GLG launches AI-built client workspace, moves expert research onto an agent layer

GLG has rebuilt its client-facing product around a set of AI tools that sit on top of its expert sourcing, survey, and content library functions, according to a press release issued on PRNewswire. The release frames the changes as a single workspace covering question-to-expert routing, call moderation, and post-call synthesis.

The move puts GLG into the same product territory AlphaSense and Tegus have been pushing into since the 2024 merger: a research surface that bundles transcripts, expert calls, and AI summarization rather than selling each as a standalone line item.

What the product covers

Three components are named in the release.

The Strategic Research Agent takes a client question and returns expert recommendations, relevant surveys, and Expert Content Library matches. The release specifies that every input also routes to GLG's white-glove service teams, which suggests the agent is a triage layer rather than a replacement for human sourcing.

The second piece is an AI moderator that can run real-time conversations with experts and follow up on answers, positioned alongside, not instead of, 1

The third is a synthesis engine that ingests calls, GLG content, and client-uploaded materials (video and audio included) and produces themed summaries with citations back to timestamps and source quotes.

Context: where the category sits

Expert networks have spent the past 18 months trying to move up the stack from "we source the expert" to "we run the workflow." AlphaSense's acquisition of Tegus in mid-2024 was the most visible bet on that thesis, pairing a transcript library with expert call infrastructure. Third Bridge has been pushing its Forum product in a similar direction.

GLG's previous positioning leaned heavily on the size and quality of its expert pool. The release signals a shift: the differentiation now sits in the tooling around the expert, not just the expert itself.

That is consistent with what buyers in the moderator and research analyst seat tend to complain about. The bottleneck on a research project is rarely "can I find an expert." It is everything around the call: scheduling, briefing, transcribing, tagging, and producing something an investment committee will accept. An agent that routes questions and a synthesis tool that cites back to source material both target that bottleneck directly.

What the release doesn't say

A few omissions worth flagging.

There is no pricing detail, no client count, and no disclosed adoption metric. The release does not name the underlying models, the vendor stack, or whether the synthesis engine processes client uploads on-prem or in a managed cloud. For buy-side compliance teams, the on-prem question is the one that determines whether the tool gets cleared.

There is also no published benchmark on transcription accuracy or latency for the AI moderator. Real-time probing only works if the model can parse the expert's response cleanly enough to ask a useful follow-up. Anyone who has sat through an automated call with a heavily-accented or technical expert knows where that breaks.

The release references a separate GLG report on agentic AI drawing on 110 implementations, though the PRNewswire summary cuts off before describing the report's findings.

What to watch next: pricing disclosure, which expert networks white-label GLG's synthesis stack (the typical pattern when a category leader ships infrastructure), and whether the AI moderator gets cleared by buy-side compliance for use on call recordings of named experts.