Welcome to our evolving guide to maximizing the benefits of Synthesis AI Search. This guide captures early best practices shared by participants in Synthesis AI Search private beta. The Knowledge Architecture team is committed to working hand-in-hand with our community to continuously refine Synthesis AI Search, and this guide.
Synthesis AI Search achieves the best results when firms maintain high-quality essential content in Synthesis, and end users regularly rate their search experiences, allowing fine-tuning of content based on their feedback, and providing Knowledge Architecture with insights into search performance. This collaborative approach will allow us to unlock the full potential of Synthesis AI Search.
Before and During Beta
Below you will find a group of tasks you will need to complete to prepare for Synthesis AI search beta, and then systematize to improve your content during beta and beyond.
Migrate Videos to Synthesis Native Video
Synthesis AI Search will only search the content of videos in Synthesis Native Video. There is a wealth of knowledge in educational videos, technical documentation, town halls, and other video types which can be unlocked and put to use by Synthesis AI Search. The content of embedded video, such as Vimeo and YouTube, will not be searchable.
Upload Missing Essential Content
Are there are categories of essential content (project management manual, proposal archive, brand guidelines, etc) you’ve been meaning to upload to Synthesis but haven’t prioritized? Are there questions which you know people will be excited to ask Synthesis AI Search, but the content to answer those questions lives elsewhere? It would be a great time to upload that content to Synthesis.
In other words, let Synthesis AI Search beta be a catalyst for rounding out your firm’s knowledge base.
Review Existing Essential Content
If you don’t have a formalized process for reviewing your intranet’s content on a regular basis, consider creating a system prior to beta. You can use our approach to Essential Content Maintenance, or invent your own. Reviewing your intranet for redundant, outdated, or unused content will be beneficial during beta and beyond.
One of our Synthesis AI Search private beta participants discovered that the HR team was keeping all of the past Open Enrollment and Benefits content on their intranet going back several years. After pointing out the challenges that would present to end users who were searching for the latest information, the HR team moved the content from prior years to an archive server.
Review Old Posts
Some post topics — project stories, awards, company milestones, etc. — are evergreen and should remain on your intranet. Other post topics — policies, software versions, procedures — can become outdated quickly and should be deleted to prevent confusion for your end users.
One of our Synthesis AI Search private beta participants targeted HR, IT, and Design Technology Posts for review, as well as a couple key hashtags used to communicate timely information. Using filtering and bulk delete features of Stream Administration, they were able to quickly eliminate outdated information and improve the search experience immediately.
Consider Converting Documents into Pages
Synthesis AI Search will search across both Synthesis Documents and Synthesis Pages, but the predictable structure (headings, collapsible sections, blocks, tables) of Synthesis Pages will provide a better search experience. Some Synthesis Documents w/ graphical requirements—past proposals, graphic standards, specifications, company templates, etc.—should remain documents in Synthesis. However, other documents—policies, procedures, technology guides, etc.—can easily become Synthesis Pages. By converting Synthesis Documents into Synthesis Pages you not only improve the search experience, you make the content easier to update. If you have a choice, use Synthesis Pages.
Pro Tip: Adding structure (especially headings and collapsible sections) to your Synthesis Pages can help make the content perform even better in Synthesis AI Search.
Consider Joining the Community AI Program
Community AI is an opt-in program that empowers Knowledge Architecture to create AEC-specific AI Infrastructure by training AI models on client content for the mutual benefit of participating firms, while also protecting confidentiality and intellectual property.
When released in 2025 to members of our Community AI Program, our AEC-specific embeddings model will be used during the retrieval phase of AI Search to find the most relevant intranet results to summarize. The AEC-specific embeddings model will understand industry jargon, acronyms, synonyms, and the meaning of common terms in our industry better than a generic model.
If you haven’t already opted in to our Community AI Program, you can review our Community AI Program: Overview + FAQ, and opt in here: https://www.knowledge-architecture.com/learning/community-ai-program-overview-and-faq
During Beta
Once you have access to the Synthesis AI Search beta, you’ll be able to proactively test and improve your content, as well as quickly respond to user feedback. Here are some tips to get you started.
Test Common Search Terms
One of the best ways to prepare your intranet and your employees for AI Search is to test common search terms. Here are some thoughts on how to come up with your testing list:
Use Knowledge Architecture’s Common Search Term List
Test Search Terms for Synthesis AI Search
This list provides many of the top search terms used across all clients. Feel free to adapt it to meet your needs. Caveat: this list was generated using keyword search data as it does not yet reflect any change in search behavior which may be introduced by the enhanced capabilities of Synthesis AI Search to handle natural language queries and questions.
Ask Your Team for Common Questions
One of our early adopters went to all the department heads and asked for the most common questions they get from employees, and then tested Synthesis AI Search on those questions. Brilliant idea!
Encourage Your Beta Testers to Rate Search Summaries
Let your beta testers know why rating Search Summaries is important.
Ratings Will Help You Improve Your Content
Global Administrators and Search Optimization administrators will see all of the Search Summary ratings. As much as you “pre-test” AI Search, there is no substitute for real users who have real questions they want answered. By rating their searches and providing useful feedback, your beta testers can alert your team to content gaps as well as outdated/incorrect content in time to fix them before rolling out Synthesis AI Search to the whole firm.
Ratings Will Help Knowledge Architecture Improve Our Technology
Our search quality team sees every Search Summary Rating from across all clients. In some cases the issue with a search is purely a content issue, which we can safely ignore. However, in other cases, Search Summary Ratings provide insights into search algorithm improvements we can make. Getting feedback directly from end users during private beta has already helped improve search quality, and this virtuous feedback loop will continue to help us continually improve Synthesis AI Search.
Please feel free to reach out to your Client Success Manager if you’d like more guidance with any of the above, and send any comments or questions to support@knowledge-architecture.com.