Loka helps BlocPower use AI to turn thousands of aging buildings into green machines
Industries
Climate tech
Tech & TOOLS
Amazon S3, Amazon SageMaker, Amazon Bedrock, AWS Lambda, Amazon DynamoDB, Amazon ElastiCache (Redis), Amazon Aurora (pgvector), Anthropic Claude
Teams & Services
AI & Agentic, Data & Analytics,
AWS Engineering
milestones
RAG-based AI assistant for building retrofit planning
Industry
Climate tech
Tech & TOOLS
Amazon S3, Amazon SageMaker, Amazon Bedrock, AWS Lambda, Amazon DynamoDB, Amazon ElastiCache (Redis), Amazon Aurora (pgvector), Anthropic Claude
Teams & Services
AI & Agentic, Data & Analytics,
AWS Engineering
milestones
RAG-based AI assistant for building retrofit planning
Loka’s GenAI solution helps BlocPower turn massive amounts of building data into actionable plans for cleaner, greener properties.
The situation
BlocPower is on a mission to transform aging properties into smarter, greener, and healthier buildings. By combining building upgrades, construction management, and financing, the climate tech company helps property owners improve energy efficiency while creating long-term value for communities.
Transforming thousands of buildings requires BlocPower to evaluate enormous amounts of information about individual properties. Each project can involve building specifications, images, equipment details, energy data, and other unstructured information that must be analyzed before teams can determine the equipment and installation processes required for a retrofit.
BlocPower partnered with Loka to explore how generative AI could make that process faster and more scalable.
The challenge
Planning building retrofits requires understanding large volumes of complex, unstructured data. With thousands of potential projects, manually organizing and interpreting that information can create a significant operational burden and make it difficult to scale efficiently.
BlocPower needed a way to clean and organize its existing data while making that information easily searchable and useful for project planning. The solution also needed to understand more than text. Building projects incorporate images and other property-specific information that can provide important context when determining the equipment and processes required for an energy upgrade.
To accelerate this work, BlocPower needed an AI solution capable of retrieving relevant information across its knowledge base and turning complex building data into actionable recommendations.
The solution
Loka cleaned and organized BlocPower’s extensive building data and transitioned it into a searchable knowledge base designed for generative AI. Loka then developed a retrieval-augmented generation (RAG) chatbot powered by Anthropic Claude that can understand text and images associated with building projects.
The AI assistant retrieves relevant information from BlocPower’s knowledge base and helps teams plan the equipment and installation procedures required to transform aging properties into more energy-efficient buildings.
Built on AWS, the solution uses Amazon S3, Amazon SageMaker, Amazon Bedrock, AWS Lambda, Amazon DynamoDB, Amazon ElastiCache with Redis, and Amazon Aurora with pgvector. The architecture provides a scalable foundation for bringing generative AI into BlocPower’s building transformation workflow.
What we delivered
AI-ready building data: Loka cleaned, structured, and organized BlocPower’s large volumes of unstructured data, transforming it into a searchable knowledge base that can support AI-powered workflows.
Multimodal RAG assistant: The Claude-powered chatbot uses retrieval-augmented generation to understand text and images, retrieve relevant building information, and provide context for individual retrofit projects.
AI-assisted retrofit planning: The solution helps teams determine the equipment and installation procedures required for building upgrades, making complex project information easier to access and apply.
Scalable AWS architecture: Loka built the solution using AWS services designed to support secure data storage, AI and machine learning workloads, vector search, caching, and scalable application processing.
Repeatable knowledge architecture: By organizing BlocPower’s data into a searchable AI knowledge base, Loka created a foundation that can support growing volumes of building information and future GenAI use cases.
Project results
BlocPower is using generative AI to make building transformation planning easier, faster, and more scalable, supporting its broader mission to turn aging properties into climate-friendly spaces. The RAG-based solution helps teams work more efficiently with massive amounts of building data, retrieve relevant information, and plan the equipment and installation processes required for energy upgrades. Key results include:
5,000+ buildings retrofitted and counting
Faster access to relevant building data through an AI-powered interface
Less manual effort navigating fragmented information
More efficient planning for equipment and installation requirements
A scalable foundation for accelerating climate-friendly building transformations
