Use TypeSafe AI with MonstarX
In shortPaste a TypeSafe API key into the TypeSafe AI row in Backend → Cloud. MonstarX's builder can then ask Jev, TypeSafe's model, up to 20 typed questions at once — yes/no, pick one of several options, or a score — and get back answers with probabilities. The key is for the builder only.
Typed yes/no, choice and score answers from Jev.
- Type
- Business tool · API key
- Category
- AI and media
- Connect with
- API key
TypeSafe AI makes Jev, a fast AI model that does not write text: it answers questions with a typed answer and how sure it is. “Is this review a complaint?” comes back as a probability of yes. “Which team should handle this message?” comes back as one of the options you gave, with a probability for each.
Connect your own TypeSafe key and MonstarX’s builder — the AI you chat with — can put those questions to Jev while it works: sort a pile of feedback, check which rows of your data match a rule, or score items so it can build around the results. The app you build never gets this key.
What can MonstarX do with it?
Section titled “What can MonstarX do with it?”| What MonstarX can do | When |
|---|---|
List the models your key can use (such as jev-latest) | Any mode |
| Ask a yes/no question about some text or data, and get the probability that it is true | Any mode, when you ask for it |
| Ask a choice question — which of your named options fits best, with a probability for each | Any mode, when you ask for it |
| Ask a score question — where something sits on levels you describe, like “not urgent” to “very urgent” | Any mode, when you ask for it |
One request can hold up to 20 of these questions about the same text or data.
What it never does: put the key into your app’s code, generate free text, or change anything in your TypeSafe account.
Before you start
Section titled “Before you start”- A TypeSafe account and API access. Sign in at the TypeSafe console.
- Costs. TypeSafe charges per input token (output is free). Current prices and limits are on TypeSafe’s Models page.
- You own the MonstarX project. Only the owner can connect or disconnect a tool.
Get your key
Section titled “Get your key”-
Sign in to the TypeSafe console and open the API keys page.
-
Create a key and copy it.
To try questions yourself first, TypeSafe has a Playground and a quick start.
Connect it in MonstarX
Section titled “Connect it in MonstarX”-
Open your project and click Backend in the top bar, then Cloud.
-
Scroll to Connectors, find TypeSafe AI and click Connect.
-
Paste the key into the API key field and click Connect.
-
MonstarX checks the key by listing your models. When it works, the row says Connected. If the key is refused, you see the error and nothing is saved.
Ask MonstarX to use it
Section titled “Ask MonstarX to use it”| You type | What happens |
|---|---|
| “Use Jev to sort these 30 customer messages into billing, bookings and complaints.” | MonstarX asks Jev a choice question for each message and shows the groups. |
| “With Jev, check which of our class descriptions mention an injury risk.” | It asks a yes/no question per description and lists the likely ones with their probabilities. |
| “Score each testimonial from ‘lukewarm’ to ‘glowing’ with Jev and put the top three on the homepage.” | It scores them, picks the three highest and, in Build mode, adds them to the page. |
Security and privacy
Section titled “Security and privacy”- Stays with MonstarX. The key is stored encrypted on MonstarX’s servers. It is never written into your app’s code or given to your app.
- One project. The key works only in the project where you connected it.
- Your data goes to TypeSafe. The text or data in a question is sent to TypeSafe with your key.
- Disconnect any time. Click Disconnect beside TypeSafe AI in BackendCloud and confirm Disconnect TypeSafe AI from this project?. MonstarX forgets the key. Delete it in the TypeSafe console to stop it working everywhere.
Good to know
Section titled “Good to know”- A single request holds at most 20 questions and about 100,000 characters in total.
- Jev answers with probabilities, so you can see how sure it is about each answer.
- Jev reads text and structured data only — not pictures, audio or video.
Frequently asked questions
Section titled “Frequently asked questions”How is Jev different from Anthropic or Gemini?
Claude and Gemini write text. Jev answers a question with a typed result — yes/no, one of your options, or a score — and a probability. It suits sorting, checking and ranking.
Can my app use Jev?
Not with this connection: the key is for MonstarX's builder only. If you want a Jev-powered feature in your app, add a TypeSafe key under Backend → Secrets and ask MonstarX to build it. See Store API keys safely.
What does a question cost?
TypeSafe bills per input token, so the length of the text or data you ask about matters most. Listing models does not use tokens.