AI-Powered Product Discovery for a B2B Solar Marketplace
Overview
A B2B solar marketplace operating in the renewable energy ecosystem set out to simplify how buyers discover and evaluate solar products. The platform aggregates products from multiple solar manufacturers and providers, serving installers, EPCs, and enterprise procurement teams that often need to make decisions across a wide range of technical and commercial parameters.
Rysun partnered with the marketplace operator to design and implement Solar Query, a generative AI–powered product discovery and sales assistant built on AWS. The solution enables buyers to interact conversationally with the marketplace, express intent in natural language, and receive accurate, data-backed product recommendations—without requiring deep technical expertise in solar engineering.
Industry
Renewable Energy, Solar, B2B Commerce / Marketplace
Solution
Generative AI, Conversational AI, AI-Powered Product Discovery, B2B Sales Enablement, Data & Analytics
Challenge
United States
Client Context
The customer operates a B2B marketplace that aggregates solar products, specifications, and providers across the renewable energy value chain. The platform serves a diverse buyer base, including solar installers, EPCs, and sustainability-focused procurement teams, each with different levels of technical expertise.
Solar product selection is inherently complex. Decisions depend on multiple variables such as efficiency, performance under different environmental conditions, warranty terms, pricing, and compatibility with specific installation scenarios. Many buyers know what outcome they want but do not always know the precise technical terminology required to search effectively.
As the marketplace expanded its catalog and onboarded additional providers, the limitations of traditional product discovery mechanisms became increasingly apparent.
Business Challenge
The marketplace faced two core challenges that directly impacted buyer experience and platform scalability.
1. Product Discovery for Non-Technical Buyers
Traditional keyword-based search and static filters required buyers to know exact product attributes and technical terms. This made discovery difficult for users who wanted to express intent-based needs such as:
- “High efficiency panels that perform well in low-light conditions”
- “Cost-effective solutions for commercial rooftops”
- “Panels with longer warranty and lower degradation”
As a result, buyers often relied on manual exploration or expert assistance, slowing down decision cycles and reducing conversion efficiency.
2. Data Scale and Consistency
Solar product data was sourced from multiple providers in inconsistent formats, including spreadsheets and CSV files. Manual data preparation and schema alignment slowed updates, increased operational overhead, and made it difficult to keep recommendations current as new products were added.
Relying on subject-matter experts to interpret data and guide buyers did not scale as the marketplace grew. The platform needed a solution that could:
- Translate buyer intent into accurate product queries
- Combine structured product attributes with semantic understanding
- Scale discovery without increasing manual intervention
Solution Overview: Solar Query
Rysun designed Solar Query as an AI-powered conversational sales and discovery assistant that bridges the gap between buyer intent and complex solar product data.
Rather than acting as a simple chatbot, Solar Query functions as an intelligent product discovery layer embedded within the marketplace. It allows buyers to ask natural-language questions, refine requirements through dialogue, and receive contextual recommendations supported by real product data.
The solution was intentionally designed to augment decision-making, not replace human expertise. Buyers remain in control of final selection, while AI accelerates discovery, comparison, and understanding.
AWS Architecture for Intent-Based Product Discovery
Solar Query is built on a scalable, cloud-native architecture using AWS services that combine data engineering, analytics, and generative AI.
1. Intent Interpretation and Hybrid Query Execution
Buyer queries are processed through a serverless orchestration layer that interprets intent and determines how each request should be handled. Depending on the query, the system dynamically executes:
- Structured filtering and aggregation against curated datasets
- Semantic similarity searches across product descriptions
- Contextual reasoning that combines multiple attributes and constraints
This hybrid approach ensures that generative AI is used responsibly and only where it adds value.
2. Retrieval-Augmented Generation for Product Matching and Comparison
To generate trustworthy responses, Solar Query implements a retrieval-augmented generation (RAG) pattern. Product specifications and metadata are indexed for semantic search, enabling the system to retrieve the most relevant products before generating explanations or recommendations.
Generative AI models are used to:
- Translate buyer intent into understandable options
- Explain trade-offs between products
- Provide context around why certain products are recommended
This ensures responses remain grounded in authoritative marketplace data rather than generic model knowledge.
3. Automated Product Data Engineering and Catalog Refresh
Solar product datasets are ingested through automated pipelines that standardize source files, manage schema evolution, and optimize data for analytics and search. This enables faster updates, reduces manual effort, and ensures new products are available for discovery without delay.
4. Governance, Security, and Observability
The platform includes built-in monitoring, access controls, and operational visibility to ensure secure, reliable, and scalable buyer experiences. Usage patterns, performance, and data quality can be continuously observed and refined as the marketplace grows.
Outcomes and Business Impact
Solar Query significantly improved how buyers interact with the marketplace and how the platform scales discovery.
Buyer Experience Improvements
- Buyers can discover suitable products using natural language rather than technical filters
- Non-technical users can explore complex solar specifications conversationally
- Decision cycles are shorter due to faster access to relevant options and comparisons
Platform and Operational Impact
- A majority of discovery interactions are handled autonomously without expert intervention
- Response times for product queries are substantially reduced
- Dataset updates are faster and more consistent through automated ingestion pipelines
- Marketplace teams can scale catalog size and user volume without increasing manual support
Overall, the solution transformed product discovery from a static, filter-driven experience into an interactive, intent-based journey aligned with how buyers actually think and decide.
Why AWS and Rysun
AWS provided the scalable foundation required to combine data engineering, analytics, and generative AI into a single, production-ready solution. Managed and serverless services enabled the platform to scale dynamically while minimizing operational overhead.
Rysun brought deep expertise in building enterprise-grade AI systems for data-rich, technical domains. By combining retrieval-augmented generation patterns with strong data engineering practices, Rysun delivered a solution that is accurate, explainable, and designed for long-term growth.
Summary
Solar Query demonstrates how generative AI can unlock smarter product discovery in B2B marketplaces where data complexity and buyer intent rarely align. By enabling conversational, intent-based exploration of solar products, the platform improves buyer experience, reduces operational friction, and supports scalable growth across the renewable energy ecosystem.
