How Expert Networks Are Evolving: From Brokers to AI-Powered Platforms

The expert network industry has operated on roughly the same model for two decades. A client firm — a hedge fund, a private equity shop, a consulting practice — submits a research request. A broker searches an internal database, identifies a handful of candidate experts, coordinates schedules across time zones, and connects the parties for a one-hour call. The client pays somewhere between $500 and $1,500 for that call. The broker keeps a margin. The expert receives a consulting fee. The cycle repeats.

This model has been durable because it worked. The major networks built real defensibility: large proprietary expert databases, compliance infrastructure, institutional relationships with hundreds of client firms. For the better part of twenty years, nobody had a fundamentally better answer to the same problem.

That is now changing. The forces reshaping the industry are not subtle, and the direction of travel is not ambiguous. Three structural pressures — compliance, cost, and scale — are compressing the broker model from multiple directions simultaneously. And the platform layer emerging to address those pressures looks very different from what it is replacing.

How Traditional Expert Networks Work

The traditional broker model bundles five distinct services into a single per-call fee: sourcing, matching, scheduling, compliance screening, and call moderation or synthesis. The economics of this bundle have been the industry's core value proposition and, increasingly, its central vulnerability.

On the supply side, expert networks maintain curated databases ranging from several hundred thousand to over a million registered consultants. Building and maintaining that supply is expensive. Experts must be recruited, credentialed, kept engaged, and cycled through compliance checks on an ongoing basis. The largest networks — GLG, Gerson Lehrman, Guidepoint, AlphaSense/Tegus — have invested hundreds of millions of dollars in this infrastructure over their histories.

On the demand side, client firms pay a premium for access to that supply plus the coordination overhead. A fully-loaded expert call through a major network typically runs $800 to $1,500, depending on the expert's seniority, the network, and the complexity of the request. Annual contracts with the large networks can reach into the hundreds of thousands of dollars for heavy users. Research teams at large asset managers routinely spend $1 million or more per year on expert network access.

The margin structure is correspondingly generous. Estimates of broker margins on individual calls range from 40% to 60%, though exact figures vary by firm and contract structure. For decades, that margin was defensible because the alternatives — building your own expert relationships from scratch, or paying full consulting rates for equivalent access — were more expensive.

The math is beginning to shift.

What Is Forcing the Industry to Change

Compliance Pressure

Regulatory enforcement in the expert network space has intensified materially since 2020. The SEC and DOJ have made clear that informal information flows between public company employees and investment professionals are a priority area. Several expert network-related enforcement actions — most prominently the wave of cases following the Galleon Group prosecution, but extending well into the current decade — have focused compliance attention across the industry.

The practical consequence for network operators is that self-certification and spot-checking are no longer adequate compliance postures. Clients want documented evidence that every call was screened, that experts were validated against relevant trading restriction lists, and that call content was monitored for material non-public information. The broker's historical approach — relying on human judgment, periodic training, and after-the-fact review — does not produce that evidence reliably.

Automated compliance creates genuine audit trails. Real-time MNPI detection during live calls, pre-call conflict screening against expert employment history and financial disclosures, and complete session-level documentation are the compliance baseline that sophisticated clients now expect. Producing this infrastructure through human processes is cost-prohibitive. Producing it through software is scalable.

For network operators, compliance has shifted from a cost center to a product requirement. The platforms that build compliance in as core infrastructure — not as a manual layer operated by humans — have a structural cost advantage over those that do not.

Cost Compression

Client-side pressure on per-call economics has been building for years. The immediate trigger in most cases is budget: research teams face the same flat or declining research budgets while their mandate to generate differentiated insights remains unchanged.

The pressure is also structural. Subscription-based pricing models — pioneered by Tegus and widely adopted across the industry — have reset client expectations about what expert access costs at scale. When clients can pay a flat annual fee and access a library of pre-recorded expert calls, the economics of paying $1,000 per custom call become harder to justify for standardized research use cases.

The response from traditional networks has been predictable: expand service offerings, add technology layers, emphasize quality of matching over price. These are rational defensive moves. They have not resolved the underlying pressure. Research teams comparing cost per insight from a high-volume AI-assisted workflow against the cost of equivalent broker-mediated calls are finding the numbers increasingly difficult to ignore.

Scale Demands

The third force is perhaps the most fundamental. Investment research is a scale problem.

A portfolio manager building a thesis on regional banking dynamics might want to speak with former branch managers, regional credit officers, fintech competitors, and enterprise software vendors across a dozen different geographic markets. Doing that thoroughly — ten to twenty calls with genuinely relevant experts — through a traditional broker process takes weeks and costs tens of thousands of dollars. Most research teams do not have the budget or the time.

The result is that primary research is rationed. Analysts make fewer calls than they should. Due diligence is shallow where it should be deep. Investment theses rest on fewer data points than the evidence would support.

The scope of what is actually feasible expands dramatically when per-call costs drop and scheduling friction disappears. Teams that have shifted toward AI-assisted expert research consistently report the same pattern: they are not just doing the same research faster, they are doing research they would not have attempted before because the cost-benefit calculation did not work at traditional prices.

The AI-Moderated Model

The AI-moderated approach to expert interviews replaces the broker's five bundled functions with a platform architecture where most of the coordination, compliance, and synthesis work is automated, and the human component — the expert's knowledge — is preserved and made more accessible.

AI agents conduct the interviews themselves, following structured question protocols that the client team designs in advance. The expert dials in or joins through a web interface. The AI agent introduces the research objective, asks questions, follows up on interesting threads, and maintains a natural conversational flow throughout. No human moderator is required.

Scheduling is handled end-to-end by automated workflows. The expert receives a link, selects a time, and confirms. There is no back-and-forth email coordination. Experts in different time zones can schedule and complete calls at times that work for them, including evenings and weekends, without anyone on the client side needing to be present.

Compliance screening runs before the call starts. The expert's recent employment history, known board positions, and public financial disclosures are cross-referenced against relevant restriction lists. During the call, real-time monitoring flags language that may indicate material non-public information or other compliance concerns. After the call, a complete documented record — full transcript, compliance flags, session metadata — is automatically preserved.

Transcription and synthesis happen immediately. Within minutes of a call ending, a structured summary with key takeaways is available for review. Quotes are tagged by topic. Moments that map to the original research questions are highlighted. The analyst does not wait until they have time to write up notes — the output is ready before the next call starts.

What is not automated is the expertise itself. The value in these conversations is the expert's knowledge of their domain: what they have observed, what they believe, what they have done. AI-moderated platforms are designed to extract that knowledge more efficiently and at greater scale. They are not trying to replace it.

What This Means for Different Stakeholders

Investors and Research Teams

For research teams, the headline benefit is throughput. More expert conversations per analyst, at lower cost per conversation, with faster time to actionable output. Teams that previously budgeted for fifteen to twenty expert calls per quarter are finding that the same budget covers three to four times as many calls through AI-native workflows.

The subtler benefit is consistency. A broker-mediated call depends heavily on the quality of the specific expert the broker found and how well the conversation was moderated. An AI-moderated call follows a standardized protocol, ensuring that every expert is asked the same core questions, that follow-up probes are consistent, and that the resulting outputs are directly comparable. Analysts who want to triangulate across multiple expert perspectives are working with apples-to-apples data rather than a set of loosely related conversations.

Expert network software comparisons for 2026 increasingly reflect this shift: platforms are evaluated not just on network size but on workflow automation, compliance infrastructure, and output quality.

Experts

The expert experience in the traditional model has friction points that the AI-moderated approach largely eliminates. Scheduling coordination is the most obvious: responding to availability requests, negotiating times across time zones, confirming logistics — this overhead is real and compounds across a busy expert's week.

More significantly, AI-moderated platforms remove the requirement for experts to be available during business hours in the client's time zone. An expert in Singapore can complete a call for a New York research team at a time that works for both parties, without anyone coordinating the handoff. Experts who might be good candidates for calls but are habitually unavailable during traditional business hours become accessible.

Language is another dimension. Expert networks serving international markets often struggle with calls where the expert's first language is not English and nuanced knowledge gets lost in translation. AI-moderated platforms can conduct interviews natively in the expert's preferred language, then deliver translated and synthesized output to the client.

Compliance Teams

For compliance officers at client firms and expert network operators alike, the shift to AI-moderated platforms changes the risk calculus significantly.

The primary compliance risk in traditional broker-mediated calls is inconsistency. Human screeners apply different levels of scrutiny to different calls. Post-call documentation is sparse or inconsistent. When a regulator requests evidence of compliance monitoring for a specific call from eighteen months ago, the answer is often some combination of memory, notes, and hope.

Automated platforms produce complete documentation as a byproduct of operation. Every call has a transcript. Every compliance check has a timestamped record. Every MNPI flag has a corresponding clip of the call where the language appeared. This is not just a defensive benefit — compliance teams are increasingly using this infrastructure proactively, reviewing flagged moments across batches of calls to identify patterns rather than reacting to individual incidents.

Expert Network Operators

For the expert network operators themselves, the transition is existential in some respects and an opportunity in others. Broker-only networks — operations that depend on human intermediaries for every function in the workflow — face cost structures that cannot compete with platform-based alternatives at scale. The labor required to handle sourcing, matching, scheduling, compliance, and synthesis manually scales roughly linearly with call volume. Technology-based operations scale sub-linearly, which is the core of the competitive threat.

But the operators who move toward technology-enabled models are finding that the economics improve materially. When the coordination and compliance overhead is automated, the expert relationships and proprietary network supply that traditional operators have spent years building become more valuable, not less. The constraint that prevented scaling that supply — the human labor required to match and coordinate every call — is removed.

InsightAgent's platform for expert networks is built specifically for this transition: operators bring their existing expert relationships, and the platform handles the automation layer that makes those relationships scalable.

Where the Industry Is Heading

The near-term trajectory for the expert network industry points in several directions simultaneously.

Consolidation of broker-only networks. Operators that have not invested in technology infrastructure will face compressing margins and increasing difficulty competing on price for the high-volume, structured research workflows that AI-native platforms serve efficiently. Some will be acquired. Some will exit. The broker-only model will survive in narrow niches — highly relationship-dependent sourcing, sensitive and confidential expert conversations, markets where technology penetration is low — but it will not be the industry's dominant form.

Hybrid models as the mainstream. The majority of the industry will not move to purely automated operations. The realistic steady state for most expert network platforms is a hybrid architecture: AI handling coordination, compliance, transcription, and synthesis, with human expertise — both the experts themselves and the network staff that curates and manages expert supply — providing the knowledge layer that automation cannot replicate. This is consistent with how expert networks explained the direction of the industry well before automation pressure became acute.

AI-native entrants competing for volume. New entrants building on AI-native architectures from the start — no legacy broker infrastructure, no institutional overhead, no path dependency on existing workflows — will compete aggressively for the high-volume, structured research use cases: channel checks, market sizing interviews, customer reference calls, regulatory landscape mapping. These are the workflows where the broker model's relationship advantages matter least and where cost and speed are the primary decision factors.

Compliance as differentiator. As regulatory scrutiny of the expert network space continues to intensify, compliance infrastructure will shift from table stakes to genuine competitive differentiation. Clients with sophisticated compliance requirements — large asset managers, regulated institutions — will actively prefer platforms that can demonstrate systematic, auditable compliance processes over platforms that describe compliance as a general commitment. The platforms that invest in compliance infrastructure now will have a durable advantage as the regulatory environment tightens.

Commoditization of transcription and synthesis. The quality gap between different platforms' transcription and synthesis capabilities will narrow as the underlying AI models improve. Differentiation will shift toward the quality of expert supply, the depth of compliance infrastructure, and the integration of expert interview outputs with broader research workflows. The raw call transcript, on its own, will be a commodity. What firms do with it will be the product.

The industry that emerges from this transition will look different from the one that existed five years ago. The core transaction — a knowledgeable expert sharing what they know with a researcher who needs to know it — does not change. The infrastructure around that transaction, and the economics it produces, are being rebuilt from the ground up.

---

InsightAgent is an AI-moderated expert interview platform built for research teams and expert network operators. Learn more about how we're changing how expert calls work.