PYRAMYD vs AlphaSense
AlphaSense reads the market faster. PYRAMYD tells you its next move.
AlphaSense is AI search over premium documents — filings, transcripts, expert calls. PYRAMYD is a typed graph of the enterprise software market with a game-theory engine on top: predictions, prescriptions, and an exploitability score for every competitor move.
Side-by-side
Different substrates, different questions.
AlphaSense answers what has been said. PYRAMYD answers what happens next. Many teams that can afford both run both — the honest comparison is about which question your team is actually paid to answer.
| Dimension | AlphaSense | PYRAMYD |
|---|---|---|
| What the substrate is | A document corpus: SEC filings, earnings transcripts, broker research, news, and a proprietary expert-call library — with AI search and summarization across it. | A typed knowledge graph: 250K+ software products, 2,546 categories, 2.4M+ reviews, and the relationships between them — entities you traverse, not documents you search. |
| Coverage shape | Breadth across every public company and sector — finance-grade coverage that goes far beyond software. | Depth on one vertical: enterprise software. Every vendor, product, feature, and category in the market, including the private companies filings never reach. |
| Predictions & prescriptions | Search and summarization are retrospective: the answer is a synthesis of documents that already exist. No forward model of competitor behavior. | Two-sided game-theory engine: a probability over each competitor move, an exploitability score with blind-spot amplification, and counter-moves ranked by expected value — runnable live at /game-theory. |
| Question it answers best | "What has been said about X?" — across management, analysts, and experts, faster than any human reading program. | "What does X do next, and what should we do about it?" — with the evidence chain attached. |
| Citation model | Answers link back to source documents and passages within the licensed corpus. | Per-claim citation chain to the graph node and source URL — retrieval timestamp, model, quality, confidence — Article 50 ready. |
| Competitive deliverables | Research workflow: saved searches, alerts, dashboards, document monitors. Battlecards and win/loss live in other tools. | The deliverables are the product: live battlecards (from $90, no account needed), competitive matrices, win/loss, RFX responses, analyst-placement forecasts. |
| Analyst-report posture | Licensed broker and analyst research is searchable inside the platform where your firm holds the entitlements. | You upload the analyst reports you are licensed for; placements are read into the graph across six frameworks and projected forward to the next edition. |
| Pricing shape | Enterprise contracts, sales-led; seat pricing widely reported in the tens of thousands per year. No self-serve tier. | Free Foundation, hubs from $150/seat/mo, and a $90 single battlecard you can buy as a guest — the cheapest way to evaluate either platform. |
When to pick each
Reading program vs forward model.
Pick AlphaSense when
- Your coverage universe is broader than software — you need finance-grade breadth across sectors, filings, and expert calls.
- The job is research synthesis: what was said, by whom, across thousands of licensed documents.
- Your firm already holds the broker-research entitlements AlphaSense unlocks.
Pick PYRAMYD when
- Your market is enterprise software and you need the private companies, products, and features filings never mention.
- You want a forward model — what a competitor does next, what it is worth, and the counter — not a faster reading program.
- The output you need is a deliverable: a battlecard, a matrix, an RFX response, a placement forecast — not a search result.
- You want to start at $90 rather than an enterprise contract.
Bottom line
AlphaSense compresses reading. PYRAMYD compresses deciding.
If the question is "what was said," AlphaSense's corpus is the strongest in the market. If the question is "what do they do next, and what do we do about it," a document index cannot answer it — a typed graph with a game-theory engine can, and shows its work.
Bring a competitor. Watch the engine price their next move.
30 minutes. We'll run the two-sided synthesis on a competitor you actually face — the posterior, the exploitability, the ranked counter-moves — or try the public engine yourself first.