EffectiveSoft

Client and challenge

The application of our client, Kilowatts Co., was designed for Tesla owners to interact with their vehicles while driving. However, manual input diverted drivers’ attention from the road, leading to safety risks. The EffectiveSoft team was chosen to integrate a GenAI-powered voice assistant into the application to help drivers stay focused on the road.

Solution

Our team created a voice assistant and integrated it into the Kilowatts application. Deployed on Microsoft Azure, the solution’s back-end architecture was built as microservices. OpenAI Realtime API powered by GPT-4o-Realtime LLM and NLP techniques like sentiment analysis allowed for voice streaming and processing in real time.

How the voice assistant works:

• Through continuous audio streaming and audio-to-text transcribing, it creates a natural conversation flow and improves accuracy.

• Records and preserves unique speech characteristics.

• Shows understanding of different topics and contexts.

• Generates suitable responses, adapting to drivers’ emotional state.

• Performs different commands like ordering items, booking rooms, and creating routes.

With the help of the voice assistant, drivers can use voice input to create orders in stores, hotels, restaurants, and coffee shops or find the nearest charging station. The GenAI-powered assistant can also plan routes and send them to the vehicle’s navigation system, control battery level, and give proactive suggestions to drivers.

Result

The EffectiveSoft team successfully completed the project, enriching the Kilowatts app with a voice assistant. This invisible AI companion enables safety on the road through hands-free interactions with drivers while also fulfilling drivers’ requests, anticipating their needs, and showing empathy.

Tech stack

C#, Go, JavaScript, RESTful APIs, WebSockets, Azure Functions, .NET, CLR, Node.js, OpentAI Realtime, OpenAI Assistant, OpenAI Chat, Docker, Azure

Client
Our client, a high-touch brokerage company (under NDA) operates in a transaction-driven environment with multiple external stakeholders. The client aimed to modernize its operations by replacing fragmented manual coordination with a unified digital platform.

Challenge
The company’s daily operations relied on email threads, messaging apps, spreadsheets, external data platforms, and manually prepared documents. As transaction volume increased, these disconnected tools slowed coordination and made workflows difficult to scale. The client envisioned a centralized web-based system capable of integrating external data sources, managing documentation and payments, supporting real-time monitoring, and applying AI to reduce repetitive work. However, the technical feasibility of such a complex system—and the boundaries between automation and core system logic—were unclear.

Solution
EffectiveSoft conducted a structured discovery phase to assess feasibility and design the platform architecture. Through a series of workshops, we mapped the full operational workflow, defined integration layers, and established clear boundaries for AI-driven automation. A key architectural principle separated deterministic backend logic—responsible for calculations and critical validations—from AI capabilities focused on parsing requests, structuring communications, and assisting with document preparation. The discovery phase resulted in a modular, browser-based platform architecture designed for phased scaling, with centralized communication ingestion and structured integrations with external data providers.

Result
At the end of discovery, the client received a comprehensive Software Requirements Specification (SRS) translating the target operational model into implementation-ready functional and technical requirements. The document formalized integration logic, automation boundaries, and architectural principles while outlining risk mitigation strategies.

Context

An international credit management company supports customers with contract compliance, invoicing, and repayment control. The team needed higher operational efficiency and the ability to handle more customer requests within existing capacity. EffectiveSoft was engaged to identify high-impact AI use cases and outline an implementation plan.

Our approach

We started with an end-to-end mapping of the company’s tasks and processes. We interviewed employees, reviewed the tools the team uses to identify automation opportunities, and documented the sources of rework, slow decision-making, and extended response times. 

From there, we pinpointed operations where AI-powered data analysis, predictive analytics, and intelligent decision support would deliver the strongest impact. For each integration option, we defined the required AI technologies and solutions, then assessed feasibility, resource needs, and expected ROI. 

Finally, we prioritized initiatives, set the execution sequence, and built a strategic roadmap with timelines, milestones, and required resources.

Result

We delivered a step-by-step AI implementation plan with clear priorities, sequencing, and resource requirements. Our first delivered solution—a RAG-based LLM for customer queries—handled around 60% of incoming queries without human involvement, reduced average response time by 38%, and lowered cost per request by 18%. Further initiatives from the road map are underway.

Tech stack 
Jupyter Notebook; Google Colab; Tableau; PowerBI; Hugging Face; Python; Tensorflow; AWS; Azure; GCP

We developed an AI-enabled chatbot to increase the efficiency of company processes.

Challenge

With the growth of our company, some departments, including HR, IT, and Personnel, faced an increasing amount of inquiries from employees. Moreover, the internal knowledge base grew, making it problematic to find the necessary information quickly. These issues led to operational inefficiency, increasing the time for query processing and delaying the completion of the primary tasks by employees. We decided to create a 24/7 service that would support and assist company employees by answering their questions and providing queried information while also reducing the load from HR, IT, and other departments.

Solution

Our team designed a virtual assistant for internal company use. The chatbot is powered by a generative AI model to enable interactive, human-like conversation. The solution is implemented using RAG architecture and an LLM model that is hosted within the company’s infrastructure. The chatbot uses the internal knowledge base, reporting system, HR platform, and other company resources to generate responses. To enable instant access for employees, we integrated the assistant into the internal messaging system. Using the assistant, the company employees can:

• Quickly and easily access information that is stored in the internal knowledge base.

• Request information about the company and employees. 

• Get notifications about their work logs. 

• Receive important company announcements and reminders.

• Get information about their leave balances and make leave requests.

Result

The solution we created is an easy-to-use and always-available service for the company employees. The virtual assistant improves employees’ productivity and streamlines processes by automating tasks and reducing the time necessary to find information. 

Tech stack

Python,  LangChain, Chroma, llama3.1-70b, Sentence Transformers.

Context

Together with a London-based B2B SaaS company, we are building an ML-based investment analytics platform for the maritime domain. Users build dashboards to compare multiple vessels across different parameters, including valuation and depreciation trends and forecasting. As dashboard configuration was still largely manual, even a standard workspace could take hours to assemble.

AI-assistant

To reduce that effort, we considered BI tools first, but they couldn’t support the required level of customization. Instead, we implemented an AI assistant that converts natural-language requests into JSON configurations. Under the hood, there are two agents:

• A configuration agent that generates JSON for charts, tables, and dashboards
• An intent agent that detects requests requiring a specific output and routes them to a predefined backend function

To keep outputs consistent and reliable, we kept the AI layer out of calculations. We also described each JSON parameter and its purpose, added a set of few-shot examples to the prompt, and implemented two-step validation (required fields + logical checks).

Result

Based on client usage analytics, the pilot showed a ~60% reduction in dashboard configuration time for complex, multi-vessel scenarios. The next step is to extend the assistant’s capability, enabling users to return with follow-up inputs, adjust dashboards iteratively, and work across multiple projects within a single workspace.

Tech stack

Azure OpenAI Service; Microsoft Azure; .NET Core; Python; React.js; Highcharts.js; MongoDB; Docker

Client

Our customer is a startup that created iOS and Android native apps combining real-time multiplayer gaming with a messenger. In these apps, people can play and at the same time communicate with friends one-on-one and in group chat rooms or make new friends in public chat rooms.

Challenge

As the number of users and their message history were growing, the customer started experiencing operational challenges related to the server-side and the performance of an Android native app. To scale and deliver new features faster, the company decided to replace the monolith server-side written in the Dart programming language with microservices. The customer’s team also needed an admin panel to maintain the app more effectively.

Apart from the server-side migration, the customer wanted to add new functionality to the Android client app and resolve issues such as bugs, crashes, and slow rendering.

The engineers at EffectiveSoft joined the client’s team to: write lightweight microservices using the Go programming language (also referred to as Golang), create an admin panel for managing all microservices, and participate in the further development of an Android app.

Result

Our engineers have been working as part of the customer’s team for almost 4 years.

After migration to microservices:

-the time to market shortened as it became easier to develop, test, and deploy the app

-the app’s resilience improved as independent communicating services are easier to change, maintain, and scale.

The work on stabilization, optimization, and advancement of the Android client app resulted in:

-a higher crash-free rate – from 95% to 99.7%

-a consistently high frame rate for users with a lot of data on the screen

-organic growth of the app’s audience

-revenue generation from in-app purchases.

Client and challenge

Our client is a fintech startup that successfully runs a crypto trading platform. Initially, the platform was created as a web app but its design was not adaptive. As a result, platform users who wanted to conduct their operations via a mobile phone faced difficulties and inconveniences. Due to our extensive expertise in creating mobile applications, the client selected us for the project.

Solution

We started with wrapping the website in a mobile shell. However, this was a workaround, and the client was looking for another option that would be suitable for end-users and could be launched within a tight schedule.

We developed a hybrid mobile application for iOS and Android on top of a web platform backend system. We focused on thoughtful, intuitive design with clear user flows.

To provide liquidity from multiple exchanges and ensure fast order fulfillment, KoinKoin has partnered with BitHolla, a company that provides infrastructure for trading software, wallet management and digital marketplaces. Security is ensured due to blockchain, which is used to store all digital assets. Multi-stage verification and two-factor authentication enable compliance with security standards.

To extend a user base, we developed a referral program. For real-time communication between users, we integrated a Trollbox chat.

Result

The MVP solution was a success. As a result, the next project stage — including a significant extension of the app’s functionality — was agreed upon, and work has already begun.

Tech stack

React Native, React Navigation, Form React Hook, React Native Firebase, React Native Picker, Redux Toolkit, Moment, React Query, Sentry, Axios, i18next, React Native Async Storage, React Native Vision Camera, React Native WebView, Styled Components

MedScope America Corporation is a privately held medical alert company in Pennsylvania. EffectiveSoft has customized Microsoft Dynamics CRM system, developed all the important medical portals (such as call support, partner, advanced installer, orders fulfillment and job posting portals), automated a great amount of external and internal tasks enabling fast PERS and care delivery.

https://www.effectivesoft.com/portfolio/microsoft-dynamics-crm-customization.html

EffectiveSoft developed a Windows-based trading platform and mobile trading apps for City Index. The desktop version of the trading platform has the following benefits: powerful analytics tools, high speed, advanced charts, due to the optimized RAM usage, the application works without interruptions for a long period of time. Trading mobile app, City Trading™, is developed for iOS and Android, it can be fully and securely accessed from phones and tablets. City Trading™ mobile app won the title of ‘Innovation of the Year’ and was awarded as “Best Mobile Trading Platform” by the MoneyAM Online Finance Awards.

https://www.effectivesoft.com/portfolio/city-index-trading-applications.html

DICOM Viewer software is medical imaging software which helps medical specialists read, analyze, transmit medical images in the DICOM format. It is a scalable and customizable solution able to integrate with PACS (Picture Archive and Communication System) and RIS (Radiology Information System).

https://www.effectivesoft.com/portfolio/algom-dicom-viewer.html

Sentral Clinical Research Services, LLC (SCRS) is a US company specializing in medical research of new drugs and medical devices. EffectiveSoft developed a secure unified document management system – SentralBinder. It is intended to speed up document exchange and ensure participants' mobility in clinical trials due to the access from different devices – PCs, iPads.

https://www.effectivesoft.com/portfolio/sentralbinders.html

EffectiveSoft developed the front- and back-end of Instrument Console from scratch on the basis of the provided essential functional specifications and wireframes created by the company’s designer. The Console is based on HIPAA-compliant features which ensure a strict data access model for user/group/role/permission mapping and data access. It means that users are only allowed to see the data they need for performing their tasks. All the information is encrypted with the help of the RSA algorithm.

Key features:

• Top-level administration and role-based access
• Patients and data management
• The ability to choose scan patterns with a different number of points to examine the eye texture, etc.

https://www.effectivesoft.com/portfolio/bosstm-instrument-console.html

EffectiveSoft enhanced the functionality of the server and enriched the trading software system with the key features: customized traders’ reports, traders’ performance evaluation tools, order management, order execution algos, prime brokers integration, the ability of the system customization according to trader’s screen specifications and more.

https://www.effectivesoft.com/portfolio/edge-trade-execution-system.html

EffectiveSoft developed enterprise software – CloudController which is intended to automate virtual machine deployment, virtual SAN/SDN deployment, application virtualization and deployment. The company’s designers developed AJAX based design which boasts the ability of advanced charting representing a full overview of the system load with real-time statistics. CloudController includes a web interface, Microsoft Windows services, custom tools and is available in 8 languages. CloudController was named “Best of VMworld” in the category of “Cloud Computing Technologies”, the world’s first independent cloud management platform.

https://www.effectivesoft.com/portfolio/incontinuum-cloud-automation-platform.html