DABYTE · ABOUT
What this is, who runs it, and why the data is shaped this way
Answer. DABYTE is a data desk. We measure how often AI assistants name SaaS & AI Tools brands when buyers ask category questions, we index what the industry press reports, and we write our own analysis of events where we have a measurement to attach to them. Everything we publish is also available as data you can call rather than scrape — JSON, CSV, markdown and an MCP descriptor, free and without a key, under CC BY 4.0. DABYTE is published by VECTORY, and it is one of 2 properties we run.
Why the data is shaped this way
Most industry data reaches you as a web page: a number you wanted, wrapped in navigation, banners and markup. Anyone building on it writes a parser, maintains it, and burns compute turning a document back into the structure it came from. We publish the structure directly, so that step does not exist.
Concretely: every ranking, every brand profile and every analysis piece exists in parallel as JSON at a stable URL, as CSV, and as a markdown mirror. Field names are documented and do not change without notice. There is no key, no signup, no rate limit to negotiate, and the licence is CC BY 4.0 — attribution is the only condition. If you are building a research agent, a dashboard or a newsletter, you can read our numbers in one request instead of maintaining a scraper.
We also publish an MCP descriptor at
/.well-known/mcp.json listing tools that resolve to those same endpoints,
so a client that speaks MCP can discover what is available without being told.
What we do not do: we do not republish anyone else's articles, we do not paywall the data, and we do not treat the link to a source as optional. Where an event comes from someone else's reporting, that reporting is named and linked, and the reader is expected to go read it.
Every surface, in one place
| Surface | What it holds |
|---|---|
/api/aiv.json |
The full index: every brand, share of answer, per-engine breakdown, rank, commercial intent, quadrant, and the measurement date. |
/aiv.csv |
The same ranking as a spreadsheet, for people who would rather not write code. |
/api/articles.json |
Our analysis pieces with the measurement each one rests on, the sources that reported the event, and how long we consider it decision-relevant. |
/api/topics.json |
What the monitored press reported, deduplicated across outlets, with brands from the index tagged. |
/index.md |
A markdown mirror of this site's main page. Every brand profile has one too. |
/.well-known/mcp.json |
MCP tool descriptor pointing at the endpoints above. |
/llms.txt |
A plain-text summary of what lives here and where. |
The network
We run 2 properties, each covering one industry with the same method and the same open-data commitment. They share a publisher, a measurement pipeline and a codebase, and we say so plainly rather than presenting them as unrelated outlets.
| Property | Industry | What it covers |
|---|---|---|
| DABYTE (you are here) | Tech & SaaS | AI visibility for 20 SaaS and AI-tooling brands, plus product events read against that measurement. |
| DABLOCK | Crypto & Web3 | AI visibility for 24 exchanges, wallets, protocols and analytics providers, plus incidents, releases and governance events. |
Editorial policy and disclosure
Who publishes this. DABYTE is owned and published by
VECTORY, a company that works on
AI visibility. That is a relevant fact about us and it is stated here, in the footer of
every page, in /humans.txt and in our structured data.
Nothing here is paid. No brand pays to appear in the index, to move within it, or to be written about. There is no advertising on these pages and no affiliate links. If that ever changes, paid placement will be labelled as such and will remain outside the measurement: a ranking you can buy into is not a measurement.
What we measure and what we do not. The index records how often an engine names a brand across a fixed panel of prompts. It is not a quality rating, not a recommendation, and not a statement about any product. A low score means the brand is rarely named in AI answers — nothing more.
Corrections. If something here is wrong, tell us and we will fix it and say what changed. Reach us through VECTORY. We would rather correct a number than defend it.
How the analysis is produced. Events are collected automatically from named sources. A piece is written only when we have our own measurement to attach to the event, drafted with language-model assistance against a fixed template, and checked by an automated gate before publication — sources must be independent of each other, the subject must match the source, and the measurement must exist. Pieces that fail the gate are not published.