General Benefits of Synthesis AI Search
Q: What is Synthesis AI Search, and how does it benefit AEC firms?
AI Search combines next generation search technology with AI to instantly generate rich answers to natural language searches and questions by summarizing the most relevant content on your Synthesis intranet. In addition, Search Summaries include citations, enabling your users to verify AI-generated answers and dig deeper into relevant content.
Synthesis AI Search enables principals, project managers, and teams to quickly find actionable answers to their questions, saving time and improving decision-making.
Q: How does AI-powered search differ from traditional keyword-based search?
AI-powered search goes beyond matching keywords. It understands the context and semantic meaning of queries, enabling users to ask natural language questions like, “Why is my Revit model slow?” and receive synthesized answers from multiple sources with citations, rather than just a list of documents.
Privacy + Risk Management
Q: Which Large Language Model (LLM) does Synthesis AI Search Use?
Synthesis is built on Microsoft Azure and we're using the Azure OpenAI Service with ChatGPT. The version itself will change with new models, pricing, and availability. We're currently using ChatGPT 4.1 mini.
Q: Is our data shared with the ChatGPT LLM for model training purposes?
As you can see in this documentation from Microsoft, the Azure OpenAI Service does not use customer data to retrain models.
Q: What measures are in place to ensure the privacy of sensitive firm data?
Synthesis AI Search adheres to your firm’s permission structures. Only authorized content is passed to the language model for processing, and no data is shared externally.
Q: How does Synthesis AI Search generate accurate answers?
We use Retrieval-Augmented Generation (RAG), which grounds search results in your firm’s data. Our private large language model only generates answers based on the most relevant content retrieved from a real-time search of your internal knowledge base, ensuring responses are up-to-date, accurate, and verifiable.
Q: What is Retrieval-Augmented Generation (RAG), and why is it safer than direct LLM-based search?
RAG combines real-time retrieval with AI generation. It ensures the large language model only generates answers using the data retrieved from your firm’s sources, avoiding inaccuracies or hallucinations common in standalone LLMs.
Q: How does Synthesis AI Search handle permissions to ensure users only see what they’re authorized to?
Permissions are honored during the retrieval process. Real-time searches ensure that only data the user is authorized to access is passed to the language model, so sensitive or restricted information remains protected.
Q: How does Synthesis AI Search ensure the freshness and accuracy of results?
By executing real-time searches, Synthesis AI Search retrieves the most up-to-date and relevant content for every query. This minimizes the risk of outdated or stale information appearing in search results.
Beta Provisioning
Q: Do we have to make Synthesis AI Search available to all our users at once, or can we roll it out gradually?
Synthesis AI Search can be rolled out gradually. You can start with a smaller group of users, such as leadership or a pilot team, to test the system and gather feedback before expanding access. This approach allows your firm to fine-tune the search experience and ensure it aligns with your needs before a full-scale rollout.
Community AI
Q: What’s the difference between AI Search and Community AI?
Synthesis contains AI powered features already and will continue to contain more over time. You do not opt in or out of this code, it is just part of the product. Synthesis AI Search is our newest AI feature.
While Synthesis AI Search is in beta you have the choice to opt in or not, once it goes into GA it will become part of the product for all firms.
What you opt into (or not) with Community AI is allowing us to train our AI models on your data, and the benefit you receive from that (if you opt in) is getting versions of AI powered features that are trained on AEC data.
If you do not opt in to Community AI you can still use all of our AI powered features, but they will be powered by generic models instead of AEC-specific models.
Q: What is the Community AI program, and how does it benefit participating firms?
The Community AI program is an opt-in initiative where participating firms contribute de-identified content to help train AEC-specific AI models. In return, they benefit from enhanced search relevance tailored to architecture, engineering, and construction (AEC) contexts.
Q: Can my firm opt out of the Community AI program while still using the core features of Synthesis AI Search?
Yes, participation in the Community AI program is optional. By default, all firms are opted out and use generic AI models unless they explicitly choose to join.
Q: What are AEC-specific AI models, and how do they enhance the search experience for architecture firms?
AEC-specific AI models are fine-tuned on industry-specific terminology, concepts, and data. They provide more precise and relevant results for queries related to architecture, engineering, and construction, improving the overall search experience.
Q: Where can I learn more about Community AI?
You can learn more about our Community AI program by visiting the Community AI Overview + FAQ page on our website.
Q: How do I sign up for the Community Program?
Once your team is ready to move forward, you can sign up here.