Settings

workspace + admin configuration

About this section

Theme + font + language preferences, password policy, notification sender, default timezone, plus the company profile + branding setup.

Theme picker (light / rust / slate / auto)
Per-user font + language preferences
Workspace-wide password policy + login config
Company profile + branding (logo, accent)

ADMIN · SETTINGS

Settings

Workspace defaults, authentication, and AI-provider config. Per-user preferences (theme, language) live on your Profile.

Company context in AI conversations

When on, every new chat starts by silently feeding the persona your Company profile (industry, products, goals, challenges). Personas reference your specifics in their answers instead of generic advice. Stored per browser — flipping it here also flips the in-chat checkbox.

Include company context in conversations

Loading…

— Edit company profile

Default timezone

Used for cron schedules + memo timestamps for users without their own preference.

Timezone

Password policy

Workspace-wide password requirements. Enforced on register + password change.

Minimum length

Require uppercase Require lowercase Require digit Require special character

—

Entra ID single sign-on

When enabled, users see a "Sign in with Microsoft" button on the login page. Uses an Azure AD app registration in your tenant — different from the Foundry service principal under AI provider.

Enable Entra ID SSO

Tenant ID

Client (App) ID

Client secret

Redirect URI

—

Notification sender

"From" address used on outgoing emails (password reset, alerts). Sent via Microsoft Graph using the same Entra credentials configured above.

Sender email

Image-generation rate limit

Per-user images per hour. 0 disables image generation entirely.

Per-user / hour

Desktop installer (Flowpipe for Windows)

Publish a fresh build of the Tauri client. The .exe always updates the in-app download; add the matching .sig to also publish the auto-update so existing clients self-update. Version is read from the filename (Flowpipe_0.2.0_x64-setup.exe). Max 200 MB.

Loading…

Installer (.exe / .msi)

Signature (.sig) — optional, enables auto-update

Release notes — optional

Mobile notifications

Get pushed on your phone when a flow needs you — a review to approve, a failed run. Install Flowpipe to your home screen first, then enable notifications here. On iOS this only works from the home-screen app (iOS 16.4+).

Checking…

Foundry connection health

Live diagnostic of credentials, endpoint reachability, and deployments visible to Flowpipe. Click Refresh to re-run; everything below auto-populates from the same data.

Credentials · —

Endpoint · —

Deployments · —

—

Model selection

Pick which deployed Foundry model fills each role. Options are your live deployments (the same list the agents use), so you can point any role at any model in your tenant. Changes apply after a service restart.

Loading deployments…

—

Foundry endpoint

The Azure AI Foundry resource Flowpipe talks to. Tenant identifies the Azure account; Endpoint is the project-scoped URL Foundry hands you when you create a project.

Foundry endpoint

Tenant ID

—

Authentication method

How Flowpipe authenticates against Foundry. API key is simplest (one secret, set-and-forget). Service principal is recommended for production — RBAC, rotation, audit trail. Azure CLI is for dev only — uses your local az login session.

Client ID (Application ID)

Client secret

Recommended for production. The service principal needs Cognitive Services User + Azure AI Developer roles on the Foundry resource. Tenant ID + Subscription are taken from the card above. Leaving the secret blank keeps the saved one.

Visible deployments

Models the current credentials can call. Personas pick from this list when admins set their model field. Refresh after a credentials change.

                Click Refresh on the health card to load deployments.
            

Azure AI Search

Connect an Azure AI Search service so agents can be grounded on a search index — hybrid vector + keyword retrieval — and so indexes can be created and managed from Flowpipe. The search service is provisioned once in Azure; everything else is configured here.

Search service URL

Reuses the Foundry service principal configured above — no extra secret to store. Two one-time steps on the Search resource in the Azure portal:

Enable role-based access — Search service → Security + networking → Keys → API access control → choose Both (or Role-based access control). Without this the service rejects the service principal with a 403 even when the roles below are assigned.
Assign roles under Access control (IAM): Search Service Contributor (create & manage indexes) + Search Index Data Contributor (read/write documents). Query-only grounding needs just Search Index Data Reader.

Agent grounding connection: —

Create a connection to this service once in the Foundry portal (Connected resources → New connection → Azure AI Search) — Flowpipe finds it by URL, nothing to paste. Its identity needs Search Index Data Reader on the service.

—

Indexes create & fill keyword indexes here — agents can then be grounded on them

—

Paste text, or upload a PDF, Word (.docx), .txt or .md — text is extracted, chunked, and indexed.

API keys

Programmatic access to the Flowpipe API. Use header Authorization: ApiKey YOUR_KEY. Each key is labelled, can have an expiry, and tracks its own usage.

Active keys

—

Total calls across all your keys

—

Loading…

Label *

Expires (optional)

Scope

Read-only (recommended)
List flows, get definitions, trigger runs, monitor status. Safe default. Read + write ⚠
Additionally create / update / delete / enable flows via MCP. Mutations still require per-flow opt-in.

Restrict to specific flows — useful for CI / Claude Desktop keys that only need one flow. Off = any flow the owning user can otherwise access.

Use with Claude, ChatGPT & more (MCP)

Drive your flows from any MCP-capable assistant — list, run, monitor and read results without leaving the chat. Create an API key above (read = list / inspect / monitor; write = also run, cancel and edit), then pick your client:

Replace YOUR_API_KEY with a key from above — it is shown only once at creation. The MCP endpoint is .

Try it

Test any endpoint without leaving the page. Same RBAC, same audit trail. Works against your existing key, your session, or no auth.

Auth

API key

Request

Presets:

Response headers

How to use it

Send the key in the Authorization header on every request.

curl -H "Authorization: ApiKey YOUR_KEY" \
     https://live.flowpipe.ai/api/personas

import requests

resp = requests.get(
    "https://live.flowpipe.ai/api/personas",
    headers={"Authorization": "ApiKey YOUR_KEY"},
)
resp.raise_for_status()
print(resp.json())

const r = await fetch("https://live.flowpipe.ai/api/personas", {
    headers: { "Authorization": "ApiKey YOUR_KEY" }
});
console.log(await r.json());

Security guidelines

Treat the key like a password — anyone holding it acts as you.

Store in a secret manager or .env file — never commit to Git.
Don't paste into client-side JavaScript or screenshots.
Rotate periodically, or immediately if you suspect a leak.
Revoke unused keys — fewer keys means smaller blast radius.
Each call is logged with your user ID. Audit trail in User management.

Company context in AI conversations

Login gate

Default timezone

Password policy

Entra ID single sign-on

Notification sender

Image-generation rate limit

Desktop installer (Flowpipe for Windows)

Mobile notifications

Foundry connection health

Model selection

Foundry endpoint

Authentication method

Visible deployments

Azure AI Search

API keys

Use with Claude, ChatGPT & more (MCP)

Try it

How to use it

Security guidelines