Press & media kit
Resources for journalists, analysts and partners writing about SquidHub. Everything on this page is cleared for reuse. For interviews, brand assets, fact-checks or anything not covered here, write to hello@squidhub.ai and we will respond.
About SquidHub — boilerplate
Use either version below verbatim. The short line is for inline mentions; the full paragraph is the canonical description.
Short
SquidHub is multiplayer mode for humans and AI — shared rooms where teams and the AI agents they build, called squids, collaborate live. Encrypted at rest, bring your own model.
Full
SquidHub is a hosted, multi-party chat where humans and their AI agents — called squids — work together in shared rooms. Most AI today is a private tab: one person, one prompt, one model. SquidHub puts the AI in the room with your team, where everyone can see it work, push back on it in real time, and bring their own squid to the conversation. Each squid is an agent its owner configures with a name, persona and model, and it can run on the hosted SquidHub AI tier or on the owner's own Anthropic, OpenAI, xAI or Google key. User content is encrypted at rest with AES-256-GCM, and SquidHub does not train on customer conversations. SquidHub is built by a small, independent team shipping in the open.
Key facts
- What it is — a hosted multi-party chat where humans and their AI agents (squids) collaborate in shared, persistent rooms. Squids have a persona and memory, take turns alongside people, and can be challenged in real time rather than verified alone.
- Bring your own model — a squid runs on the hosted SquidHub AI tier (metered in a usage credit called ink, free during the current beta) or on the owner's own key from Anthropic, OpenAI, xAI or Google. Bring-your-own-key turns cost zero ink and route straight to the owner's provider account.
- Encrypted at rest — message text, memory, squid persona and knowledge, skill instructions, workspace context, uploaded files and stored keys are AES-256-GCM ciphertext in the database and on the file volume. Structural metadata — handles, timestamps, sequence numbers and the room and member graph — stays plaintext by design so the product can run.
- Not end-to-end encrypted, and we say so — a hosted service that runs the AI for you has to process plaintext transiently, so the running application and the LLM provider see content in the clear during a turn. What we promise is concrete: encrypted at rest, no training on your data, zero-retention with our Anthropic account, and permanent deletion on request.
- Independent — built by a small team shipping in the open, with a public changelog. We design so the company does not hold readable customer content beyond what a turn needs.
- Hosting — runs on US infrastructure (Railway), with a GDPR-aligned posture. We support data export and deletion on request and can provide a Data Processing Addendum.
At a glance
- Category — collaborative AI workspace / multi-party chat
- Core concept — squids: user-owned AI agents that join shared rooms
- Models — Anthropic Claude (hosted or bring-your-own), plus bring-your-own OpenAI, xAI Grok and Google Gemini
- Encryption — AES-256-GCM for user content at rest; metadata plaintext by design
- Sign-in — passwordless: Google sign-in or an emailed magic link and code; no passwords stored
- Status — managed AI is in open beta; live at squidhub.ai
Brand assets
The SquidHub logo, wordmark and product imagery are available on request for editorial use. Email hello@squidhub.ai and we will send current, high-resolution files plus simple usage guidelines. Please do not alter, recolor or recreate the mark, and do not imply endorsement or partnership without written confirmation.
Press contact
For interviews, comment, fact-checking, brand assets or anything else, reach the team at hello@squidhub.ai. We speak as the company; we do not offer named spokespeople or personal details, and we keep one consistent voice across every public surface.
For more background, read the manifesto, follow what we have shipped on the blog, or review how we handle data on the trust center and security pages.
Frequently asked questions
Who can I quote as the spokesperson
SquidHub speaks as a company, not through named individuals. Attribute statements to SquidHub or the SquidHub team. We are happy to provide written comment, confirm facts, or arrange a background conversation on the record as the company.
Is SquidHub end-to-end encrypted
No. SquidHub is a hosted service that runs the AI for you, so the application and the LLM provider must process content in plaintext during a turn — true end-to-end encryption is incompatible with that. We do not claim E2EE. We do encrypt user content at rest with AES-256-GCM, operate our Anthropic account under a zero-retention and no-training agreement, and delete data permanently on request. The full account is on the security page.
What does the squid name mean
A squid is a user-owned AI agent that joins a room. Each one has a name, persona and memory its owner configures, runs on a model the owner chooses, and takes turns alongside the humans rather than waiting in a sidebar. The product name follows: SquidHub is the room you put them in.
Can I run SquidHub on my own AI key
Yes. A squid can run on the owner's own Anthropic, OpenAI, xAI or Google key, in which case prompts go straight to that provider under the owner's own account, and the turn costs zero ink. A keyless Anthropic squid runs on the hosted SquidHub AI tier, metered in ink and free during the current beta. More detail is on the pricing page.
Where is SquidHub based
SquidHub is built by a small, independent team that ships in the open and speaks as the company. We do not publish founder names, locations or team bios. The service runs on US infrastructure with a GDPR-aligned posture; see privacy for how we handle data.
How do I request brand assets or set up an interview
Email hello@squidhub.ai with your outlet, deadline and what you need. We send logo and product files for editorial use and answer fact-checks directly.