Skip to main content

​
Overview

Home is the dashboard landing page for your workspace. It shows high-level operating signals — live articles, review backlog, resolved conversations, and CSAT — plus shortcuts into review work, Inbox triage, and the bots in your workspace. Each bot is an agent: its own knowledge scope, its own Behavior, its own Rulebook, and its own deployment surfaces. Most teams end up running multiple bots side by side — one for customer support, one for internal IT, one scoped to a specific product line, one for a specific brand. Same workspace, different lanes.
Home page in the Fini Demo workspace showing workspace metrics, Generate an Article and Go to Inbox shortcuts, review queue drafts, and Your bots cards for test and Venmo.

​
Managing bots from Home

The Your bots section lists the bots in the workspace with conversation, resolved, and resolution-rate snapshots. Use + Create bot when your workspace needs a new agent, or open an existing bot to continue configuration.
Name bots by function from day one. Support-Billing scales better than Bot-1, especially once the same workspace has multiple deployment surfaces and review queues.

​
The agent selector

Most pages in Fini (Knowledge, Inbox, Test Suite, Analytics) scope their data to a single agent using the agent selector in the product shell. The selector defaults to whichever agent was last active; switch it to scope the page to a different agent’s data. The selector behaves slightly differently per surface, but the principle is the same:

​
Knowledge

When set to All, the Knowledge page shows every article across every agent in your account. Switch the selector to a specific agent to filter the folder tree and article list to only the knowledge connected to that agent. See Articles for how per-agent scoping actually works (toggle attachments, don’t duplicate articles).
Knowledge page showing the selected agent, Assign folders button, Venmo folder tree, and empty article pane

Knowledge with the agent selector set to All. The full folder tree across all agents is visible.

Knowledge page filtered to a single agent, showing Venmo knowledge folders and the Assign folders button

Knowledge filtered to a single agent. The folder tree now shows only the sources and articles connected to that agent.

​
Inbox

The Inbox scopes conversations to a single agent the same way. Once you select an agent, use the date range picker and filter chips (Ticket Id, Conversation status, Channel, Source, and + Filter) to drill into exactly the conversations you need.
Inbox page filtered to a single agent, showing the Ticket Id, Conversation status, Channel, Source, and date range filters, plus an empty conversation list state

Inbox filtered to a specific agent. The agent selector, date range, and filter chips are all visible in the toolbar.

Use the Channel filter in Inbox to compare conversations per deployment. This is useful when the same agent runs on both Zendesk and the widget and you want to understand behavior differences by channel.

​
Test Suite

Test suite runs test sets against the selected agent. See Test Suite.

​
Analytics

Here you can scope metrics to the selected agent. See Analytics.

​
What to configure next

Once an agent is selected, the next steps fall into three groups: what it knows, how it behaves, and where it runs.

​
What the agent knows

Knowledge

Connect sources (URLs, files, Drive, Notion, Confluence, Zendesk), review and publish articles, and scope which articles this agent can see.

Attributes

Bring in customer context — plan tier, order status, account flags — that the agent can reference when answering.

​
How the agent behaves

Behavior

Persona, tone, escalation triggers, and the instructions that shape how the agent reasons and responds.

Rulebook

Deterministic flows for high-leverage intents — order status, password reset, cancellation. The Rulebook runs before the LLM gets the question.

Reply Rules

Reply-time controls: when to send, when to suppress, when to escalate, when to take an action.

Tags

Categorization for conversations — used by Analytics to slice volume by topic and by Rulebook to gate behavior.

​
Where the agent runs

Deploy

Connect the agent to Zendesk, Intercom, Front, Salesforce, the widget, your help center, or any of the other supported surfaces.

Test Suite

Validate the agent against scenarios before it goes live — and continuously after, as you change knowledge or prompts.

​
Working across multiple agents

The same knowledge graph powers all your agents — articles are workspace-wide, and each agent’s scope is controlled by attachment toggles on folders and articles. Don’t duplicate articles to scope to an agent; toggle which folders the agent has access to. See Articles for the full per-agent scoping model. Agents can share knowledge, share Actions, share Attributes — but their prompts, rules, and reply behaviors are independent. That’s the unit of separation: same data, different behavior.