First metrics of the Virton AI beta version

It's been quite a tumultuous month for Virton AI. There were many successes, as well as difficulties in general. Let's take a look at what the project has achieved.

Below we've put together a brief summary of Virton AI metrics for you:

• 4,500+ DAU - A record for unique users in a day;

• 2,500+ DAU - Average DAU;’

• 30,000+ MAU - At the time of publishing the news;

• 15+ Pilots and 500+ Products;

• +65% - Up for generation quality;

Analytics for stores based on Virton AI - new functionality

Just today, a new functionality was launched for all Virton AI-powered stores - analytics.

This functionality allows store owners who have connected the fitting room to view real-time efficiency. Figures on views, unique users, number of fittings, conversions to purchases (both from the widget and from VK and Telegram applications), as well as the history of recent events.

In addition, UTM tags are now transmitted to the store when a widget or app transitions to a purchase to help evaluate traffic metrics.

New possibilities for customization

Virton AI updates widget customization capabilities every iteration, both for regular users and White Label solutions. You can now customize video blocks, covers, and upload your own sample selfies to the widget.

New Virton Shopping product launch

Neurosell today launched the beta test of its new Virton Shopping product. It is a cross-platform marketplace that will be available to all online clothing stores working with Virton AI.

This marketplace will include both online shopping and a fitting room in one person. In the future, full-fledged payment acceptance, delivery and synchronization with all popular e-commerce platforms are planned. In the meantime, every store can try it out for free.

You can try Virton Market in Telegram Mini Apps.

Other wins and results of the month

The first month of beta testing has been quite busy for both Neurosell and Virton AI. Below we have collected a list of other wins of the month:

• Virton AI was honored as the best startup of the week at Product Radar, taking the top spot;

• Neurosell has been named one of the top 500 companies in the world by Crunchbase;

This is far from the end, just the beginning of team and product development. We thank all partners and users, as well as our team for their boundless support of Neurosell projects.

Read More about Neurosell | Read More about Virton AI

Why did Neurosell enter the competition?

The flagship product of startup Neurosell is AI-powered virtual fitting rooms. The team has great expertise in development, including in AI organization. The Virton AI startup is already working with pilot projects and testing many hypotheses, weekly algorithms. So why being an already growing startup - Virton AI was sent to the competition?

First and foremost - the company enters contests to get expert feedback and community involvement. For Virton AI it is important to get not only customer experience, but also support and feedback from industry experts.

Alexander Khopyorsky - Founder at Brat AI, no code for AI Agent - became the project hunter. Including the exchange of experience with other industry representatives is an important point in the formation of the company's product strategy.

What did the Virton AI team get out of the competition?

First and foremost, our team gained experience with the community. It was important for us to get feedback both from other founders in related fields and to learn more about working with communities like Product Radar. Such experience helps us adjust our product strategy and internal processes. - Elijah Rastorugev, Neurosell & Virton AI CTO

Naturally, this is just the beginning of the journey for Virton AI, but already, after just one month of an active public beta, the team has secured many key partners, including pilot projects with various brands.

To learn more about Virton AI, visit the product's website:

https://virton.tech/en/

New options for customizing store widgets in Virton AI

Neurosell today launched new functionality for additional widget customization in Virton AI. Now stores can not only customize the content of the video block on the homepage, but also upload their own sample selfies for users.

In addition, now you can view possible and future integrations with various platforms directly from your personal cabinet, as well as leave a request to connect to VK Mini Apps and Telegram Mini Apps.

Also in the near future we will launch additional functionality for generating models and product cards directly from the personal cabinet of the store. In this way, brands will be able to save money on photo shoots for their new collections.

New Brand at Virton AI and partners section on the product website

Today Neurosell announced a partnership with ZANOVO brand, a fashion company that develops and sells clothes from Karelia. The companies will cooperate in e-commerce and virtual fitting rooms.

More information about the brand can be found on the official website:

https://za-novo.ru/

In addition, Neurosell has launched an updated section of the Virton AI website that will display all new key partners in the virtual fitting room space.

The list can be found on the official website:

https://virton.tech/en/brands/

Integrations with new brands and examples of Virton working as a cross-platform solution

One of the major Virton AI collaboration announcements among fashion companies was the launch of a pilot with premium clothing brand DeBacker's. The brand integrated with a fitting room based on Shopify, which is the first such integration.

Now the Neurosell team is actively working on other integration options, including Telegram and VK mini-apps. Examples of such integrations can be seen below:

• DSM Icons Brand at VK Mini Apps;

• NashLocь - Virtual fitting room in Telegram Mini Apps;

These integrations with new sales channels for an apparel brand can be up and running in a single day, which is one of the benefits of going full-fledged with Virton AI.

Participating in the battle of startups on Product Radar

Also this week, Neurosell began competing in a battle for the title of Startup of the Week and Month on the Product Radar platform with Virton AI. The project hunter was Alexander Khopyorsky, founder of Brat AI.

You can vote for Virton AI on the Product Radar page:

https://productradar.ru/product/virton/

New records and Virton AI users

Well, the latest news from Neurosell is that Virton AI's daily user traffic grew over 100% this week to over 3,000 daily users.

The product is currently in training and in collaboration with brands to develop its algorithms.

More details can be found on the Virton AI website:

https://virton.tech/en/

Virton AI-powered martkeplaces announced

Today Neurosell announced a new product - a marketplace based on Virton AI. This functionality will work in parallel with the usual fitting rooms, but in addition to the usual fitting rooms, a marketplace based on applications for Google Play, Android, as well as VK Mini Apps and Telegram Mini Apps will be launched.

Every store that connects a fitting room will immediately be placed in the marketplace, thus opening up a new audience for Virton. The marketplace is scheduled to launch in June 2025.

Virton AI analytical tools

Advanced analytics will be available to all Virton AI users in the coming weeks. This will include an overview of popular products, a look at the number of generations, generation of picks based on product popularity, and an overview of the audience cross-section, including a breakdown by day.

Already now, there is functionality to track UTM fitting room tags, which are generated automatically for each product.

New pilot projects and partners

Neurosell has also announced partnerships for Virton AI virtual fitting room pilots with brands such as Nishtyak Bratok, Zanovo, Murka, NashLosь, DSM Icons, Knives and other apparel brands. This helps in training the neural network even faster and better. Pilot projects are planned for six months, during which time the generation quality is expected to improve by more than 90%.

More detailed information is always available on the official Virton AI website:

https://virton.tech/

This is gonna be interesting, let's go!

Virton AI's massive new update

Today we launched a major new update to our artificial intelligence-based virtual fitting rooms - Virton AI. It includes new functionality for online store owners, updated algorithms for working with photos, widgets and internal tests of video fitting in motion. Below, we tell you about everything in order.

Online Store Dashboard

Started a personal store management cabinet to implement a fitting room widget without programming knowledge. To do this, you only need to customize the widget appearance, fill in the products and connect the widget to the site.

You don't need to fully customize products like in regular catalogs of online stores - you only need to customize the product photo, name and link for purchase, specifying categories.

Appearance can be adjusted to your site - upload logos, adjust the color palette. Changed your site design? No problem - the widget automatically updates all settings when you save them in your control panel.

Virton AI virtual fitting room algorithm update

We've also updated our algorithms, making them even more accurate for trying on products. And the fitting widget itself has become clearer, more user-friendly and simpler.

In addition, we have internally tested the online fitting room algorithm based on video in motion. The result of the video sample from the internal tests can be seen in this video:

The launch of new algorithms for fitting accessories, hats, shoes by photo is planned for June, but with video fitting - we plan to start in August. If you have your own online store, right now you can join us on preferential terms (the first months are free), and then be one of the first to receive priority updates.

Neurosell and Virton AI partnership areas

Starting Monday, Neurosell will begin working with partners like Shopify, InSales and other online store building platforms to make it quick and easy for anyone to build full product card integration into their website.

In addition, we can already now individually connect the fitting room to your VK community or Telegram bot.

Pilot programs and new integrations

Virton AI is scheduled to roll out next week as pilot projects with 8+ apparel brands who will help train the algorithm at a discounted rate. For now, anyone can test our algorithms unlimitedly by installing the widget for free:

As a final note, I'd like to extend a huge thank you for the support of all the apparel brands who are helping with the development of Virton AI, as well as the Neurosell team for their prompt work on the algorithm.

Try Virton AI for free

See you in a week!

Beautiful Generations

Let's start with good fitting room generations. Users self-assess - and we select for you.

Unusual and funny

And now, let's get to the most interesting - unusual and funny generationals that made us at least smile.

Today, the Neurosell team is announcing the launch of the next round of Seed investment of $800,000 to improve the product and scale it to market. To date, the company has already raised $60,000 Pre-Seed investment for initial hypothesis tests.

What's the plan to do?

In addition to expanding the development team and launching marketing campaigns, Neurosell plans to take the accuracy of the algorithms to the next level, providing a new quantum leap.

Key product readiness criteria based on Seed round highlighted for Virton AI:

• Increase in generation speed by more than 115% - to an acceptable 5-6 seconds;

• Improving generation quality and its average rating up to 90%;

• Provision of tariff plans for businesses;

• Expansion of generation options in new projections (back, side, etc.);

• Expansion of supported product categories (shoes, jewelry, hats, etc. will be added);

It is worth noting that Virton AI is part of Neurosell's ecosystem of products, which allows it to have big data to learn from e-commerce partners. Also, in the coming days, the company plans to launch a personal account for businesses, allowing them to create widgets and integrations for any online store in a few minutes.

You can follow Virton AI news in the official resources:

- Neurosell Blog;

- Crunchbase Page;

- Neurosell Telegram Channel;

- Bluesky Profile;

Hi everyone, we are here with the Neurosell team. In today's article we would like to share our experience in training neural networks on the example of our product - Virton virtual fitting room. We will tell you about the complexities of the algorithm, user experience and training opportunities.

First things first, let's review what our goals were:

• Maximum preservation of context to get an accurate result of fitting in the image (for example, the user wants to match a T-shirt to his shorts in style;

• Accurate product image transfer with preserving poses, adding details of fabric folds, etc;

• Relative economy of algorithms;

What do we use for training and work?

• UNet-based generative and reference models;

• PyTorch under the hood, transformers and diffusers;

• SCHP for pose detection and correction;

• DensePose for body pixelization;

• NVidia Cuda-based servers;

Now that we have a quick look at the goals and technologies, let's break down what we are doing to train the algorithms, looking at the main challenges and current results.

Training algorithms, metrics in current format

Initial training of the algorithms was performed on the large DeepFashion database, which offers many variations of different fashion items. Other sources for pose detection were also used. However, there is a disadvantage in using such databases - they do not provide real user experience and train models in ideal conditions.

What to do in such a case? Run beta versions of the algorithms!

As a result, real users start trying on clothes, creating new generations and pre-training the algorithm. And in the case of pilot products on online stores, we can make the algorithm even better, getting new styles, different users with their own understanding of how the services work.

So, what we have achieved in two weeks of training the algorithms:

• At launch and in the first week of operation - generation accuracy was in the neighborhood of 30%;

• On the second week of work, some correction of UX and internal algorithms - the accuracy increased to 60%;

Thus, by pre-training the algorithms on real users we bring the results to commercial values and this is only for 2 weeks of work.

What affects the accuracy of an algorithm?

Under ideal conditions, of course, we get great results, but in the real world it doesn't work that way. We get different users who often take photos that are not quite right for the neural network to work. And all of this has to be taken into account.

In our case, accuracy is affected by:

• The quality of the original photos;

• Compliance with color rules (if the person in the photo starts to merge with the background, or if he or she has monochromatic clothes that are difficult to segment);

• Presence of other people in the frame;

• Using photos that are not full-length;

• Posing (crossed legs, arms, etc.);

The more of these indicators are combined together, the lower the quality of generation becomes. What to do in this case:

• First work through the UX and explain to the user in clear language under what conditions they might have a bad result;

• Do-train models on segmentation in various non-ideal poses for the algorithm;

• Implement Multi-View Pose Transfer (for sideways photos and other body rotations;

• Implement body part intersection detection, e.g. using pose detection algorithms (there are good examples on Tensorflow) and alert the user to problems;

Generation examples - from successes to epic fails

And as the final block of this article - we decided to share with you the results of generation, among which there are excellent, and there are quite funny and even strange.

Conclusion

What I would like to summarize in the end: any neural networks trained on ideal data and various open databases, as a rule, do not take into account the user factor, which definitely affects the result. Therefore, it is worthwhile to densely engage in pre-training of models on your real users, work with UX and look for new options to improve algorithms (for example, by introducing new additional tools).

And as always, discussions and your questions are welcome!

So, let's get to the hottest updates!

Virton AI virtual fitting rooms updates

Just last week we launched the beta version of Virton - a virtual AI fitting room for online stores, and already this week we are rushing to tell you about the new update and development roadmap for the next month.

So what's new in Virton?

• Redesigned the user interface of the demo app and widgets;

• Updated the algorithm and pre-trained it for more accurate generation of underwear;

• Accelerated generation at the same capacity by 15% - now the average generation speed on the demo bench is 60 seconds;

• Started working on pilot projects, launched a demo version of the app in VKontakte mini-apps;

• Worked out a development roadmap for May;

What to expect in May:

• Reduced generation time by increasing project capacity by 50-70%;

• Algorithm updates and support for different poses (back view, side view, support for redrawing complex poses);

• Widget builder for business (to be presented at the Ecom Expo 2025 conference in Moscow);

• Full-fledged marketplace applications for end users in beta version;

• Pilot project launches with clothing brands;

Thus, we are trying to develop the product from new directions, setting ourselves the ambitious goal of the emergence of AI fitting rooms as industry standards.

Read more About Virton AI Fitting Rooms

A major release of NeuroCart, a cloud-based e-commerce ecosystem, is in the pipeline

What New await in May?

• We are preparing dozens of new integrations with popular payment acceptance services, CRM systems, ERP and delivery;

• Public release of content AI modules, smart product feeds, and integration with Virton's virtual AI fitting rooms;

• Updated customizations and basic store templates;

• Support for VK Max messenger and Parallelix library update;

• Going Open-Source and public documentation for developers;

• Start preparing a cloud-based control panel for extensions and store instances;

• New use cases with large ecommerce clients;

Read more about NeuroCart Ecosystem

An Open-Source solution for Peer-to-Peer neural network training will be released in May

We work a lot with artificial intelligence, and, in fact, we are already more of an AI company than a traditional IT business. It is very important for us to train algorithms not only on open databases, but also on real cases from users, while optimizing the training processes themselves.

In May, we will provide a new open source and completely free tool for training neural network models on a Peer-to-Peer basis. What it means. Every user of your website/service/store will be able to contribute to the development of your AI services by sharing part of their power for model training.

How it works?

• The cloud server assigns the tasks of computing certain models to training;

• Computations are distributed among connected clients on the web (background work via services and webworkers is possible);

• After the computations, the data is sent back to your cloud server and augment the trained model with new data;

At our core, we are launching decentralized free libraries for developers to take some of the computation off the servers and offer their training to clients.

Follow our projects and libraries on the official blog

Other Neurosell news

Finally, let's run briefly through the news that didn't make it into the full digest, but may also be of interest:

• Virton is officially verified in SaasHUB;

• We are looking for pilot projects for AI fitting rooms;

And on that note, we will conclude our digest of the top news in the Neurosell ecosystem. Follow us on social media and share your experiences!

We are always open to new ideas and feedback.

For more than half a year now, we have been developing various e-commerce solutions. We want to create a full-fledged ecosystem using modern AI technologies, and the idea of introducing fitting rooms that can be available for any online store has been plaguing us for quite some time.

We did quite extensive market research in this area to understand which technologies might be suitable for us. We considered VR and 3D, augmented reality and photo processing.

So why did we settle on the latter option, processing photos with AI for fitting?

• A simpler and more accessible option for users - just take a suitable photo once, then try on different things and select images.

• Compared to augmented reality - the result is more interesting and accurate.

• We almost immediately stopped considering VR and 3D, as this method is quite expensive to implement in online stores, and not so accurate for users, as the 3D model does not reflect how the buyer will look in clothes in reality.

And so, a few months ago we started active development, and today we are launching the first public beta test. Let's talk about how we created the product and how it can be useful in the future.

General concept

We all remember that nowadays there is a standard in online stores - you can add items to Wish-list or to your shopping cart. But to try things on, you usually have to either go to the showroom or wait for delivery in order to at least understand whether the item fits your style.

What do we want to decide?

We want to make it easy to try on an item before you buy it. It's cool enough to see a real-life example of how a sweater will fit your look by uploading a photo and getting the result on the real thing. That's what we're doing with AI algorithms that accurately transfer the item to your photo.

The interface should become as intuitive as simply adding an item to your cart - you select the product, attach your photo and get the result.

How does it work?

We won't go into the technical features of the product - we'll write a separate article for that in the future. Here we will simply describe the main features of the algorithm step by step.

Let's take a look at Virton's step-by-step algorithm:

1. The user chooses a photo of the product (or uploads his own);

2. Uploads his own photo for fitting;

3. Algorithm analyzes and segments the body parts in the photo;

4. Based on the segmentation, a mask is created for further work of the generative AI;

5. Using algorithms for working with images, the photo of the product is transferred to the mask;

6. Further post-processing, smoothing and minimizing artifacts are performed;

The user then receives the final result of the algorithm and can download it to his/her device / share it with friends, etc.

How can Virton be used?

At the moment, our product is at the stage of open testing. We are launching pilot projects with online stores and further training the algorithms to achieve the most accurate results.

We plan for our product to accomplish the following tasks:

• Increase the performance of online stores by providing users with a completely new AI algorithm-based online fitting functionality. No waiting for delivery and no showroom visits - everything is done one photo at a time.

• Reduced costs for photo studios and models to create product cards, whether for their own store or for posting on marketplaces.

• Improved user experience, combined with other AI tools in online shopping, creating a next-generation ecosystem (smart product feeds, image matching assistance, and more).

We hope that we will be able to achieve this through our experience with AI, as well as pilot projects with which we will jointly try to improve the quality of the algorithm.

Virton Roadmap

As of today, we have a clear roadmap for product development for the coming year. We are confident that we can make fitting rooms an even better and more useful product that can be integrated into any store.

What we plan to do:

• Improve mask generation algorithm and accuracy of clothing transfer to the photo;

• Add support for other projections in the algorithm, not just frontal view;

• Run several pilot projects on large online stores;

• Optimize generation speed to 10-20 seconds, increase the number of servers for generation;

• Provide widgets for easy integration into any online store;

• Launch Virton client application with catalogs of pilot stores;

• Add search and image generation based on user preference, photo and online store catalog.

The plans are really big and ambitious, but we have come a long way in building the ecosystem and are confident that everything we do is doable, it's just a matter of time.

Where can I try the product?

At the moment - Virton is in open beta testing. You can try it absolutely free on our website:

https://virton.tech/

We have also attached a few variants of the fitting done on the current version of the algorithm:

At the time of beta testing, we are also happy to invite online store owners to run pilot projects and integrate Virton fitting room on your website.

Read more about Virton for online stores:

https://virton.tech/ru/ecommerce/

Hi everyone. Today we have launched in beta a new toolkit for cross-platform web application development called Parallelix. This toolkit was born as a separate library, important in modernization of our products, as we daily face the problem of controlling dozens of SDKs for different platforms of our web applications.

So - meet a unified library for working with your projects.

GitHub Repo | Documentation | Demo | Roadmap

About Parallelix

Parallelix Toolkit - Written in Vanilla JS, a modular library that allows you to forget about the existence of different client SDKs when developing your cross-platform web applications, be it VK Mini Apps, WeChat Mini Apps, Progressive Web App or Apache Cordova based applications.

Write the same code for all platforms, changing only the parameters to execute methods according to the platforms' documentation.

Why Parallelix?

• Cross-platform operation of your web applications on different platforms (VK Mini Apss, Telegram Mini Apps, Cordova etc.);

• Unified API for working with authorization and other functionality of platforms;

• Modular connection of various libraries required for platforms' operation (for example: VK Bridge);

• Simple to use and easy to learn;

Get Started with Parallelix

Before you start, download Parallelix libraries from the release page from our official GitHub.

Or clone repository into your project:

git clone https://github.com/Neurosell/parallelix.git

Now connect the main script to work with Parallelix in body tag:

<script type="text/javascript" src="/libraries/parallelix.js"></script>

And start connecting the client in your JS application:

// Configure Parallelix
const client = new Parallelix({
    baseModulesPath: "/libraries/",            // Where your modules located
    supportedPlatforms: ["vk", "telegram"],    // Modules to initialize
    vk: {},                                    // VK Module Config Example
    telegram: {}                               // Telegram Module Config Example
});

// Add Error and Initialization Handler
client.OnError = (error) => {};
client.OnInitialized = (data) => {
    // Work with Parallelix Here
};

// Initialize Client
client.Initialize();

Done. We've got the basic connection sorted out. If you want to learn more about configuration options - visit our section in API Reference.

Connect your Modules and Usage

Modules are connected automatically when you specify their name in the Parallelix object configuration. The module name is also contained in the script (parallelix.MODULE_NAME.js).

If you don't need a particular module, you can simply remove it from the configuration.

Once connected and initialized, you are notified that the client is ready for use. This is how you can start using it:

// Your Client Initialization Handler
client.OnInitialized = (data) => {
    // Publish story to social Network
    client.Platform.PublishStory("mediaURL", {
        vk: {
            // Platform Specific Request Parameters from VK Bridge Docs
        },
        telegram: {
            // Platform Specific Request Parameters from Telegram SDK Docs
        }
    }, (data)=> {
        console.log(data);
    }, (error)=> {
        console.error(error.message);
    });
};

What’s Next?

We look forward to your comments, feedback and of course - new ideas. All detailed information can be found in our GitHub.

About Help Center

This help center has been created to organize all documentation and articles that may be useful for your study. Here you will also find links to all the most relevant open source products and solutions from the Neurosell team.

For Contributors

Learn how to work with repositories and projects with developers from the Neurosell team and community.

Visit documentation center

It's no secret that these days, AI (or to be more precise, big language models), have become ubiquitous. They already know how to perfectly answer user queries (especially when using manifestos to achieve goals), generate good images, search for information on the Internet and do many other things.

In today's article, we would like to talk about how things are going in terms of programming. How well progress is moving there, and whether it is realistic to apply AI today in actual product development.

A short about us - We are in the e-commerce product development business and have already been implementing AI tools everywhere based on our experience - smart product feeds, product card generation and more. But even though we've been using AI tools for enterprise tasks for quite a long time and tightly, we've only recently introduced them into development. And let's see how things are going with the application of AI in the development process?

How can AI help in development?

When can AI be useful?

Artificial intelligence has significantly changed the approach to software development in recent years, providing developers with new tools and capabilities that can significantly improve the efficiency and quality of the products they create.

One of the key capabilities of AI is the automation of routine tasks. Modern AI systems are capable of performing many tasks, such as writing basic modules or templates, testing, writing documentation and debugging.

We've highlighted a few things that help us specifically:

• Replacing Stack Overflow and Google search (noise reduction) in a number of basic tasks;

• Help in comparing algorithms and patterns for subsystem design;

• Help with basic creation of user interfaces, text and translation work;

• Technical support;

• User context and recommendations;

• Auto-complete and code writing;

• Generating templates for development;

• Writing documentation;

• Basic algorithm development;

• Refactoring;

• Testing;

When can AI be harmful?

Above we have highlighted the points where AI can be useful. However, it is unwise to leave all the work to AI. Language model technologies, whatever their context, are far from perfect. Therefore, you need to recognize when AI can harm your development:

• When the developer is unable to clearly identify and set his task;

• When the developer has bad competence in the issue he/she is solving with AI;

• When it would take less time for the developer to write the code than to further debug and verify the code written by the AI;

All of this needs to be taken into account when applying AI implemented into your processes. You should always remember that AI is not a replacement for a programmer, but a great tool for simplifying development processes. It is foolish not to use these advantages, but it is also unwise to rely on them completely.

Applied and indirect tasks

First, let's understand the tasks involved in research and user data, rather than writing the code itself.

Stack Overflow replacement

Search engines are imperfect, and there are times when AI can understand the context of your query more easily and better than you can Google it and start scouring the pages of Stack Overflow and other forums. In most cases at this point, AI is already getting better at finding answers to a lot of common questions, but complex systems are still difficult for it.

It's important to note here that AI is simply speeding up the search for information. In addition, today it is possible to make a squeeze of current queries and shorten the answer.

Architectural planning assistance

Compare several approaches in architecture without flipping through 200 pages describing key features of these or those approaches? Of course, it is easier to do it on the basis of AI, as it copes quite well with outlines of already existing information.

You feed it your ideas + source of information about these or those solutions and get a brief, comparative extract of what suits your tasks. Save time and nerves for yourself. With ready examples based on your task, not Hello World from a book.

User interface and translations

AI helps you solve basic issues of rapid prototyping of user interfaces, it can help you draw it, layout it, as well as write basic texts, translate them into different languages and much more.

At this point, AI can help you create a full-fledged layout or templates for a product MVP, reducing your time on this routine task.

Context for users (recommendation algorithms)

Another point that is important for the end user is personalization. In combination with social profiles (for example, a user authorized through VK gives access to his data and interests) and AI with a set role of a consultant, you can give the user the necessary content already on the first entry into the application.

How does this roughly work?

• Let's say you have a database of products (clothes) for all ages, genders and in different styles;

• A user goes to your app and authorizes via social network;

• AI with a given consultant context selects the necessary products on the main page of the application (so that a man in his 30s would be given relevant products in priority depending on the season and other parameters);

Technical support

Another method of reducing costs in the development department is AI-based technical support. We connect a chat (for example, Chat GPT), feeding it our knowledge base - and it automatically tries to solve the user's problem. In 70% of cases - this is enough to avoid contacting a real specialist.

Tasks directly related to programming

Okay, now let's get to the most useful area - the development itself. Here we will look at solutions to tasks that come in handy on a daily basis.

Context-based auto-completion

Classic IDEs have had intelligent auto-complete built in for a long time, but the way it works now with all the context of the project, your recent actions and the AI - it works quite fast and snappy.

The only thing to consider is security. If you have any critical code that needs to be confidential - I would consider deploying LLM with auto-complete and writing code based on the entire project context - inside your corporate network (or locally), goodness Open-Source solutions are already plentiful.

Generating code templates

You can quickly generate some code templates based on your finished work, project context or completely from scratch. For example, this can be useful when creating new modules of your product based on some existing architecture (for example, generating models, controllers, views and routings based on the project context with a description of your API scheme).

Naturally, such patterns are worth testing, like anything AI does. But, in general, routine code writing becomes much faster if you don't generate huge classes (which should be avoided in programming based on the principles of SOLID).

Development of basic algorithms

Yes, of course you can use search, but sometimes it's easier to ask GPT to compare several variants of algorithms, run a test for you (for example, by executing on JS code) and based on the most suitable algorithm - create the necessary function for you. Of course, you should check the code for correctness afterwards, but it helps to speed up the research process.

Refactoring

During the product lifecycle, you always have to change some things. For the most part, refactoring should be done manually, but with strict adherence to a normal Git Flow, many routine tasks can still be automated.

As a rule, AI has already learned to identify the main bottlenecks in popular languages and helps to solve them during refactoring. The development of basic algorithms described above also helps.

Writing individual non-critical MVPs

In some cases, AI can also help you develop non-critical MVPs that will not affect your business processes. For example, to sketch a test application for your new service or a basic library to test some hypotheses.

Of course, sometimes AI still gets confused and makes a mistake somewhere. However, if you provide more context and clarify the tasks, the result can be quickly transformed into a basic MVP for your new test project.

Testing and debugging

Yes, you can (and even should) use a combination of human QA, auto-tests, and of course AI. The AI itself will help you cover the necessary classes with the help of the same auto-tests and tell you how to set them up correctly. And where can you go without code debugging.

Here we can also mention the documentation of the code. It is useful to make sketches based on your API schema for the end user and beautifully format it all, both inside the code itself and external documentation files.

Useful tools for you with links

And not to be completely unsubstantiated, we have collected a list of used and tested solutions for you:

• Cursor - one of the coolest IDEs with code generation, refactoring, auto-completion at the press of Tab button and context of your project;

• DocuWriter - A great tool for writing AI-based documentation;

• UX Pilot - perfectly working tool for rapid UI prototyping with support for exporting to figma;

• Replit AI - development assistant and agent for your projects;

• Bito AI - Code review and Refactoring tool with cool plugin for VS Code;

Let's summarize

Today, AI is not replacing the developer, but it is unleashing the developer. Accelerating development through tools based on large language models directly gives you an advantage in the market, but don't forget that you should in any case pump up your skills, because without understanding what's going on, it will be hard to work with AI.

Hello everyone. My name is Elijah, I am the CTO of Neurosell. For the last 7 months I've been developing technological products for ecommerce.

In this article I would like to share a case study of transition from gamdev to fully commercial development in the B2B segment. We will touch upon both difficulties and successful cases.

Before I start, a bit of background: I've been working in game development for 16 years, taking part in projects of various scales - from social browser games to AAA projects on PC and consoles. During these years I've been involved in many things - client and server side, AI implementation, architecture and infrastructure creation and many other things. I managed to work both in small teams and in big studios.

So, here we go.

Why did this happen in the first place?

It all started banal and simple - from fatigue. Somehow, on another heavy Friday night, after a passed release, we met with my good friend Max. Behind the bar we discussed life, different perspectives and projects.

And here, remembering some old ideas for creating applications for VK and Telegram, we gradually (it took more than one evening in the bar), we came to the fact that we wanted to try to change our range of activities and adopt the idea of developing mini-applications for social networks, which later grew into a large product for e-commerce using AI.

Having decided to launch my startup, I, as the person in charge of the technology part, went away to generally remember what it is - product web development and to forge an outline of the plan. And there was a lot to think about - work organization, architecture, technology stack, and more.

Eventually, after a few months (and a few rewrites of critical architecture), we came up with the following stack:

Let's break down why this particular stack?

• NodeJS - In general, in my bearded years I wrote back on puffy, but we live in the 21st century and I wanted to take something faster and more flexible, understandable for further development state and universal. Yes, of course there are still different Java, Python and Go, but here the question of experience, feasibility of some solutions and the possibility of hiring a team quickly arises. Of course, we use Python (for AI services) and C# (more on that below), but the main products work on Node.

• Express - Staying on the node. Why Express? After all, there are more modern solutions. However, my hands, beaten off in gamedev, say that the simpler and smaller the framework, the better. I didn't want to drag dozens of libraries, some of which won't be used in 99% of cases. Simple rationality. And Express is known by all noders.

• C# - For some heavy calculations, we use C#. I think the advantages of compiled languages over interpreted languages don't need much explanation, besides, we use Shared code (partially overflowed from gamedev projects), which can still be deployed in WebAssembly, where JS just can't achieve the results you need in the browser.

• JavaScript - Yes, I'm a stickler for pure JS, however I have to use it minimally. Mainly because there are people in the team who are used to working with typed languages and don't know all the pitfalls of JS.

• TypeScript - It is used for basic development of critical components. So that there are no errors with types. It may be unnecessary, but it often saves life.

• Tailwind - Why Tailwind and not Bootstrap? We actually use Bootstrap too, just not on all projects. Tailwind is more convenient. Bootstrap allows you to develop simple applications faster. That's all for now.

• Sass - For the most part, have given up on pure CSS. It is more convenient to use selectors and work with variables. And in any case we will get CSS at the output.

• Webpack - Why him? It's as old as the world. Well, in fact, even though it is ancient, it is well studied, and it copes with its tasks very well. When properly configured, it does not cause any problems. Hot Reload also helps.

• React - This is where I am impressed with the similarities in our UI frameworks in the gaming industry. React bindings and all that. Yeah and it's just already an industry standard.

• PostgreSQL + Sequelize ORM - We mostly use PostreSQL, for we have our hands full. But in general, nowadays there is ORM and we use Sequelize. Less headache when designing queries, convenient work with schemas and models.

• GraphQL - For internal work of services - GraphQL is used because of its convenience, but as external handles of work with API - Rest is used, because it is more common and understandable to other developers.

• Rest API - For external requests we use the more familiar Rest API.

In general, the base has been discussed. Now let's look at different approaches and experience in both spheres - game development and commercial software development.

What surprised you about the move into commercial B2B product development?

In fact, there are no differences in development approaches. The same modern standards are used everywhere - Git Flow, sprints, retro and so on. However, if you dig a little deeper, on average across the market, there are still differences:

• Reduced development pace - Perhaps, at first glance, the difference may be insignificant, but as my personal research and experience has shown, the gamedev industry has a stricter development pace (especially when it comes to mobile games), which is due to many factors, but the main one is, no matter how ridiculous, a more A/B-oriented and experimental approach to product development, aimed at maximizing profits.

• Customers - In gamedev, as a rule, your product has one or maximum two customers, in the form of a producer and/or publisher. Here (if we take B2B sphere), every new client is a new customer. And that customer may have another customer and so on. This leads to the following point.

• Customers at Customers - In addition to being responsible for the success of your product and business, you are also responsible for the customers at your current customer. The implication of this is that all development and support is broken down into two areas - support for the product itself and its end users (customers), and support that extends to the customer's end users.

• Lower entry threshold - As it may seem, it's much harder to get into gamedev, so there is an unpleasant situation where there are a lot of low- and mid-level low-quality staff on the market (even more so than in gamedev), and as a consequence, there are difficulties in recruiting really cool specialists.

• Market overheating - Huge stack of technologies (let's say in gamedev there are 3 main game engines and a couple more solutions for back end, and in Enterprise it's dozens of technologies just for the front end), a large number of different companies and outsourcing studios, freelancers, market dumping and much more. Well, and also - gamedev community is more close-knit, narrower, though not much inferior in market volume.

Thus, the differences are mostly in the business processes themselves. However, there are also technological solutions that are squeezed to the fullest in game development. We will talk about them below.

What kind of experience from gamdev helps?

Now let's break down what experiences help us the most when transitioning from game development to creating enterprise B2B products.

Complexity of technologies

The first and one of the most important is the complexity of technologies. In the game industry, developers have to study and work with many complex subsystems inside game engines, such as working with graphics, physics, audio data, and optimization. Optimization plays a particularly important role, as developers sooner or later have to delve into a fundamental understanding of optimal handling of CPU, RAM and many other aspects, as games are very complex software where every byte of memory has to be saved.

Architectural solutions and optimization

The second most important thing is architectural decisions. The game, in essence, is a huge monolith with hundreds of different subsystems, which you cannot scatter into microservices on different servers. Some of these systems are quite computationally heavy and it is very important that your architecture perfectly handles all the bottlenecks of the project.

Network solutions

Networking is also one of the most important aspects (especially in online games), where algorithms for instant response of subsystems need to be designed. Unlike corporate solutions, where a notification or message may arrive with a delay of up to half a second, which is not so critical, in real-time multiplayer games, such delays will destroy the product (especially in dynamic projects like shooters).

Teamwork

Another important aspect is teamwork. Game development, as a rule, implies cooperation between different specialists on the whole way of creating a product: programmers, artists, designers, sound engineers and scriptwriters. This teamwork experience is quite more versatile than in the usual corporate development, where a programmer usually communicates with a smaller number of different specialists.

Working with innovation

Working with innovations that often originate in the gamedev itself. Such examples can be new GPUs and new trends such as VR / AR. In general, game development strongly drives progress in hardware, and game developers more intelligently understand how to use its resources more efficiently.

Gamification

Gamification is the whole basis for product-user interaction in gamedev. Hundreds of mechanics for user retention, increasing the average check and time spent in the application were born there. All of this can be used in classic business development.

Thus, the experience gained in game development allows us to create really optimized, high-quality and large-scale projects, using technologies and approaches that have successfully proven themselves in products with multi-million audience.

What experiences get in the way?

In general, from the point of view of the development department, the transition to corporate B2B solutions from gamedev - from a general point of view simplifies the processes of product creation and technology implementation, somewhat reduces the simultaneous load, but on the other hand - some business processes change, burnout factors increase due to double responsibility.

Moments we highlighted for ourselves:

• Business communications with development and understanding of technology are weaker developed than business communications within gamedev, which creates an additional burden due to the extra step of onboarding new customers;

• The business is growing more shakily due to less adoption of product metrics, which 99% of product in gamedev is covered by (don't count indie developers, there's more of a creative story there);

• In gamedev, any mechanics are built around the psychology of the user, their behavioral factors;

• Products are more difficult to launch because of the presence of an additional link in the development chain (the customer);

• The quality of MVP launches should be an order of magnitude higher, particularly if the product approach has not been communicated to the client;

• Less automation in processes. Still a lot of things are done manually.

• Poorly justified technical decisions that may shoot off the knee in the future (chasing unjustified technologies).

It is clear that everything I described is quite individual for each business, I just tried to make an average estimate based on my experience (IMHO).

If you have any other examples or contradictions, I would be happy to discuss them with you.

What would we like to see in the future?

In general, we like to implement all the practices that we have carried with us from game development. However, in the future, we would like to implement the product approach of game development as much as possible, minimizing losses, while restoring the clarity of all processes and development chains to the same level as before.

So far, this has been hindered by some differences in business processes, but I am sure that everything can be adjusted to using a rational approach. It is clear that somewhere something will have to be dropped or completely removed from the work.

What are we doing today?

Today we are developing a whole range of services to create an ecommerce 3.0 ecosystem. We are developing new solutions for e-commerce based on AI, gamification methodologies (with transferred monetization mechanics in gamemade) and personalization of many processes through social mechanics.

You can read more about us on our website.

The first next-generation e-commerce platform with AI on board. Huge range of features. Fast customization. Automatic Promotion.

Try it on your own

A flexible turnkey solution for setting up stores in one day

Handle all online sales, handle shipping, product statuses and other useful features without the need for a large CRM or ERP.

Customize the process as you need it to be

Up to full automation and integration with the payment systems and delivery services you use. Everything will work like clockwork.

• Fully customizable process

• Integrations of your choice

• Personalization of discounts and sales

Your Store as a Service

We will place your store on cloud servers, conduct round-the-clock monitoring and comprehensive data protection. We provide support and help with store improvements.

Why NeuroCart?

NeuroCart is not just a platform for creating online stores, it is also a complete team that will help you make your online sales process perfect

What else is included in cloud services?

Extended support - We provide full technical support, which includes assistance in finalizing your store, integration with the necessary services, round-the-clock monitoring of applications and site, as well as basic assistance for your users.

Assistance in migration to our technology - Do you already have a website but want to move to our platform? We'll help you migrate all the data you need without losing it, including integrations with your automation services (getting product balances, email newsletters and more).

Promotion of the store in catalogs - We will help you place your mobile app in Google Play, Apple AppStore, VK Mini Apps and Telegram Mini Apps. We will integrate all the necessary services, pass moderation and help you to get fichering with promotion.

Economical consumption of resources - We are fully committed to green IT approaches and strive to reduce our carbon footprint as much as possible. Our services undergo regular optimizations and certifications to reduce our environmental impact.

Read More about Cloud >

Team as a Service

We provide more than just a platform for creating stores. You will have a dedicated team of professional developers working for you, ready to always improve and maintain your store.

Many years of experience in developing products of varying complexity around the world

The team members you will be working with have more than 12 years of experience in app, website, game and AI development. We regularly update our knowledge, organize events and share knowledge.

We have worked and successfully created projects for brands such as VK, Microsoft and many other major market players, providing the highest quality of our services. Now - we stand guard over your online store, creating the future of e-commerce.

A team that is always ready to solve any of your problems in online commerce

Would you like to be part of the team?

Take a look at our vacancies >

About Features

Discover the power of Shop as a Service and use the new ecommerce 3.0 approach to successfully trade online.

Not just an online store

Powerful and extensible platform - A modern AI-based tool for creating online stores with a mobile app included. One-stop sales center and integration with the most necessary services.

Marketing and promotion - Publishing to multiple platforms to promote your online store through Google Play, AppStore, VK Mini Apps and Telegram Mini Apps catalogs.

Your own personal development team - Always help with store setup, import and synchronization of products, orders and delivery. Will create new functionality for you or improve the existing one.

Simple management of the entire trade

Our platform is ready to launch and publish your store app on Google Play, Apple AppStore, VK Mini Apps, Telegram Web Apps and other platforms. Get promotion with organic traffic.

Read more about Features >

Open Demo

We have prepared for you a demo website and store control panel based on the NeuroCart Platform.

Go to Demo Page >

Want to get your store up and running with NeuroCart?

We will advise you on any issue of the platform and its components.

Contact Us:

https://ncommx.com/en/contacts/

The main idea was to create an application for users of vkontakte services, where everyone could emphasize the benefits for themselves in terms of creating a business and getting additional advice, including with the help of AI.

We concept an app that helps a start-up business keep their daily diary of successes with tips from AI, learn more about opportunities to start a new business, keep up with news (also picked up by AI) and more.

Realization

The application runs on the core modules of our NeuroCart platform and has a modern NodeJS-based stack under the hood.

As a client part we used the usual JS + Bootstrap with its own components and internal API calls, based also on the services of Bank Tochka and Neurosell, as well as VK Bridge.

Future Roadmap

In the future, we want to bring new elements and functionality to the app that could further help our assistant users, such as:

• Gamification Elements - build your own virtual business, play with friends and win real-world prizes;

• Full-Featured AI Helper - additional AI functionality for businesses such as - assistant in generating names, ideas, product packaging and more;

First Results

Unfortunately, due to some store policies, we weren't able to fully drive organic traffic to our app, but already today (just a week after launch) over 1000 business people are using it.

You can try the service on your own inside the VK Mini Apps:

https://vk.com/app53006182

About Cloud Infrastructure

Cloud infrastructure gives organizations the ability to scale their resources based on business needs. This is especially important for companies looking to adapt to changing market conditions. Cloud services allow organizations to quickly scale up capacity, add new services, or even completely change the architecture of an application without the rigors of time and significant financial investment. This gives businesses the flexibility and adaptability they need to respond to new challenges and opportunities faster than ever before.

Setting up CI/CD practices in a cloud environment greatly simplifies the application development and deployment process. CI/CD is a methodology that allows development teams to automatically integrate code changes, test them, and deploy new software. Using cloud-based platforms for CI/CD solves a number of problems such as reducing the likelihood of deployment errors, improving code quality, and reducing the time it takes to bring new features to market. Cloud providers typically offer tools and infrastructure optimized for CI/CD, minimizing setup and configuration time.

For example, we at Neurosell use roughly the following infrastructure:

Logic and Data Separation

One key aspect worth noting is the separation of logic and data. This concept refers to organizing the application architecture so that business logic is separated from data storage and management. This improves the scalability and maintainability of applications, and makes it easier to implement changes to both components. Cloud platforms provide a variety of data management solutions, including Database as a Service (DBaaS), which allows companies to utilize the most appropriate storage option for their data without having to manage the infrastructure.

To give an example, services may be divided as follows:

• General Application Service;

• Authentication Service;

• Database Server;

• Static Data Servers (CDN);

• AI-Services;

Separation of logic and data combined with cloud infrastructure provides greater security and access control. Users can restrict access to data, providing only those resources required for specific business logic. This, in turn, facilitates compliance with legal and security standards that are becoming increasingly important in today's world, where data breaches can have catastrophic consequences.

In addition, cloud solutions provide companies with the ability to integrate with various APIs and third-party services. This allows them to use off-the-shelf solutions to fulfill specific business processes, reducing the time required to develop their own solutions and allowing them to focus on their core tasks and business goals. These integrations are often done through CI/CD, which automates the process of deploying new functionality and reduces the likelihood of errors associated with manual configuration.

Don't forget about the cost aspect. Cloud services are often provided on a pay-as-you-use model, allowing companies to save on unnecessary resources and optimize costs. This is especially beneficial for SMBs that may not have access to large upfront investments to build their own data centers. With cloud infrastructure and CI/CD capabilities, small companies can quickly launch their products, get feedback from users and make changes, making them more competitive in the marketplace.

In conclusion

In conclusion, cloud infrastructure, together with CI/CD practices and logic/data duality, are powerful tools for optimizing business processes, increasing agility, and enabling rapid response to market needs. Companies that understand and capitalize on these benefits will be able to dramatically improve operations, reduce costs, and secure a sustainable competitive advantage.

You can learn more about how we use cloud services in our other upcoming articles, or on the website of one of our products, NeuroCart.

A little bit about the changes

Briefly combined information about what we do, leaving only the most important things in plain sight - AI-powered app development and gamification of your business processes.

We talk about the key benefits of hybrid app development with the Neurosell team, including high security, end-user promotion, and modern proprietary solutions for store management (and beyond).

Briefly shared why it is important to utilize gamification tools in business. We have been working in the game industry for 16 years, and during this time we have gained quite a lot of experience in increasing retention and profits through meta-gameplay, which fits perfectly into business mechanics.

Share information about your own technologies that make content easier to work with in your apps and stores. Customization has never been easier or more convenient.

Clearly demonstrated a roadmap for working with each project. We value our clients and try to be as transparent as possible, including minimized intrusion into business processes when developing and implementing new solutions.

New Jobs Section

And of course, we have a new section - vacancies. For those who are ready to develop new business solutions in e-commerce and create new technologies based on AI.

Shall we talk about the changes?

We are interested in your opinion on whether the description of our work through the new Neurosell website has become clearer and more transparent.