Publicly listed companies, Investor Relations (IR) Team.
01. Experience (December 01, 2025 - Present)
Investor Relations Expert (IR Expert) – Bittnet Group
- Represent the company in direct interactions with investors via website, WhatsApp and email — available 24/7, no wait time.
- Build trust through professional, coherent and multilingual responses, tailored to each investor's profile.
- Analyzes and interprets annual, quarterly and ESG reports, providing clear and relevant summaries upon request.
- Supports informed investment decisions by correlating data from multiple sources and organizational perspectives.
02. Education and Training (in progress)
Advanced training in capital markets and IR strategy
- In-depth knowledge of financial statements, ESG terminology and international reporting standards.
- Continuous update with market data, legislative changes and trends in the Investor Relations sector.
03. Personal Skills
Multilingual
- Communication in over 90 languages — Romanian, English, French, Spanish, Chinese, Arabic and others.
Financial communication
- Specialized professional language, calibrated for the audience — from the individual investor to market institutions.
Processing capacity
- Correlate and synthesize information from over 2.000 financial reports and documents.
Adaptability
- The tone and style of communication can be configured to reflect the organization's identity and standards.
04. Technical Architecture
Nenos Investor Relations AI Agent is built around a financial-focused RAG engine, powered by LlamaIndex, on a knowledge base that indexes annual reports, quarterly reports, ESG reports, market releases, and company strategy documents. Semantic retrieval allows the agent to answer complex questions like 'how did EBITDA evolve in the last four quarters' or 'what are the ESG commitments for 2026' with precise references to the source documents. Anthropic's Claude models, accessed through Amazon Bedrock, ensure the correct interpretation of financial terminology and the construction of coherent answers in over 90 languages, without degrading translation accuracy.
The interface is built in React/JavaScript and exposed simultaneously across multiple channels — web widget, WhatsApp Business API, and email — through a unified routing service that maintains the investor’s conversational context regardless of the channel used. Flow orchestration is done with LangGraph, with dedicated nodes for identifying the type of question (operational, financial, ESG, strategy), routing to the appropriate index, and synthesizing the final answer. An LLM abstraction layer allows switching or combining models without changes in business logic, and per-channel rate limiting prevents overloading and ensures constant availability. The infrastructure runs on AWS with an on-premises deployment option for organizations with strict market data confidentiality requirements.
Agent security is treated as a stand-alone architectural layer, integrated from the design phase and validated before each delivery. Each instance undergoes a rigorous adversarial testing process that covers the main attack vectors specific to LLM systems, including full coverage of the vulnerabilities cataloged in the OWASP LLM Top 10.