Skip to main content

Release Notes: AI Search Public Beta

Christopher Parsons
  • Updated

Current Version: Beta 1.1 

(Released: Public Beta - June 6, 2025)

The core focus of Synthesis AI Search Beta 1.1 is Advanced Employee Search.

We've improved AI Search results for employee skills, licenses, certifications, roles, office, project experience and many other query types.

Here are some examples of queries we've improved in Beta 1.1: 

  • “who is our bluebeam expert?”
  • “who can help me with contract negotiations?"
  • “who is the go to person for mass timber?”
  • "how many PEs do we have in Iowa?"
  • "who is on the sustainability committee?"
  • "who is the managing principal for Seattle?"

More detailed release notes on Advanced Employee Search can be found here.

Next Version: Beta 1.2

We're actively working on the next evolution of AI Search — Advanced Project Search, part of our upcoming Beta 1.2 release.

This release will support richer project, client, and consultant queries, including questions like:

  • “How many healthcare projects have we completed in the last five years?”

  • “What projects have we completed at UCLA with ABC Engineering?”

In addition, Advanced Project Search will introduce support for project experience queries, allowing users to ask:

  • “Who worked on [Project X]?”

  • “Who has experience with [Project Type Y]?”

  • “Who’s worked on projects in [City Z]?”

We don’t have a release date yet, but development is underway. 

Future Versions: Additional Development Priorities for 2025

Synthesis LMS Query Support 

As we move toward the public beta of Synthesis LMS Version 1 at the end of the year, we’re also preparing to upgrade AI Search to include content from the LMS in its results. This will allow users to ask questions and discover relevant learning content directly through search — another step in connecting knowledge, people, and learning across your firm.

Advanced Query Support - Temporal Queries

We plan to support temporal queries like “what site tours did Ann lead this year?” or “who won research grants last year?” 

 

Known Limitations + Bugs

Hashtags, Community Home Pages, Contacts, Companies, Opportunities, Directory Views, and Library Views are Not Included in AI Search Yet

We don't have a timeline for making this change yet.

 

Synthesis Assumes Your Firm is the Author/Subject of all Documents

This is a known bug which just emerged recently. You might have external documents on your intranet (best practices, case studies, or other reference materials) which were written by and/or about another company. From time to time Synthesis may assume that external documents are about your firm and cite those external documents as sources. We don't have a timeline for fixing this issue yet, but your examples will be helpful in developing and testing a fix, so please rate them and provide feedback should you encounter this issue.


Finding All the Things vs. Finding the Best Things

We have initially designed Synthesis AI Search to return the best answer to a question, not necessarily a complete list of results. For most queries, this works well—people are typically looking for a specific piece of information, like:

  • Where can I find the latest proposal template?
  • What’s our PTO policy?
  • Who leads our sustainability initiatives?

However, some users expect search to return all relevant results, such as:

  • All posts from last year with a specific hashtag
  • All training videos on Revit

Right now, Synthesis AI Search is not designed for exhaustive, list-based queries like these. Over time, we’ll evaluate expanding search to better support queries that require returning all relevant content, such as office locations, training videos, or historical posts.

If there’s a specific "all the things" search that’s critical for your firm, let us know—we’re always looking to improve.

 

Office Locations Queries

We are aware of an issue affecting Synthesis AI Search's ability to accurately answer questions about office locations, such as "How many office locations do we have?" or "What are all of our office locations?" This issue stems from two main factors:

  1. A lack of structured, canonical data for office listings, which currently relies on inference from page hierarchy and navigation blocks.
  2. The existing focus on providing the best answer rather than all possible answers (see above), which impacts comprehensive query results.

We don't have a specific timeline for fixing this issue, but are committed to resolving it in the future.

 

Changelog

06.06.25   Added Advanced Employee Search to AI Search
04.04.25   Added Search Analytics to AI Search
02.14.25   Added Autoscroll to Page Sections to AI Search
01.23.25   Added Search Filters to AI Search
01.06.25   Added support for External Links to AI Search
12.23.24   Added support for Private Communities to AI Search
12.21.24   New/edited resources are now indexed immediately by AI Search
12.16.24   Synthesis AI Search Public Beta
12.14.24   Improved content weighting based on content type, freshness/recency, and best bets to  make sure that the most relevant content is used to generate AI Search Summaries.
12.04.24   Added Targeted Clickthroughs to Cited Sections of Videos to AI Search
11.27.24   Synthesis AI Search Private Beta Round 2
11.18.24   Added support for Documents to AI Search
11.18.24   Added Vector/Semantic Search capabilities to AI Search
11.05.24   Added Search Summary Ratings to AI Search
10.21.24   Synthesis AI Search Private Beta Round 1

Additional Questions

Please email us at to support@knowledge-architecture.com.