AlphaSense vs Tegus (2026): They're the Same Platform Now

If you came here to work out which of these two vendors to buy from, the answer is that the question no longer has an answer. AlphaSense acquired Tegus in 2024. They are one company, one platform, and one commercial relationship.

That is not a technicality to note and move past — it changes what you should actually be asking. The useful question is not which of the two to pick, but how the transcript library and the wider research environment behave differently inside the same subscription, and which one a given research task should send you to first. This piece answers that, and then covers what AlphaSense's in-house AI-moderated call capability signals for the rest of the category.

AlphaSense Acquired Tegus in 2024

The deal closed in 2024 at $930m, and it was not AlphaSense's first move into expert content — it had already acquired Stream in 2021 to seed a transcript library. Tegus was the larger and better-known of the two, and after the acquisition its archive was folded into the same research environment that indexes filings, earnings calls, broker research, and news.

So Tegus today is a product, not a firm. It is the brand for AlphaSense's searchable library of prior expert interviews, sold as part of the AlphaSense platform rather than as a separate expert-network contract. Whatever the packaging or tiering looks like on any given deal, it is one vendor relationship: there is no scenario in which the two are bidding against each other for your budget.

This matters practically, not just as trivia. Teams still carrying the pre-2024 mental model tend to make two mistakes. The first is running an evaluation as though they are choosing a supplier, which produces a decision that cannot be acted on. The second is the opposite error — assuming that because they are one platform, the distinction between the library and the wider environment has dissolved. It has not. The two surfaces answer different kinds of question, and knowing which is which is the part that actually affects your research day.

The Two Products Inside One Platform

What the Tegus library is. A compounding archive of expert call transcripts — conversations commissioned by subscribers over years, transcribed, indexed by company and topic, and searchable. You are reading interviews that already happened rather than commissioning new ones.

Where the library earns its place. Immediacy on covered names. A researcher with a question about a well-followed company can be reading practitioner views within minutes, without waiting on a calendar or a compliance clearance, and typically without a separate per-call charge on top of whatever the subscription already covers. Because the archive is text sitting inside a search environment, it behaves like every other document class on the platform: a passage can be pulled out, set beside a filing, and read in sequence. A category's operator commentary rarely sounds the same in one year's interviews as in the next, and the library is where that drift shows up. It is also where the specific texture of expert commentary lives: how a former channel manager actually describes a distribution shift, in their own words.

Where the library runs out. Every other document class on the platform exists because somebody was obliged to produce it — a filing gets filed, an earnings call gets held, an analyst publishes on a schedule. The transcript archive has no such generator behind it. It exists only where a subscriber once paid to ask, which leaves its shape uneven in a way the rest of the corpus is not: dense on large-cap and widely-followed names, sparse on private companies, unusual supply-chain nodes, emerging categories, smaller geographies. Fit is uneven for the same reason. Each interview was scoped to somebody else's brief, so you inherit their question set rather than writing your own, and the right expert on the right company may have been asked about something adjacent to your problem. Currency is the third gap. A filing carries a date that means something; a transcript's date tells you when the conversation happened but not whether its subject matter still holds, and a result list presents a stale interview and a current one with equal confidence.

What the wider AlphaSense platform is. A document-search environment covering filings, earnings transcripts, broker research, news, and company documents, with the expert content indexed alongside it rather than in a separate silo.

Where the wider platform is strongest. Breadth and triangulation. A question about a company's margin trajectory can be worked across the filings, the earnings call language, the sell-side view, and the expert commentary in one search context, and the platform is built so those sources sit next to each other rather than in separate tools. For the large share of research questions where the expert view is one input among several, this is the surface that does the work.

Where the wider platform shows friction. Breadth cuts both ways. A search across every document type returns a great deal that is not expert commentary, and on questions where a practitioner's operational view is the whole point, filtering down to it is an extra step. Documents also tell you what companies and analysts said publicly, which is a different thing from what an operator will say on a call — the two are complementary rather than substitutable, and the platform's integration can make it easy to forget that.

Choosing Between the Library and the Wider Platform

Since both surfaces come with the same subscription, this is a routing decision inside your workflow rather than a purchase decision.

Recurring coverage of a defined company universe leans on the library. If you return to the same set of names quarter after quarter, the archive is working in your favour: much of what you need has probably been captured by someone else covering the same companies, and every transcript you read compounds against the next question. Public-markets coverage, competitive monitoring, and continuous tracking programmes get the most out of this, because the hit rate on a well-covered universe is high enough that starting in the library is usually the fastest route.

Broad research across document types leans on the wider platform. When the question spans regulatory filings, disclosed financials, analyst commentary, and market context — a sector initiation, a thematic piece, an unfamiliar industry you are getting up to speed on — the expert transcripts are one input among several, and the environment that holds all of them together is the right starting surface.

The honest caveat is that on genuinely novel questions neither surface has an answer waiting. Private-company diligence, a category that did not exist three years ago, something that happened last month: the archive has not covered it because nobody has asked, and the document corpus has not covered it because it was never disclosed. That is the point where a new conversation with a practitioner is the only route, and it is the same point at which most research teams reach for an expert network.

What AlphaSense's AI-Moderated Build Means for Expert Networks

AlphaSense did not stop at distribution. Alongside the acquired archive, it built AI-moderated call capability in-house — an AI agent conducts the interview, and the output is transcribed, summarised, and indexed alongside every other document type on the platform.

If you operate an expert network, that build is the part of this story to sit with — not as a threat, but as a piece of information you did not have to pay for. A firm that had just spent $930m acquiring a stock of interviews someone else had already conducted turned round and invested in producing new ones without a human moderator. Read the two decisions next to each other and you get a fairly direct statement about where the platform concluded the durable cost sits: not in owning transcripts, but in making them.

It also moves the baseline for what a research team assumes is normal. Someone who reads expert content on the platform gets it structured, searchable, and already written up as the default condition — that is simply what expert material looks like there. Carry that habit into a network relationship and an hour of raw audio with a page of handwritten notes stops reading as the deliverable and starts reading as the raw material for one. Nobody raises it, and the expectation is set anyway.

Notice, though, what the platform's own limitation tells you. Its archive is deep where subscriber demand has been heaviest and thin at the edges — and the edges are where boutique networks live. Proprietary access to practitioners in a niche nobody else has interviewed is the one thing no shared library can manufacture, because a library can only redistribute conversations that already happened. That asset is not under pressure from any of this.

What AI-moderated calls change is the cost of turning that access into something with the properties of a document. The expert joins by phone or web at a time they choose, the agent runs your question set and follows up on the answers, and the call resolves into a transcript, a structured summary, and a compliance record without anyone writing them afterwards. Two costs leave the per-call figure: the calendar negotiation and the moderator's hour. What arrives in their place is uniformity — the same question discipline applied to every respondent, which is what makes a set of interviews comparable rather than merely numerous.

None of that puts a network out of a job, and it is worth being precise about which layer actually moves. A library is distribution. An agent is production. Neither one is supply. Somebody still has to know which practitioner is worth an hour, be able to reach them, and carry the compliance weight of the introduction — and that is the work a network sells.

Most networks start inside their own operation rather than in front of clients, because vetting is where the cost lands first and a misstep is cheapest. Screening calls run by an agent take the coordinate-then-screen sequence out of the path at a fraction of the per-expert cost, and throughput stops being a function of headcount: a hundred candidates a month and a thousand cost roughly the same to process, with a human still reading the output before anyone reaches a client. What that produces as a side effect is an archive of your own, on a coverage area no shared library indexes. Pointing the same machinery outward is the second step — the repeat formats clients already buy in volume become something you can price as a product line, while the calls that warrant a broker still get one.

None of this is an infrastructure project. AlphaSense reached its capability by acquisition and engineering; that is one route, and it is not the only one available. The agent is configured against the client use cases you already serve, connected to the expert scheduling workflow you already run, and the calls go out under your own brand. Pricing starts at $499/mo — an operating line, not a build. The InsightAgent for Expert Networks overview covers the full capability set and how it slots into existing network operations.

If you want to judge the quality of an AI-moderated call, the fastest way is to be on one — the demo is a call with the agent, not a slide deck:

Talk to the agent — live

Or start a free trial and run your first calls on your own question set:

Get started free