user doesn’t need to know what an LLM is to make something useful.

Give users control over training, voice, and performance

Let users personalize bots with their tone, upload their own datasets, and track how bots are used.

Help easybits convert users and grow their credibility

Build a product that shows off easybits’ infrastructure and gives the company a strong case for both customers and investors.

Process: fast feedback loops in a technical space

We kicked things off by gathering user needs, business goals, and technical constraints. I worked as the solo designer, collaborating closely with the PM, engineering, and marketing.

Step 1: research & mental model mapping

We analyzed how similar tools approached chatbot building, onboarding, and training. Then we mapped out our user types:

• Founders and solopreneurs;

• Marketers and support leads;

• Developers and AI enthusiasts;

Each had slightly different needs, but all shared one thing: limited time and tolerance for friction.

Step 2: wireframes, flows, and 5+ iterations

The chatbot creation flow took the most iterations. We explored models based on linear wizards, customizable dashboards, and modular builders. After user feedback and internal testing, we settled on a progressive flow that asked only what was needed, when it was needed.

Each step - naming, training, voice & tone, channels - was its own focus. We prototyped, tested, and adjusted at least five times before landing on something that felt smooth but still flexible.

Step 3: aligning with the brand system

I built the product UI using the same foundation as the Bits Design System from the main site. That kept the experience consistent and also helped devs work faster using shared tokens and components.

Key features and design decisions

Chatbot creation flow

We broke down the flow into clear, digestible steps:

1. Select the preferred module;

2. Give your bot a name, description and default messages;

3. Upload data (manual entry for the MVP). Choose a tone of voice;

4. Choose integrations (Telegram, WhatsApp, etc.)

5. Launch & share.

Each screen emphasized clarity and gave lightweight context. No jargon. Just guidance.

Personalization & training

Users could upload their own data and teach the bot how to respond in context. We gave them:

• A live training preview

• A tone/tuning section with suggestions

• Support for question variants and fallback logic

This gave users the power of fine-tuned AI without having to know anything about tokens or vector databases.

Analytics Dashboard

Once launched, users could track how the chatbot was performing with:

• Message volume over time

• Channel-specific usage

• User feedback loops (coming in future iterations)

The dashboard wasn’t overwhelming. It was just enough to give people confidence and actionable insight.

Results & impact

The app launched on app.easybits.tech and became a powerful layer in easybits’ overall offering.

• Helped generate over 200 user signups in the first few months

• Played a key role in lead generation and product validation

• Helped the team demonstrate the value of their infrastructure in real-world use cases

Some data remains confidential under NDA, but internal feedback was very positive - especially from the sales and marketing teams who now had a working demo they could actually show off.

What I learned

This project was a masterclass in simplifying the complex. It also reminded me that:

• UX is often about subtracting, not adding;

• Internal testing and feedback loops speed up good decision-making;

• Developer collaboration early on saves weeks later;

Most importantly, I saw how good product design can unlock a business model - by making something previously too complex, finally usable.

Conclusion

The Easybits web app takes something inherently technical and makes it feel natural and even empowering. It’s not just a tool. It’s a platform that lets anyone tap into the power of AI with confidence.

Curious how this applies to your project?

Let’s chat.