Videos

From zero to the first answer

Duration 1:39Narration: English

A fresh installation of Mislata on the firm's own servers opens a setup wizard. It asks for a one-time token printed in the server logs, then the licence key, then the firm's Microsoft Entra ID app registration, which Mislata tests before saving it. When setup is complete, Sarah signs in with her Microsoft work account and creates the organisation.

She connects the AI model: the firm's own Azure OpenAI deployments, one for answers and one for search (embeddings). Mislata tests both before saving them. She uploads the first nine documents, which are read and indexed on the firm's servers, and asks which portfolio companies depend on Kestrel Data Labs and what the vendor due diligence found. The answer cites the source of every fact. Keys and secrets are not shown at any step.

What the video shows

A setup wizard on a fresh installation, opened with a one-time token from the server logs
Sign-in through the firm's own Microsoft Entra ID app registration, tested before saving
The firm's own Azure OpenAI deployments for answers and search, both tested before saving
Documents indexed on the firm's servers, and answers with every source cited

Transcript

Halberd Capital Partners, London. Mislata runs on the firm's own servers.
A fresh install opens with a one-time token from the server logs.
Then the licence key.
Next, the firm's Microsoft Entra ID app, tested before it is saved.
Setup is done. Sarah signs in with her work account.
She creates the organisation.
Then she connects the AI model: the firm's own Azure OpenAI deployments, one for answers and one for search.
Mislata tests both before saving.
She uploads the first documents. Mislata reads them and indexes them on the firm's own servers.
The first question.
And the first answer, built from the firm's documents, with every source cited.
Installed on your infrastructure, connected to your identity and your AI model, and answering from your documents.