Powering and Protecting Our Clients in the Age of AI

AI-powered tools allow research teams to analyze voluminous inputs faster than ever. The resulting flow of information threatens to overwhelm compliance teams, many of which still rely on review models and workflows that simply do not have the capacity or scale to keep up. Even where the underlying risks resemble those from a pre-AI world, we consistently hear from clients that technology has fundamentally changed the speed and scale at which those risks propagate.

Scaling our tools for a rapidly evolving research landscape

Many research teams that want to unlock the benefits of new AI tools are asking for more call transcripts, both from traditional one-on-one expert calls and AI-moderated engagements. For GLG clients in financial services and beyond that are navigating the evolving regulatory landscape, the ability to search, reference, and synthesize expert conversations at scale is rapidly becoming a central component of their process. For compliance teams, the decision to move towards blanket transcription hinges on crucial questions: How do you build workflows at scale that can reasonably detect and assess information that warrants compliance review? What happens when a transcript gets ingested into a data lake, synthesized by an AI tool, or surfaced through an MCP integration, but your compliance team has a question about its source? In an AI-accelerated environment, one problematic fragment of content can be disseminated downstream in seconds and may be difficult, if not impossible, to extract, depending on system design.

The tension between the benefits and risks associated with expanded transcription has been a theme in countless conversations with our clients and their compliance officers. Many investment professionals urgently want broader access to transcripts to scale up their research and make decisions faster, but compliance teams need to maintain a reasonable level of control and review at scale, balancing finite time and resources.

At GLG, our response has been to listen to our clients and build the controls they need directly into our platform, with functionality that dovetails with their evolving processes around the transcripts they increasingly rely on to power their research.

With our latest suite of transcript compliance features, clients can facilitate their compliance program initiatives and policies concerning transcripts at no additional cost:

• AI Flagging automatically scans transcripts for potential compliance concerns in the categories a firm chooses, so compliance teams can focus their attention on what needs to be reviewed rather than reading every word of every call.  

• Redaction Options let compliance officers remove specific content before it reaches end users, with a clear indication in the transcript where a redaction has occurred. These redactions apply from the transcript through the audio recordings, and downstream in AI summaries, so the redacted passages are never released into a firm’s data.  

The new capabilities build on existing ones in GLG’s platform, such as transcript holds, which let firms gate access to any transcript until their compliance team has reviewed and approved it; summary-only access, which allows for only AI-generated summaries to release to end users; and flexible retention settings, which set transcripts, summaries, and/or recordings, respectively, to expire within customizable time frames.
‍

These features are already making GLG research and content more accessible to our clients where and how they prefer to work. Our MCP connector, which lets users query and analyze their call transcripts and GLG’s Expert Content Library directly inside Claude and other AI assistants, can flow through the same compliance settings. Before a transcript surfaces in your firm’s AI-powered workspace, it can be cleared through your firm’s compliance approval settings.
‍

Enabling transcripts also empowers clients to take full advantage of our AI-Moderated Calls offering, which conducts structured expert interviews autonomously, with GLG’s industry-leading compliance framework embedded into its design. Running calls around the clock, the capability can help research teams generate dozens of insight-rich transcripts in just a couple of days, and we built it to work seamlessly within the most stringent compliance workflows, with the added option of permitting compliance teams to complete their review before the transcript is accessible by any member of their research team.

Differentiated trust at every stage of the workflow

Beyond our focus on facilitating the work of our clients and their compliance teams, we are also helping them solve for a more fundamental challenge as they design AI-powered research tools and workflows: “garbage in, garbage out.” As new data offerings and untested niche vendors appear in the market almost daily, firms should be careful not to conflate effort put into building incredibly powerful AI systems with effort to assess the quality and veracity of the inputs feeding them. Facts and real human insights matter as much as ever in research...and perhaps even more than ever as clients seek to distinguish differentiated, supported conclusions from AI slop. Plugging untrustworthy inputs into research workflows will eliminate ROI anticipated from the adoption of the AI tools ingesting those inputs, no matter how well designed or skillfully deployed. (Bad fuel causes sputtering and bad performance, whether you are driving a Corolla or a Ferrari.)

These risks multiply for firms building proprietary libraries, where the quality of any single transcript will impact research far beyond the discrete project for which it was initially commissioned. Vendors that do not invest adequately in screening processes, that do not effectively leverage vetted third-party data and proprietary tools to identify risks, and that cannot rely on large historical and up-to-date data sets concerning the experts they deploy (of which GLG has more than any peer), bring with them outsized risks by introducing inputs from unvetted sources into melded AI-powered research outputs. The risks such vendors present are hard to justify given how integral an expert’s identity and experience is to the value of their input.

While some players in the market anonymize the identities of the “experts” on whose assertions their transcripts hinge, GLG has always maintained that an expert’s identity is inextricable from their insight – that the details of their identity and experience lend their perspective the essential quality that makes it so valuable. Vendors that seek to hide the identities of the sources of the information that they sell may argue that AI-powered tooling can be designed to disregard outlying data points from unverified or untrustworthy transcripts and data sets. This stance understates important risks. It can be hard or impossible to say which data points are outliers on research of topics or unique angles that are not extensively covered by more trustworthy inputs. Further, even on topics on which there are numerous inputs, research diligence requires additional assessment of whether these problematic sources impacted the conclusions of the AI tooling. Such risks should cause business leaders to seriously question why they would introduce anonymized or otherwise undependable inputs and untested vendors in the first place as they evaluate solutions.

At GLG, we will continue to raise the standard for trust and safety in the industry that we invented and turn our focus towards building the next layers of protection, while our industry-leading screening and matching systems continue to guard the front door of our clients’ AI-powered research environments.

Prioritizing trust and safety in the future of primary research

The research landscape will only grow more complex from here. AI will keep accelerating the pace at which data is collected and moves through analytical systems, empowering researchers further, putting more demands on compliance teams, and requiring unwavering diligence around the quality and traceability of sources.

At GLG, where we hang our hats on compliance, trust, and safety, and strive to earn our clients’ confidence every day, we are excited to meet every challenge AI presents to professionals in the primary research space while helping them unlock its most promising opportunities. We remain focused on building the best infrastructure in the industry alongside our clients to keep them ahead – not just to mitigate today’s biggest risks, but to give them confidence in the face of whatever comes next as they leverage the irreplaceable power of human insight.

‍