Why an ‘organisational AI’ model, trained on your own data, is the next big step for housing providers’ operations.
ChatGPT and other AI large language models (LLMs) are revolutionising the way we seek the answers to questions. They’ve enabled access to general best practice and other knowledge in the public domain in a usable and engaging way.
However, while this is fine for general information, what many crave (or need) is something similar for internal organisational queries to support management and operations, improving speed, accuracy and relevance, improving employees’ working environment and efficiency and, ultimately, tenant outcomes.
Proprietary information
For example, an employee asking ChatGPT or Copilot, “What is our ASB policy?”, will yield a response along the lines of, “I do not have information on your own organisation’s ASB policy but I can provide you with general ASB policy information. Would you like me to do this?”.
The reason for this response is because, as stated in the AI’s reply, this type of AI doesn’t have access to your ASB policy documents because they are proprietary to your organisation. The AI hasn’t been trained on them nor would you want it to be because you probably don’t want to share that information with the rest of the world.
Instead, what’s required is a means of augmenting your own current, clean and up-to-date information with the AI’s natural language processing (NLP) capabilities to provide a comprehensive, contextual (to your business) answer to the query. To do this you need to implement your own organisation’s equivalent of ChatGPT using AI RAG.
What is AI RAG?
RAG stands for retrieval-augmented generation. This enhances AI LLMs by combining (or augmenting) two key capabilities:
- Retrieval – The system searches through a locally-controlled knowledge base (of your documents, databases, etc) to find relevant information based on the user’s query.
- Generation – The AI LLM model then uses both the business-specific retrieved information and its own training to generate a coherent, accurate and business-specific response.
In practical terms, RAG works like this:
- When a user asks a question, the system first retrieves relevant documents or passages from a knowledge base that is specifically trained on your data.
- These retrieved passages are provided to the AI RAG.
- This then generates a response that incorporates both this retrieved information and its own general knowledge.
The big advantage of this approach is that your data remains secure. The AI RAG system embeds your proprietary information within its locally-controlled knowledge base and uses it to respond to the user’s query.
The benefits of AI RAG
For housing providers, AI RAG provides numerous benefits:
- Improved accuracy & relevance – An AI solution specific to the business, allowing responses to be grounded in your actual data rather than generic information.
- Up-to-date information – Unlike traditional AI models, which are limited by their training cut-off dates, RAG can access and use your latest company information, documents, and data.
- Reduced hallucinations – By retrieving factual information before generating responses, RAG significantly reduces the risk of AI ‘making up’ (hallucinating) and providing inaccurate responses.
- Domain-specific expertise – RAG enables the AI to become an expert in your services, processes and policies by accessing your proprietary information in a secure and confidential way. It can also incorporate social housing papers, reports and other resources to build a comprehensive expert knowledge base within your own organisation.
- Transparent sourcing – RAG implementations can cite the exact sources used to generate responses, providing an audit trail and helping to build trust with users regarding the answers provided.
- Cost efficiency – RAG can reduce the need for expensive fine-tuning of AI models by using retrieval to handle business-specific proprietary knowledge.
- Knowledge preservation – RAG helps businesses capture historical organisational knowledge from documents, manuals and other sources, making it accessible through a conversational AI interface.
- Integration flexibility – Modern RAG systems can connect to various data sources including documents, databases, APIs and internal tools enabling not only information ingestion, but governance and control over what is used to train the system.
What can AI RAG be used for?
AI RAG solutions offer significant value across a variety of housing functions:
- Customer support – Self-service chatbots using approved company information for handling routine inquiries.
- Internal knowledge management – Employee policy and procedure hub, cross-department knowledge sharing and preservation of ‘corporate memory’.
- Legal & compliance – Contract analysis, legal Q&As, policy and regulatory guidance, and potential compliance risks.
- Research & development – Analyse trends, research and whitepapers.
- IT support & operations – Self-service access to technical expertise from FAQs and other documentation sources.
- Finance & accounting – Tax guidance and compliance, audit support, financial reporting, analysis and insights.
- HR & people operations – Service staff enquiries with a self-service chatbot, HR compliance assistance, and policies and procedures.
- Tenant marketing & communications – Web and content creation, grounded in your own guidelines and style.
- Executive decision-support – Business strategy and operational planning assistance.
- Supply chain & operations – Vendor information such as contracts, emails, licensing and SLA documentation.
The most successful implementations typically start with well-defined use cases that have clear RoI potential, abundant documentation and measurable outcomes. As these sources of information evolve, the system automatically learns from the new material provided, ensuring reliable and up-to-date content, thereby driving efficiency, achieving insights and delivering successful tenant outcomes.
To know more about AI RAG, please don’t hesitate to get in touch with me via colin.sales@3cconsultants.co.uk.
Colin Sales is the CEO of 3C Consultants.
