Safety

Six things knectar will not do.

You give knectar your AI keys and your prompts. It will not do these six things with them.

  1. 01

    Hand your AI key to an app

    Apps only ever get a knectar key. Your AI key is stored encrypted on our servers and stays there.

    because apps only ever hold a knectar key

  2. 02

    Show your AI key again

    After you save it, the dashboard shows only its last four characters, to you and to everyone on your team.

    because the full key is never sent back out

  3. 03

    Write your keys into a log

    When sign-ups open, your knectar keys and your AI keys stay out of request logs and out of the audit log.

    because keys are stripped before anything is logged

  4. 04

    Leave your stored keys open to the internet

    Your stored keys and settings sit on a private network. From the internet, only AI requests, a status check and the dashboard reach knectar. The dashboard needs you to sign in.

    because key storage is on a private network

  5. 05

    Save answers without being asked

    Nothing is saved for reuse unless the request turns saved answers on.

    because saved answers are off unless you ask

  6. 06

    Share saved answers with another customer

    Saved answers are kept per customer and per knectar key. If knectar cannot tell whose request it is, it skips them.

    because saved answers are kept per customer

Not built yet: knectar does not hide personal data in prompts. It also keeps each prompt and reply for 24 months, and you cannot turn that off yet.

What is ready, and what is not built yet.

Ready, not live yet means the item is built, and customers can use it once sign-ups open.

ready, not live yetcomingnot built

StatusItemWhat it means
Ready, not live yet knectar keys One for each app. Create, swap, switch off or delete them in the dashboard. You see the full key once.
Ready, not live yet Your own AI keys Paste your OpenAI, Anthropic, Gemini, Groq or OpenRouter key when you create a knectar key. It is stored encrypted and tied to that one key.
Ready, not live yet Backup models Name a backup model in the request. If the first AI provider fails, knectar sends the same request there. A backup on Anthropic needs your Anthropic key on the same knectar key.
Ready, not live yet Routing rules Set a backup model for each key in the dashboard, or send a share of traffic to another model. No code change.
Ready, not live yet Saved answers Turn them on per request with one header. A request that is the same as an earlier one, or very close to it, gets the saved answer. Kept per customer and per key. Best for repeat jobs like tagging, not for chat.
Ready, not live yet Logs One row for every AI request with the model, time taken, tokens and cost. Filter by key, model or status. An admin can delete a single request.
Ready, not live yet Spending caps and limits Cap what each key can spend per day, week or month, and how many requests or tokens it can send. Your AI provider still bills you as it does now.
Ready, not live yet Playground Try any model you connected, in your browser, with your knectar key. No code. Replies arrive whole for now, not word by word.
Ready, not live yet Audit log See who created, changed, swapped or deleted a key, and who added or replaced an AI key. Admins only.
Coming Model providers page One page showing each AI provider your account can use, which key serves it and its models.
Not built Hiding personal data Removing personal data from a prompt before it leaves knectar.
Not built MCP tools Letting your AI call MCP tools through knectar.
Not built A setting to stop storing prompts Today each prompt and reply is kept for 24 months. A setting to stop storing them is not built yet.

knectar is running and answering requests. No customers use it yet. The waitlist hears first when sign-ups open.

Get your knectar key when sign-ups open.

Join the waitlist