AI & Developer infrastructure

AI assistant for ERM workflows

"I've been very impressed at their openness to innovate, their openness to collaborate. And not just be an organisation that's there to do what they're told, but an organisation that's there to make our organisation much more technologically stronger."

Matt Hyne, CTO at Protecht

Client

Protecht is a leading enterprise risk management software and advisory company, headquartered in Sydney, Australia, with operations across the US, UK, and APAC. Founded over 25 years ago by risk management experts, Protecht serves 400+ organizations - including governments, regulators, and central banks - across banking, financial services, insurance, fintech, manufacturing, and beyond. With 200,000+ users on the platform and a 98%+ customer retention rate, Protecht is one of the most trusted names in GRC globally. Hotovo has been a key development partner for over 15 years.

AI assistant for enterprise risk management blog Cognita Protecht

Case study

The need

Protecht was developing an AI-powered chatbot to extend capacity in a market where AI-driven assistance is becoming standard.

In enterprise risk management, incomplete or inaccurate data does not only slow teams down, it can also create compliance blind spots that expose organizations to regulatory penalties and undetected threats. Protecht was focused on developing an AI assistant that could guide users in real time, reduce form errors, and support accurate data entry.

The solution also needed to improve usability while strengthening Protecht’s position as an AI-enabled GRC platform.

Core issue

The development team in Košice led the React frontend, collaborating across multiple timezones.
The chatbot also had to integrate into an existing large-scale enterprise application in active development for 20 years, with 15+ modules and 400+ customers, without disrupting ongoing development.
The solution was delivered on time and launched without any integration failures despite parallel development across two continents.

Our solution

Hotovo took full ownership of the frontend, with additional contributions to backend development and QA delivering a production-ready solution on time.

Key technical capabilities delivered:

Real-time streaming AI responses — answers begin appearing immediately as they are generated, with the ability to interrupt generation
Smooth streaming text experience achieved through adaptive rendering that preserves word boundaries and readability during generation
Scalable conversation history handling using virtualized message lists to maintain UI performance even with large chat histories
Context-aware AI suggestions system integrated directly with 10+ form fields for intelligent field-value autocomplete and contextual navigation
Multi-format AI interactions supporting plain responses, contextual suggestions, knowledge base answers, form assistance, and save/submit confirmations
Rich text AI responses with support for structured lists, headings, and other formatted content for improved readability
Per-module enablement controls to activate or deactivate the chatbot contextually
Error boundaries isolating chatbot failures from the main application, ensuring chatbot issues never impact core risk management workflows that are critical for enterprise reliability and compliance.
Extensive 90%+ test coverage across the chatbot codebase to ensure reliability and maintainability
On the backend side, built the initial proof-of-concept AI service architecture with Spring AI integration, contributed to register schema generation for the LLM model, implemented supporting UI changes in the existing application layer, and improved error handling
QA team was involved from the early stages, building robust test automation coverage across E2E functionality, quality of AI responses, and LLM-specific scenarios

Key results

Cognita - Protecht's flagship AI product, now available to 400+ organizations and central to their GRC market positioning as an AI-enabled ERM platform
Full audit trail and explainability built in - every AI-assisted field is traceable, meeting compliance obligations with regulators and central bank clients
Faster task completion for end users working with complex forms and risk management workflows
Reusable, well-tested codebase as a foundation for expanding AI across the platform and supporting Protecht’s product roadmap
Accelerated delivery through AI-assisted development across the entire team
Delivered on time and showcased at the Gartner Enterprise Risk, Audit & Compliance Conference in the USA, attended by 1900 participants.

Tech stack

Frontend: React, TypeScript, Redux Toolkit with RTK Query, Material UI, Jest, React Testing Library
Backend: Java, Spring Boot, Spring AI, AWS Bedrock, REST
QA: Robot Framework, Playwright, ReadyAPI, custom tools (made with NodeJS, NextJS)
AI IDE: VS Code + Roo Code (LLM Provider: Requesty), IntelliJ Idea (Cline AI Agent)

Team size

4 Front-End Developers (including AI UI/UX integration specialists)
1 Back-End Developer (AI service architecture & LLM integration)
2 QA Engineers (including LLM-specific testing & AI response quality evaluation)
1 DevOps Engineer
1 Business Analyst
AI-assisted development across the entire team using Roo Code and Cline AI Agent

Explore other projects