Rezolve Ai Investor Day: Commerce Infrastructure Technology - Sauvik Banerjee - REZOLVE AI PLC

Financial Performance:

• No specific financial metrics (revenue, earnings, margins, cash flow, debt) were disclosed in the provided transcript segment.
• The speaker announced the intent to commercialize the "Database as a Service" (DaaS) platform, noting it was disclosed in the annual report and is "almost there," but no launch timeline or projected revenue impact was specified.

Guidance and Future Outlook:

• Future transactions are projected to occur via AI agents by October 2026.
• The company is shifting from traditional columnar/row-based databases to a distributed, infinite-scale, multi-millisecond blockchain-based infrastructure to handle massive concurrency (e.g., 20 million+ catalogs, 100 million daily transactions).
• Strategic focus is on expanding the commercial rollout of the proprietary "Database as a Service" across all technology stacks.

Business Segments and Product Lines:

• Brainpower Core: A proprietary technology stack comprising industry-specific Large Language Models (LLMs), voice orchestration, agent orchestration, and an ethics/compliance layer.
• Proprietary Models: Trained over 10 years on single-industry catalogs (300 billion parameters, 30 billion tokens) rather than general internet data to prevent hallucination and ensure accuracy.
• Traceware: A "maker and checker" framework that validates agent actions against logs to ensure reported transactions actually occurred, achieving 99.5%–100% accuracy.
• Auditable AI: A platform that benchmarks outputs against other LLMs to ensure neutral judgment, observability, and monitoring of AI behavior.
• Resolved Provenance: Technology embedding invisible watermarks into images, video, audio, and text for verification and anti-counterfeiting purposes on Product Detail Pages (PDPs).
• Resolve AI Labs: A global research division with teams in the US, Europe, South America, Asia, and India.

Market and Competitive Landscape:

• The market is moving toward agent-driven commerce where AI agents swarm to handle discovery, checkout, and loyalty.
• Current generic LLMs are described as lacking necessary industry-specific knowledge (e.g., internal catalog data like jacket use cases) and prone to inaccuracy.
• Competitors using general internet-trained models face risks of hallucination and model drift, which Resolve AI aims to eliminate through proprietary, catalog-trained models.

Risks and Challenges:

• Hallucination and Model Drift: Mitigated by training on proprietary catalog data rather than public internet data.
• Agent Reporting Errors: Addressed via Traceware to prevent agents from reporting non-existent transactions.
• System Scalability: Traditional databases are prone to choking during high-concurrency events (e.g., festive days), driving the need for a new blockchain-based architecture.
• Brand Erosion: Risks associated with AI errors impacting retailer brand reputation are managed through strict ethics and compliance stacks.

Management Commentary and Tone:

• Tone is highly confident and visionary, emphasizing Resolve AI's "DNA" of invention and pioneering deep tech.
• Management positions the company as a solution to the limitations of current generative AI, highlighting the necessity of industry-specific training and rigorous verification.
• The speaker emphasizes a "maker and checker" philosophy, ensuring no single model operates unchecked.
• CTO Shovik Banerjee frames the company's mission as solving for "problem-solving utopians" with a focus on deep technology and global collaboration.

Analyst Questions and Answers:

• No analyst questions or Q&A session details were included in the provided transcript; the segment ends with the speaker handing over to other executives (Elizabeth and Steve Perry).

Other Key Points:

• The technology stack is designed to support a "flywheel" of data: get chosen -> get paid -> get loyalty -> get chosen again.
• The architecture supports multi-form factor functionality, allowing provenance technology to work offline.
• The company is building a "database as a service" capable of handling infinite scaling and multi-millisecond latency, contrasting with legacy database limitations.