01 / Why teams choose Voidly

Three buyers, one data layer.

AI labs

Know where your models are accessible.

ChatGPT and Gemini are blocked in 13+ and 17 countries respectively. Anthropic's Claude has different patterns. Self-monitor accessibility from inside your stack via MCP, custom feeds, or signed webhooks — before customer support tickets tell you.

News orgs

Cite incidents. Embed widgets. Fact-check claims.

Every incident has a stable URL like IR-2026-0142, BibTeX/RIS export, and an evidence permalink. Embed live widgets, get RSS feeds for your beat country, and verify censorship claims through a dedicated endpoint.

Threat intel

Network-level signals for any domain in 130 countries.

TCP-RESET, DNS poisoning, blockpage, TLS reset, header manipulation, middlebox detection. Pulled from OONI + CensoredPlanet + IODA + 37+ Voidly probes. CI integration via our GitHub Action — fail your build if a target country starts blocking your assets.

Frontier AI lab? See the dedicated pitch at /ai-labs — accessibility intelligence built for Anthropic, OpenAI, Google DeepMind, Meta AI, Mistral, xAI, DeepSeek.

02 / Capabilities

Free tier vs Enterprise tier.

Honest mapping. The free tier is genuinely useful — most public research will live there. Enterprise unlocks scale, SLA, and bespoke deployments.

Feature	Free	Enterprise
Rate limits	1,000 req/min	Unmetered, fair use
Endpoints	All public	All public + custom feeds
SLA	Best effort	99.9% uptime, p95 < 500ms target
Custom probe deployment	—	Yes, with notice
Dedicated support	Community	CSM + Slack channel
Access to the data itself	Everything, CC BY 4.0, published to everyone at the same moment	Identical — there is no embargo tier and no pre-publish window. We do not sell the dataset.
Webhook delivery	Public alerts API	Signed HMAC, retry, custom filters
Managed filtered feeds	Self-serve — filter the free API yourself (/data/incidents?country=…)	We build and maintain the filter, with delivery SLA
SOC2 documentation	—	In flight (be honest — see FAQ)
Onboarding	Self-serve docs	Discovery call → 2-week pilot → rollout
Pricing	Free	Tailored — schedule a call

03 / Customers

Open-source flywheel, vendor-neutral by design.

honest position

Distribution and data depth are the moat.

The free public tier and open data layer come first. Distribution and dataset depth are the moat. We are evaluating select partners in AI labs, fintechs, and threat intel for paid pilots — talk to us if that fits.

What we will not do: list logos of companies that do not pay us, claim Fortune 500 deployments, or invent case studies. Everything we publish is verifiable.

open ecosystem signals

Where Voidly is already used.

·CC BY 4.0 dataset: 68 downloads on HuggingFace, free for academic + commercial use with attribution.
·@voidly/mcp-server: live in the Anthropic MCP Registry, integrated into Claude / Cursor / Windsurf / Cline.
·Framework adapters: 9 published for LangChain, CrewAI, AutoGen, LlamaIndex, Haystack, Vercel AI, x402, A2A.
·Awesome-list listings: 50+ curated indexes across MCP, A2A, x402, privacy, internet freedom.

04 / Compliance + security

What your security team will ask about.

data privacy

No PII. CC BY 4.0.

Source data is OONI, CensoredPlanet, IODA, and our own probe network — all network-level measurements against pre-defined domain lists. We do not collect end-user IPs, identities, or browsing behavior. The full public dataset is CC BY 4.0 (attribution required, commercial use allowed).

security

Cloudflare + D1 + R2.

The public API is a Cloudflare Worker (multi-region by default, isolated V8 sandboxes, no shared filesystem). Persistent state lives in D1 (SQLite) and R2 (object storage). ML inference runs on a hardened Vultr host (key-only SSH, fail2ban, UFW, kernel hardening, non-root service users). We can share the full hardening checklist on request.

compliance

Zero PII. GDPR-aligned. SOC2 on roadmap.

Voidly collects no personal data, so the GDPR surface is minimal by design. The full stack is auditable: open-source SDKs, CC BY 4.0 datasets, public APIs. We provide a self-attested security questionnaire and any available pen-test reports on request. SOC2 Type I is on the enterprise roadmap. Email enterprise@voidly.ai for the security packet.

05 / Pricing

Two tiers. Honest.

free tier

1,000 req/min, all public endpoints, full public dataset (CC BY 4.0), MCP server, agent SDK, RSS / Atom feeds, RSS for incidents, public webhooks. Fair use applies. No API key required for most read endpoints.

Public API docs →

enterprise tier

Tailored

Custom probes, unmetered API, SLA, dedicated support, managed filtered feeds, signed webhooks, security review docs. Pricing is set per deal based on probe footprint, delivery volume, and support depth — never on access to the dataset, which is free and CC BY 4.0 either way. Schedule a call.

Schedule a call →

06 / Onboarding

What happens after you email.

step 01

30-min discovery call

You describe what you need to monitor and why. We map it to existing endpoints, gaps, and any custom probe deployments required. No deck — just a working conversation.

step 02

2-week pilot

Sandbox API key, sample integration code (LangChain, CrewAI, raw HTTP, or your stack), and a Slack channel with the founder. We measure latency, accuracy, and fit against real workloads.

step 03

Production rollout

Signed contract, dedicated support channel, custom probes provisioned, webhook signing keys issued, and ongoing roadmap input. Renewal annually unless killed.

07 / Frequently asked

The questions buyers actually ask.

Can we deploy probes in our specific country or network?

Yes, with notice. The fleet size is public and machine-readable at /v1/probe/network — read it there rather than from this page, because it moves (it reported 40 active nodes on 2026-08-04 and 41 on 2026-08-05). Two things that endpoint cannot tell you, so we will: the split it publishes is `active_nodes` (core) versus community, and some of the community bucket is Voidly-run Fly.io machines registering through the same public path a volunteer uses — nothing in the response separates them, so read core as a floor on what we operate and community as a ceiling on third-party participation. /atlas/probe-coverage is honest that the covered countries are largely non-censoring. Enterprise customers can request additional probes in target jurisdictions — physical hosting, AS-level placement, or your own VM. Lead time depends on the region.

Can we get embargo / pre-publish access to new findings?

No — and this used to say yes, so the correction matters. Voidly is an open observatory; selling a head start on the data would make the publication date a product, which is exactly the thing the positioning says we are not. Every incident, forecast and finding hits the public RSS / Atlas feed for everyone at the same moment, free and CC BY 4.0. What an enterprise agreement buys is delivery and support around that public firehose — signed webhooks with retry, a filtered feed we build and maintain, an SLA, probes in your target jurisdictions — never earlier access to it. (Journalists: we do offer coordinated embargoes on write-ups we are drafting, free and to anyone reporting the story — see /journalists.)

Can we white-label the data, widgets, or feeds?

Talk to us. The underlying data is CC BY 4.0 (attribution required), so most embedding is already legal under the public license. White-labeling without attribution, custom branded dashboards, or rebranding API responses is an enterprise conversation.

Do you have SOC2?

SOC2 Type I is on the enterprise roadmap. Today we operate with strong informal controls: Cloudflare Worker isolation, zero PII collection, signed CI builds, and audit-logged D1 database access. For regulated buyers we provide a self-attested security questionnaire and any available pen-test reports. Email enterprise@voidly.ai for the security packet.

What's your SLA in practice?

Target: 99.9% uptime on api.voidly.ai (Cloudflare Worker, multi-region by default), p95 < 500ms latency on cached endpoints. We publish live status at /status. Enterprise contracts include credit terms when targets are missed; specific terms are negotiated per deal.

Where does the data come from? Is any of it PII?

Source data is OONI (CC BY 4.0), CensoredPlanet, IODA, and our own probe network — all network-level measurements. We do not collect end-user IPs, identities, or browsing behavior. Probes test predetermined domain lists from Citizen Lab. Zero PII in the public dataset, zero PII shipped to enterprise customers.

How quickly can we go from intro call to live integration?

Pilot can start within a week of the discovery call. Most teams are integrated end-to-end in 2 weeks. Custom probe provisioning in new countries takes 2–6 weeks depending on physical access.

How is Voidly vendor-neutral?

Every layer is open. Data is CC BY 4.0 — your integration owns the data, not us. SDKs are MIT-licensed on npm. The MCP server is open-source. 1.6M historical records and the public datasets are mirrored on HuggingFace. A Voidly Pay reference implementation runs on any Cloudflare Worker. Your stack survives independently of Voidly.

08 / Talk to us