Private AI chat that stays on the low
A private AI chat is a conversational AI workspace where your messages, files and provider credentials are readable only by your own account. ChatOTL enforces that at the database level with row-level security, keeps uploads in a private bucket addressed by your user ID, and can be self-hosted on infrastructure you control.
Most AI chat products are private in the marketing copy and ambiguous in the terms of service. ChatOTL takes the opposite approach: privacy is a property of the schema, not a promise on a landing page.
Every conversation, message, setting and uploaded file is owned by exactly one account, and the database refuses reads from anyone else. You can verify it — the access rules ship with the project and are covered by automated tests.
What makes this chat private
- Row-level security scopes conversations, messages, settings and files to the account that created them.
- Uploads live in a private storage bucket keyed by your user ID and are reached only through short-lived signed URLs.
- Provider API keys are stored per account and are never shared between users.
- The whole workspace can be self-hosted, so the deployment boundary is yours.
Privacy without giving up capability
A private workspace should not be a worse workspace. Replies stream token by token with stop and regenerate, you can switch models from OpenAI, Anthropic or Google mid-conversation, and any answer can be forked into its own branch when you want to explore a different direction.
What privacy does not mean here
When you send a message to a hosted model provider, that provider processes it. ChatOTL controls who can read your data inside the application; it cannot change a third-party provider's own retention policy. If that matters for your work, point ChatOTL at a self-hosted OpenAI-compatible endpoint and the traffic never leaves your infrastructure.
Where it fits
Work material you can't paste into a shared tool
Draft, summarise and review internal documents in a workspace that only you can read.
Personal thinking space
Keep long-running threads about health, finances or career decisions out of a shared account.
Client work under NDA
Run each engagement in its own thread and self-host if the contract requires it.
Limitations
- Hosted model providers still process the prompts you send them.
- ChatOTL does not browse the live web, so it cannot look up current information.
Frequently asked questions
Is ChatOTL actually private, or just encrypted in transit?
Both matter, but the meaningful control is authorisation. Database policies restrict every row to its owning account, and uploaded files sit in a private bucket namespaced by user ID. Transport encryption is table stakes on top of that.
Can an administrator read my conversations?
The owner dashboard reports counts, usage and errors rather than conversation content. If you self-host, you control who has infrastructure access — which is the honest answer for any hosted product.
Do you train models on my chats?
No. ChatOTL stores your conversations so you can return to them, and nothing else.
What happens to files I upload?
They go to a private bucket under your user ID. Text and code are extracted for the prompt; PDFs and images are passed to the model through signed URLs that expire shortly after use.
Start a private conversation
Create an account and your first thread is scoped to you from the first message.
Start chatting