AI-based Interactive Learning Platform for Sales and Marketing

How Brocoders helped to build an interactive digital learning platform and integrate AI-based units to create an exclusive training tool for one of the fast-growing EdTechs in Europe.

Clients since

Our client

In October 2021, David Neuendorf, CEO the European EdTech company - HeyPractice (ex PeterPitch), contacted us with a request to re-build MVP of their interactive learning platform. HeyPractice is a VC-backed SaaS product with headquarters in Germany. The story behind its idea is that David used to work in a marketing agency and therefore knew that sales skills must be constantly practiced and honed. Combining the innovative approach of artificial intelligence with best practices and recommendations for sales pitches, he aimed to create a platform to help you learn this online. Now he would like to re-build MVP using modern technical stack.


“With Brocoders, we were able to set up a highly skilled IT team in the shortest amount of time. For finding a team with similar skills we would have needed 4-6 months instead of the instant start.”

Read full review on Clutch
David Neuendorf
David Neuendorf
CEO & Founder at
In July 2020, David started as Founder & CEO at Peter Pitch. From the April of 2022 - CEO & Founder at
Syam Sadasivan Pilla
CTO at Peter Pitch /
Started in January 2022

Business idea

The project is a training course for sales represantatives, managers and other employees. It is an excellent tool that allows staff to practice pitching skills for efficiency. Furthermore, it is a chance to transform companies’ learning culture itself.

Thus, their target audience is represented by companies that have departments for staff training and development. What makes it unique is the practice-oriented approach, where you learn theory and immediately put it into practice. The training takes place in an interactive form. As you go through the course, you record yourself and your pitches while the AI analyzes your responses. AI visually and aurally identifies your strengths and weaknesses and gives automatic feedback.

It offers the product as a stand-alone solution or as a solution that provides compatibility with an existing LMS. With HeyPractice you can train digitally or in person. And also, both group training and interactive training at your workplace can be applied.

The German approach to security guarantees

The level of data protection in Heypractice meets the highest industry standards:

  • data is stored in data centers certified in accordance with PCI DSS Level 1 and ISO 9001/ISO 27001;
  • has an SSL certificate and works exclusively through secure connections;
  • complies with the GDPR.

The situation

David Neuendorf, CEO of PeterPitch, wanted to design an interactive digital learning tool with empowerment options.

He already had a MVP of a product and was looking for an IT company to relaunch and improve their software. As a VC-backed company, PeterPitch needed a skilled team to build a new version of product in a short terms, within 4-5 months.

In this regard, they were interested in qualified specialists with experience in tech-stack, such as ReactJS, NextJS, and RoR. David looked at 4-5 companies and ended up partnering with Brocoders.

Working process


Initially, the customer has already performed the development of MVP backend, and they came to us to either redo it or make it from scratch. The client also had a ready-made design for new advanced functionality for their software solution. Thus, at the time of contacting us, the main request was that to rewrite the frontend interface and add new features.

Product audit

Following our usual approach to effective development, we had to understand if the backend was built properly so we can use it for the new frontend. That's why we conducted a technical audit and gave recommendations on updating the MVP and where to start the process. We provided the Heypractice team with a technical conclusion and a justification for why it would be better to rewrite the backend of a product:

  • there was no test coverage for the backend part;
  • there was no documentation or readme file for the project running;
  • code quality was average;
  • there was no dividing into front and back sides;
  • all application logic implemented via controllers;
  • API was not built on REST principles.

That being said, we also summed up that during future development of the new product version, we could reuse the old database structure (taking models and migrations from the current application). It was a very important conclusion because time restrictions weren't allowed to rebuild the backend from scratch.

Further, our team provided the client with the following solutions on how to improve and increase the quality of the application:

  • adding test coverage, at least for business logic;
  • dividing API will allow several developers to develop in parallel and make it possible to use the API for mobile applications. It also makes the project scalable;
  • separate logic into different layers to improve maintainability and testability;
  • make stress testing to check the system operates under a lot of use and load;
  • add OpenApi to document the backend interface;
  • consider AWS for hosting to use CI/CD services.

We suggested looking at some AWS services, as all our projects are hosted on AWS. So we can provide complete expertise about it. Another good point to consider is that they offer some internal services for running and storing videos. But alternatively, we could store videos on an AWS s3 bucket, make a preview, and resize them with ruby.

David agreed that the project needed to be rewritten, and we started our cooperation.

“We saw a huge amount of time that we saved. For finding a team with similar skills we would have needed 4-6 month instead of the instant start.”

David Neuendorf

Working stages

Therefore, we put together a team with great experience and set up a development process using best agile practices.

In general, the workflow of the project took place in two phases:

  • MVP development: until March 2022, Brocoders developers released the solution into production.
  • Final development: until August 2022, our team finalized the MVP.

At the first stage of the workflow, our team worked in cooperation with the customer’s developer, who wrote the first version of the backend.

And in the second phase, we had the pleasure of collaborating with Syam, CTO at Peter Pitch. He joined the team in January 2022 and participated in frontend and backend development.

For communication and task tracking, we use the tools of Slack, Jira, and Email.

Our developers were really encouraged during the working process. Searching for suitable solutions, they not only carried out integration but also conducted thorough research on AI products and managed communications with their representatives.

This was an essential part of the job, as the training courses require users to record their pitches on audio or video. AI-powered speech recognition makes each pitch automatically transcribed and scored. After loading the pitch, the employee receives personal feedback on the assignment without waiting for manual input from the coach, which makes the learning process more efficient.

Basically, we have carried out three AI integrations to provide comprehensive feedback for user input:

  • voice recognition and transcript of audio and video records into text for further analysis;
  • text analysis for allowed or forbidden words;
  • recognition of emotional color by face image.

Thanks to AI units, we make soft skills measurable.

What did we do? Our technologies

Our team developed the main functionality of the software product with additional AI-based features. The process consisted of the following:

  • Сreating and managing infrastructure based on AWS services;
  • Setting up CI/CD pipeline;
  • Building frontend and backend applications.

We developed the product using tech-stack:

Our team

For this project, Brocoders has assembled a team of professionals with extensive backgrounds. In this way, we save clients time in finding highly qualified professionals suitable for the project.

The customer characterizes Brocoders team as "fast set-up, very communicative, highly skilled, most flexible when talking about finding a good solution for all parties." David Neuendorf, CEO & Founder at Heypractice

So the Brocoders team included:

  • Project manager

  • Frontend developers

  • Backend developers

  • Fullstack developer

  • Quality Assurance specialist

At the start one frontend developer and one backend developer started working in October 2021. They started the initialization of the project, set up the development environment, and the initial project configuration (linter, test runner, etc). In a week, two more frontend developers joined the team.
Our team has a good level of communication inside, which helps them quickly come to the optimal solution. Frontend and backend constantly communicated directly with each other and could pick up each other's tasks if necessary.

New people joined the project as needed and when new tasks appeared in the project. Thus, in December, a QA specialist came to the team, and a month later, another frontend developer joined.

Key challenges

Searching for AI solutions

One of the key tasks for Brocoders team was to integrate the AI solution. The bottom line is that you can train in the area you want and get immediate feedback from the AI. You don't have to wait for the coach to process your results. You simply record yourself on video or audio during the training process and get a quick assessment from the AI.

Therefore, our developers were looking for a solution that could recognize voices and convert audio recordings to text so that AI could analyze them.

The AI gave feedback to the user based on given parameters, such as:

  • permitted and prohibited words;
  • length of pauses;
  • volume of sound;
  • speed of speech (spoken words per minute).

Voice recognition parameters such as volume criteria, length of pauses, and speech of speech have been set on the backend.
To analyze the text for allowed and forbidden words, the client can fill in a list of do and don't words when creating a task for the user.

So, our developers were tasked with integrating an AI service that can be used for voice recognition and converting audio files into text transcription. To accomplish this, they had several options.

First, we turned to the Assembly AI provider, which had a very competitive pricing model. But it turned out that their service does not support the German language. However, during communication with our team, the Assembly AI support promised to add the German language. We did the integration, and the release date came, but still, there was no ready-made Assembly AI solution for the German language. As a result, we had to abandon this variant.

Therefore, we switched to a solution based on the IBM Watson platform. This option was chosen in cooperation with CTO and was good value for money. The main challenge in working with this provider was that they did not have JavaScript documentation. Therefore, the developers spent a lot of time figuring out how to integrate this system. However, they could cope with the task thanks to their persistence, experience and professionalism.

We integrate PRECIRE AI solutions to generate psychological results of text analysis.

SCORM standards implementation

Our team did a bunch of research and found a solution that makes course content compatible with SCORM standards so that enterprise clients can play our content in their LMS. Thus, we found and integrated Rustici Software. Why was it so important?

The fact is that often companies interested in e-learning services for their employees have their own LMS (e.g., Moodle or eLearning). For such corporate clients, it is crucial that their employees do not switch to another portal during training but remain in their LMS. And if we want them to be able to upload our training content into their LMS, the courses must be compiled with the unified SCORM.

Sharable Content Object Reference Model (SCORM) is a set of technical standards for e-learning products that helps them play together. Advanced Distributed Learning Initiative (ADL) developed it to support the Department of Defense in standardizing learning objects and making learning portable across systems. SCORM consists of sub-specifications. Content packaging defines how the content needs to be packaged and described. Run-Time specifies how the content must be launched, how the data interacts with the LMS and the data model requirements for this interaction. Finally, sequencing determines how the learner can move between parts of the course (SCO). Compliance with this standard ensures that the content is compatible with the LMS.

Thus, we needed to find a solution that would provide an additional opportunity to export courses to other LMS. Rustici Software converts content according to SCORM standards.

It is also essential that thanks to the Rustici Software solution, users do not feel they have switched to another platform when entering their credentials. We, for our part, do not receive user data.

Admin panel implementation

To manage, control and monitoring all the product processes, the team needed to find the best solution for implementing the admin panel.

There were several options for discussion:

  • move the functionality of the admin panel into a separate application;
  • built it into our solution, but to differentiate access by roles.

We still settled on an intermediate option, when we collect two separate folders for client pages and admin panel from one shared Front database.

Delivered solution

The solution we performed is an EdTech SaaS product based on AI. It is available in two languages: German and English.

As HeyPractice continues to grow, they have plans to develop the platform further:

  • backoffice for coaches;
  • community for company portal;
  • integration with Slack, Microsoft Teams.

HeyPractice customers who already used digital training on the platforms:

A mobile version of the product is also developed.

Application structure

Company portal

  • Dashboard
    The admin can track performance data on the company portal through an intuitive dashboard. Pages of Overview and Environment updates are available here. It is easy to check the overview of both teams' and members' activities. And here you can freely look through your materials, feedback, polls, and new records.

  • Courses
    This section displays a list of teams and individual participants, as well as the pitches of coaches. Here you can also set up the schedule for the courses.

  • Teams
    In the company portal, it is also possible to create and edit Teams. There you may add members, monitor the team progress and time sessions, see the team leadership board and criteria overview. You may look through the courses and records of the group's members and easily find information about the results of each employee.

  • Content
    Here you may add and organize content materials. Folders, files and lists with different materials for courses or with videos or presentations are placed here.

  • Workspace
    This is the workspace where you manage the members. Here you can find user data, their emails, and their roles. Also, there are invoices for payment for purchased products. Mind that payments are made offline.

Employee portal

  • Profile
    This is the space where users can customize their profile: change photos, set up notifications, and so on.

  • Dashboard
    Thanks to the dashboard, an employee can monitor his progress and environment updates. You can keep track of the courses you subscribe to.

  • Courses
    Here is a list of the courses you are taking. You can track progress, view available pitches, check out the pitch leaderboard, and track task scores. You see task info and reviews: feedback, comments and AI-Feedback.

  • Community
    This is a communication space where you can not only follow the updates and news of other team members but also receive valuable materials from the coach. It was not included in the MVP.

Superadmin - backoffice

  • Companies
    Here we add and edit information about:
    • companies, their names, number of employees, number of courses, contact person (who may have the super admin role);
    • all users, enter their personal data and specify the roles: super admin, admin and member;
    • teams;
    • courses;
    • content (folders, files, lists);
    • workspace; here, we indicate to whom the company gives permission to manage team and user data, view team statistics, create, edit and delete courses and pitches etc.

  • Content
    What you can do here:
    • upload files;
    • create lists of reactions, do-words, dont-words and attach them to companies;
    • add content, form folders, upload files and lists.

  • Courses
    This is where you add a new course. Teams and participants are determined for each course, and there is a space for viewing pitches and making new ones.

Courses. How it works

The application has a really user-friendly interface. Corporate clients can either buy ready-made HeyPractice courses or develop them on their own based on our platform with the help of their coaches (for example, add their tasks, etc.).

Opportunities for trainees

For salespeople and employees, HeyPractice offers digital products to train sales pitch. For example, it can be a strategic selling technique developed by coaches. The aim is to train how to create a successful Sales Strategy.

How you can improve your skills:

  • Pass the theoretical part at the beginning of the course.
  • Take interactive multiple-choice tests to test gained knowledge.
  • Take Quizzes, write short texts, and record audio tracks.
  • Start recording videos of your own pitches. For example, you may practice how to offer a new investment digital product. Your pitch is evaluated with some criteria:
    • persuasiveness (confident body language, speaking/writing clearly);
    • charisma (communicating dynamically);
    • quickness (without pause of more than 5 seconds);
    • uniqueness (unique techniques and solving problems.)
  • Get immediate AI-Feedback. When you complete the task, results are processed by AI units, and you see AI-Analytics and examine what you need to improve.

How to create a course

When a corporate client buys a company product, a workspace is created for them. Then the admin invites people and assigns user roles to managers and employees.

In the company portal, he may use editor f