Changelog

Stop is now settled by what actually happened: if the reply had already finished, you keep it and the run shows how long it took. A request you stopped is not answered later unless you ask for it.

Replies finish even when the tab is in the background.

Bonus credits show separately from your plan credits, the Pro meter counts only plan credits, and Usage keeps its history when you upgrade.

A cookie notice lets you accept or reject analytics and session replay. Change it any time from Cookie settings or Settings → Privacy.

The docs now cover everything in Sands: guides by team and industry, every capability, a page for each integration with everything Sands can do there, credits, settings and this changelog.

Where: Docs.

Clicking Upgrade to Pro twice within a few minutes could fail to open checkout. The second click now opens the same checkout as the first.

The bottom of the sidebar now shows your name and plan. Click it for Settings, Usage, Upgrade plan, Learn more and Log out. Settings also opens with ⌘, from anywhere.

The megaphone button next to it sends feedback to the Sands team. Write what you noticed; the chat you have open is attached so we can see what happened.

Where: Sidebar, bottom left.

Upload a photo, pick a Sands mark in your own shape and colors, or use your initials. Click your avatar in Settings to change it.

Where: Settings → Profile.

Plan & billing now cancels Pro at the end of the period, or resumes it, without leaving Sands. Payment methods and invoices are still one click away in the billing portal.

Usage keeps its history when you change plans: This period, Previous periods and All time, with a marker where you upgraded.

Where: Settings → Plan & billing and Settings → Usage.

If your balance cannot cover a request, Sands runs nothing and charges nothing. Your message stays in the chat with one card that explains why and offers the upgrade.

A request that spends your last credits finishes what it can and shows the card under its work. A request you stop before Sands replies or uses a tool is free.

When you ask for a change, the plan document shows it happening one place at a time: new dates, each rescheduled ticket, removed questions. Nothing is tinted or swapped wholesale; a cursor marks where Sands is working.

Undo the whole change with ⌘Z right after it lands, or from the plan's title menu with Undo last Sands edit. The chat notes the update, and the steps stay in the run.

Where: Any open plan.

New plans fill each ticket's priority and estimate, assign owners from the team you named, and date tickets inside their workstream. Schedule changes in chat reschedule the tickets they affect.

Every account starts with 500 free credits. Pro is $29 a month for 3,000 credits each month. A task uses the credits its work actually took, so a quick question costs a few and building a plan costs more.

Where: Pricing and Settings → Plan & billing.

Delete a plan from its title menu. Undo is offered right away, and a deleted plan can be restored from its chat card for 30 days.

Click a plan's title for Export, Copy as Markdown, Rename, Duplicate and Delete. Send to has one button for your last-used tool, and its arrow lists Jira, Linear and Asana plus Word, PDF, Markdown and backlog CSV.

Tables, images, code blocks and dividers get a handle for Delete, Duplicate and Move. Any section can be deleted, and Undo brings it back whole.

Where: Plan drawer header.

Every reply ends with Copy and thumbs up or down. When you send, your message moves to the top and the reply fills in below it; scroll up at any time and Sands stops following.

Tables in replies, like hours reports, appear row by row and copy into Docs, Pages, Gmail or a spreadsheet as a real table.

Send to lists every supported project tool, connected or not. Picking one you have not connected opens the connect step, then goes straight to that tool's preview of what will be created.

After a send, the chat shows a table of everything created, each with a link.

Type / on an empty line for headings, lists, checklists, tables, quotes, code, dividers, images, dates and people. Hover a table for Add row and Add column.

Priorities read High, Medium and Low, and every date opens a calendar.

A plan opens in a drawer the moment you ask for it and fills top to bottom as Sands writes: the goal, capacity from the team you named, workstreams, a ticket table, risks, key dates and open questions.

Every line is editable, like a doc. Edits save on their own, and if Sands changes the same block at the same time you choose which version to keep.

Where: Ask Sands for a plan, a spec or a sprint plan.

Assign owners, set due dates, change priorities or move the project dates in chat. Sands changes only those fields and tells you exactly what changed, including anything it could not assign.

Sands now reads each request on a stronger model and checks its reading before acting. Edits to the plan you have open stay edits; they are never mistaken for a send.

Ask for a plan and Sands drafts it, even from a short brief. Anything it does not know becomes a labeled assumption or an open question in the plan instead of a round of questions first.

You choose the project tool when you send, not before you start.

After a reply, Sands suggests next steps based on what it just found and what you are working toward. When there is nothing useful to suggest, it suggests nothing.

Create an account or sign in with your Google account.

Usage breaks credits down by apps and tools, models and conversations, with a daily chart for the period.

Where: Settings → Usage.

When a request needs a tool you have not connected, Sands shows a card naming what it will do, and one click starts the connection. After you connect, Sands carries on with the request.

If a connection expires mid-request, one card asks you to reconnect and the request resumes without doing anything twice.

Set the name Sands uses for you, and delete your account from Settings when you need to.

Where: Settings → Profile.

Invite links bring new people into the beta with their free credits ready. Sands Pro is available from Settings, and after checkout you land back in the chat you came from.

See your plan, your credit balance, when it renews, and how you have used your credits.

Where: Settings.

Answers that used the web end with the sources Sands read. Every answer can be copied or rated.

Ask Sands to research a market, a vendor or a technical question and it searches the web, reads the results and answers with sources. Research can feed a plan or a comment you ask it to write.

The + button in the composer attaches files and connects apps. The home screen suggests prompts with blanks to fill in; clicking one never sends it.

Choose how Sands handles changes to your tools: Manually approve, Automatically approve (pauses only for riskier changes), or Skip all approvals. The control sits next to + in the composer.

Asking Sands to post to one channel or person sends it right away and confirms where it went. Messages to several places or with @here or @channel still ask first.

A comment Sands writes to a ticket contains your own words or facts it read from your tools. The approval card shows the full comment text before it posts.

Each reply now opens with one line for the work Sands did. Expand it to see each step, what it read and what it changed; the answer streams below.

Mention @Sands in a channel or message it directly and it answers in the thread with your connected tools. Follow-ups in the thread do not need another mention.

Where: Settings → Apps → Slack.

Sands reads your channels and threads and sends messages and reactions from chat, alongside your project work.

Read and manage Asana projects, sections and tasks from chat, and send plans to Asana as a project with sections, tasks and subtasks.

Read and manage Linear issues, teams and projects from chat, and send plans to Linear as a project with milestones, issues and sub-issues.

Every run streams the same way: each step says what it is doing, shows what it found, and closes with a one-line takeaway.

Before the entries above, Sands worked with Jira: read projects and issues; create, assign, move and comment on work; turn a brief into a plan and send it to Jira as epics and stories; check project health; run retrospectives; and build client-ready hours reports from worklogs.