ApexMake vs GLG

A traditional expert network is a logistics business. Someone reads your brief, searches a database, screens candidates for relevance and conflicts, agrees an hourly rate, finds an hour that suits three calendars, and bills the call. GLG has done this at greater scale and for longer than anyone else. The 19 offices exist so that chain can run in every timezone.

All of that logistics is also where the delay lives. Even when it runs well, a first call is hours away at best and days away often enough. Fine when you are scoping a sector. Much less fine when the investment committee meets Thursday, the question landed Wednesday afternoon, and the answer decides whether you flag a risk.

ApexMake moves the interview earlier. Rather than interviewing an expert after you ask, a proprietary AI agent has already run a deep, structured interview with them, before the question existed. Querying that is a database read, not a calendar negotiation. Median time to answer is around six seconds.

GLG has not stood still. myGLG runs a real AI workflow now: a conversational Ask step to build the project, an ML-assisted Match step, a Gather step offering AI-moderated interviews in ten languages, and a Synthesize step that pulls themes across transcripts. There is even an MCP connector so research lands in Bloomberg Terminal.

For existing GLG clients that is a decent product, and it does compress the write-up at the end. But look at where the AI sits. It wraps the sourcing and it wraps the synthesis. The scheduled interaction is still the thing being bought, and it still has to be arranged.

ApexMake flips the stack. The AI is not a layer over a booking business, it is how the network gets built in the first place. Sourcing, screening and a structured deep interview all happen up front, and what the client touches is the resulting knowledge base. Same word on both homepages, very different product underneath.

Database size is GLG's headline advantage and it is a real one. If you need a former regional distributor of agricultural equipment in Brazil, a network of 1.2 million people will produce one and a network of a few thousand will not.

For technical questions, though, breadth and depth pull against each other. The number of people who can speak credibly about tokens-per-joule on custom silicon at production latency, or explain why a particular inference vendor lost a contract, is small. Finding them inside a 1.2 million-person database is a search problem before it is an access problem.

ApexMake bets the other way. A narrower network, screened hard for first-hand senior operating experience, with real density in AI infrastructure, inference, training and the software and infra products deal teams keep underwriting. You give up “we have someone for everything” and get back “the person we have actually ran it”.

Compliance is the part of expert networks regulators have spent a decade watching. Over the past ten years the SEC and DOJ have brought a long run of enforcement actions connected to expert network calls. Every serious provider, GLG included, has responded with expert training, conflict screening, contractual MNPI prohibitions and chaperoning on request.

Where the two differ is in what gets recorded. A human call produces a compliance process wrapped around it. An ApexMake query produces a machine-readable record of the interaction itself: query received and screened, experts matched and conflict-checked, answer delivered with citations, follow-up screened again, each one timestamped.

For a compliance officer that is the gap between attesting that a policy exists and being able to reconstruct exactly what was asked, who answered and what was said. No human call can offer the second thing, because the substance of a call is only as auditable as the notes somebody chose to take.

Per-interaction pricing does something quiet to how teams research. It makes every question expensive enough that analysts start censoring themselves. With calls benchmarked between $500 and $1,500 an hour, a team learns fast to save its questions for the ones that clearly justify a booking. The small clarifying question, the one that would have caught the problem, never gets asked.

A subscription to a queryable knowledge base costs roughly nothing per additional question. That changes what gets asked, not only how fast it comes back. Teams end up interrogating a thesis rather than sampling it.

It is also why both models sit comfortably in the same firm. ApexMake soaks up the high-volume, fast-turnaround questions, and covers the calls inside its sectors too: the expert is already vetted, their previous answers are on file, and booking takes a couple of clicks. GLG stays the right tool for the conversations that fall outside that coverage.