Skip to main content
You go from first look to a live assistant in seven steps: try the public playground, get access, sign in, connect a source, ask and check the citations, put the assistant on your website, API or MCP, and invite your team. After that the same account opens Chat, AI search visibility and AI employees. Your own part takes under an hour; the waits are for a person to set up your access and for your first sync. Each step below says what you do, what you see when it worked, and roughly how long it takes, and links to the page with the detail. The knowledge product shows the two steps to a first answer as a checklist, Set up your first answer, until your assistant can answer.

Before you have an account

1

Try the playground

Ask MITHUNAI’s own assistant a question at play.mithunai.com, with no account and nothing kept. It answers only from MITHUNAI’s public documentation.You see: an answer with the page it came from, or a plain statement that the documentation does not cover your question. About 1 minute.
2

Get access

Request access, or accept an invitation from a colleague whose organisation already uses MITHUNAI. There is no instant sign-up: a person sets up each new organisation and its first Owner. To start the conversation, book a demo.You see: an email that lets you sign in, or one that lets you accept your seat and choose a password. A person reviews each request, so it is not instant. See Invitations.

Your first cited answer

1

Sign in once

Sign in at app.mithunai.com. The same account is your way into every MITHUNAI product; Chat already lets you straight in, and the rest is in beta.You see: Home, with every product on your account and what needs you. Open Knowledge to find the Set up your first answer checklist. About 1 minute. See Sign in once.
2

Connect your first source

In Sources → Connect source, add your documentation site, a sitemap, a GitHub repository or a file. Fetching and indexing run in the background; you can leave the page. If a source fails, it says why and offers a re-sync.You see: the source with “N of M documents indexed”, then marked ready. About 2 minutes, then minutes for a small site and longer for a large one. See Sources and Ingestion.
3

Ask and check the citations

Select Set up and ask a question in the checklist. That creates an assistant over your knowledge with the default settings, which you can change later in Customizations. Then ask a real question in the Playground and open each citation.You see: an answer with numbered sources that open the passage they came from, or “I don’t know” when your content does not cover the question. About 2 minutes. See Configure an assistant and How answers work.

Live, with your team

1

Put it on your website, API or MCP

Under Integrations → Website widget, enter your website’s address, paste one line into your site and select Check my website. Or create an API key and call the API from your backend, or connect an AI tool over MCP (beta, read-only).You see: the Ask AI button on your own site. About 10 minutes. See Deploy.
2

Invite your team

Owners and Administrators invite colleagues from Settings → Members → Invite by email, choosing each person’s role. Each gets a link, valid once for seven days, to choose their own password.You see: each invitation listed until it is accepted, then the person in Members. About 1 minute a person. Invitations need your deployment to send email; until it can, the operator creates accounts. See Invitations.

Then, with the same account

Once your assistant is live, the console’s Dashboard shows What’s next with a card for each of these that your organisation can use. A card appears only where that product is switched on for you.

Chat

A chat workspace with your knowledge in every thread: sign in once, switch on Knowledge, and get cited answers. Available.

AI search visibility

How AI engines answer your buyers’ questions, and who they recommend. Beta, off by default, not yet run against a real AI engine.

AI employees

MITHUNAI Team: Support, IT service desk, DevOps, Marketer, Sales and Finance employees that do routine work and ask a person before they act. Beta, off by default.

Gateway

One endpoint in front of your model providers. For platform administrators, and for Owners and Administrators where a deployment allows it.
In Chat, start with Answers from your knowledge. For AI employees, start with Approvals and the approval policy, then open the employee for your team: Support, IT service desk, DevOps, Marketer, Sales or Finance. MITHUNAI Agents is in beta, with access on request; see Agents. Product status lists every capability and its status.

If a step stalls

Last modified on November 7, 2026