Get started
Intro to Sands
Sands is an AI teammate for project work. You hand it the work the way you would to a colleague, and it plans, tracks and carries it out in your tools. Under the hood, Sands is an AI agent.
Why Sands
Agency owners and project leads spend hours turning briefs, threads and strategy docs into plans, tickets and status updates. Sands takes on that work. It reads your boards, drafts the plan, creates the tickets, and answers questions about progress with the numbers behind them.
The result lands where your team works: a plan you edit together, tickets in Jira, Linear or Asana, a message in Slack, a table for an invoice.
A teammate you delegate to
Working with Sands feels like working with a teammate who handles project operations:
- You ask in plain language, in the web app or by mentioning @Sands in Slack.
- It knows your tools: your projects, boards, people and channels.
- It keeps the thread: the plan you are working on, what it read, and which project you mean.
- It reports back with the result and the evidence, and says plainly when it could not do something.
An AI agent underneath
Sands is an AI agent: an AI model that, for each request, decides which steps are needed, uses your connected tools to take them, checks the results, and writes the answer. That is what makes it more than a chatbot, and it is also why it works the way it does:
| Because Sands is an AI agent | Sands is built to |
|---|---|
| It can misread a request or get something wrong. | Show every step it took, so you can check its work. |
| It can act in your tools. | Only carry out changes you asked for, on items it can identify, under your approval mode. |
| It answers from what it can see. | Read your tools live for numbers, and report missing data as missing, never as zero. |
| It works when you ask. | Act on your requests. It does not run on a schedule or change things on its own. |
Sands is not a person. Treat its plans and answers the way you would a capable colleague's first draft: a strong start you review before it reaches a client. See How Sands works and Approvals.
Key concepts
| Concept | What it means |
|---|---|
| Conversation | Where you ask. One chat holds one thread of work, including the plan you are working on. |
| Run | What Sands did for one request: each step, what it read and what it changed. Shown at the top of every reply. |
| Plan | Sands' main artifact: a short document with the goal, team capacity, tickets, risks, dates and open questions. |
| Connection | Sands' access to one of your tools, granted through that tool's sign-in. |
| Approval mode | Your rule for when Sands asks before changing your tools. |
| Credits | How work is measured. Each task uses what its work took. |
What Sands does
- Plans: drafts a plan from a brief, assigns your team, and sends it to your tool.
- Tracks: reads your boards live for health checks, retros, lists and hours reports.
- Acts: creates, assigns, moves, edits and comments on work; posts and reacts in Slack.
- Researches: searches the web and answers with sources.
The full list is in Capabilities.
Sands in Slack