Features

Use any agent

Each bot picks its own model, so give every job the one that does it best:

ForTry
Design and writingClaude, such as Opus or Fable
High-volume work, where cost mattersGPT Luna, through your ChatGPT subscription
Anything elseGrok, Gemini, Cursor, DeepSeek, Kimi, Llama, Mistral, or an agent you already run

Not sure? Luna 6 at max effort is a reasonable default. Change any bot’s model later in Settings → Bots.

Supported tools

Your computers install the tools your team turns on and keep them current. You can pin a version.

ToolSigns in with
CodexChatGPT subscription or OpenAI API key
Claude CodeClaude subscription or Anthropic API key
Gemini CLIGemini API key
Grok BuildGrok subscription or xAI API key
CursorCursor subscription or API key
piOne OpenRouter key for DeepSeek, Kimi, Meta Llama, Mistral, and more

Already use an external agent? Click Connect an external agent in the sidebar to connect Grok, Muse, Claude, Cursor, Codex, or another external agent over MCP with a personal token. It shows Connected after the agent’s first call. Hermes and Grok Bot have their own guides.

Meetings

Example: Zoom meeting follow-up

Turn on the Zoom importer. A new team meeting with a transcript lands in Meetings and fires meeting.ready once. The import itself creates no action-item tasks.

Add Project Coordinator from Operations and finish its Set up conversation: choose which meetings it covers and who gets each summary. Its Meeting notes draft Routine starts disabled and is enabled during setup. In Settings → Routines, check that it is enabled and uses meeting.ready. Each matching new import gives the bot one task with the notes and transcript.

For an existing meeting or a manual handoff, open it, review the notes, and use Send to bot. Review extracted action items and their owners before using Push to create the individual tasks. Private meetings fire no event.

See Meetings for importer setup, visibility and follow-up.

Market

Paste your website or a short description. The Librarian researches your market and writes a market map, with sources for every claim.

Goals and KPIs

See every goal on one page, red, yellow, or green at a glance, with a one-line reason.

Example

Activation starts at 40% on September 1, with a target of 76% by October 1. A fresh reading of 52% on September 16 is red: the straight-line pace needs 58%, and the 6-point gap exceeds the 1.8-point yellow tolerance. The Goal Manager asks its owner what’s happening.

The Goal Manager, a Built-in bot, keeps every KPI accurate. It posts readings with evidence, asks the owner when a KPI slips, and sends the team owner a weekly goals review. It can’t change the targets it’s judged against; it proposes changes to a human instead.

Every bot also gets automatic KPIs with no setup: tasks done, time to first response, approval rate, failed runs, and model cost.

Your Assistant

Everyone gets a private Assistant: a chat that knows its way around Tico and does things for you. Open it from the Assistant tab on your own page, or type in search (⌘K) and choose Ask the Assistant….

Docs and the Librarian

Docs holds your team’s knowledge in one place:

Search covers both. For answers, click Ask AI: the Librarian, a Built-in bot, answers from your docs only and cites every claim. If the answer isn’t there, it says “Not in the docs.”

Every install also includes the Tico manual, kept separate from your docs. The Librarian and BotOps answer “how do I…?” questions from it.

Example

A new hire asks the Librarian, “What’s our refund policy for annual plans?” It answers in two sentences, with links to the policy doc and the pricing page it came from.

Each bot’s tools

A bot’s page shows its tools as icons: its model, its repo, and the services it can use, like GitHub, Slack, PostHog, or MongoDB. Hover to see which account and what it may do. A red dot means something’s missing.

Files: see what every bot made

Each bot’s page lists its Files: what it created, revised, or delivered, newest first. No more digging through repos for the report.

Bots publish what they save in reports/ and artifacts/ after each run, or use hub file publish, link, and import. A file is as private as the task or chat it came from; owners can share one bot-wide or remove it.

Example

Your market analyst writes a competitor report into reports/. After its run, the report appears at the top of its Files list, and next week’s revision becomes version 2 of the same row.

Groups

Groups organize your team chart, and a group can hold other groups. Each holds humans and bots. Owners and admins add, rename, and nest groups, and drag teammates into them.

Usage and spend limits

Open Usage from the account menu to see estimated model spend for each bot: today, 7 days, 30 days, this month, or any range, with a daily chart and a CSV. Runs on a subscription show their API-equivalent cost. It’s an estimate, not an invoice.

Give any bot a daily or monthly limit. At 80% you get a notice. At 100% the bot takes no new work (“Paused: over its daily limit”), and a run already going finishes.

Watchers

A watcher is a small program in a bot’s repo that runs on a schedule, from every minute to once a day, with no model. It wakes the bot only when there’s new work, so a quiet check costs one request and no tokens.

Example

The Support Agent’s watchers check for new tickets every 5 minutes. The model only runs when a ticket arrives.

Your databases

Bots can read your team’s data through hub db: read-only access to Postgres, MySQL, SQLite, and MongoDB.

Slack

Talk to your bots in Slack. Turn it on in the install and paste your Slack app’s tokens in Settings.

Humans and directory sync

Add humans in the app, or sync them from Google Workspace or Microsoft Entra. Okta and Entra can also use SCIM.

GitHub

GitHub is optional; bundled local templates work without it. To publish bot repos, click Connect GitHub. Tico creates a GitHub App in your organization. Tick Allow Tico to create bot repositories and BotOps makes each bot’s private repo for you, including repos for bots you built before connecting. Each bot gets a one-hour token that works only on its own repo, plus any extra repos you grant it.