Starter
$29/month
Up to 500 customers
- Daily churn and retention analysis
- At-risk customer warnings
- Retention Signals and suggestions
- Access for your team and agents
Turn your agent into a churn killer.
See who’s about to churn, what makes customers stay, and what to build next. In the agent you already use.
14-day free trial.
- You: Send a 20% win-back offer to everyone who’s cancelling.
- Using Databrush: Found 3 users who are cancelling
- Using its own tools: Sent 3 win-back offers
- Using Databrush: Marked them contacted in Databrush
- Agent: Done. All 3 have the offer.
- You: How can we retain more users?
- Using Databrush: Checked your retention signals
- Using Databrush: Read Databrush’s suggestions
- Agent: Databrush has three suggestions. The strongest is an invite prompt: users who invite a teammate in week one retain 81%, against 64%.
- You: Make that change.
- Using Databrush: Accepted the suggestion
- Using its own tools: Added an invite step and opened PR #214
- Using Databrush: Marked it shipped in Databrush
- Agent: Shipped. Databrush will measure whether more new users invite a teammate, and whether they stay.
Give your agent its growth toolkit.
MCP server
https://app.databrush.ai/api/mcp- Claude
- ChatGPT and Codex
Then ask your agent: “Set up Databrush for my Supabase app.”
Users about to churn
See who’s leaving while there’s still time.
Databrush flags every customer who is cancelling, has gone quiet, or still pays without using your product, with the revenue at stake.
Databrush Needs attention: 41 paying · $1,027 MRR. 5 paying users who are cancelling or have gone quiet, each with their name and plan, as a stack of cards. An agent marks each one contacted in turn.
Retention signals
Find what makes customers stay.
Learn what customers who stay did in their first week, and steer every new user toward it.
Chart of 7 retention signals, most certain first, by the share of users still around at day 60. Created a project on day one: 78% retained vs. 62% of other users, 16 percentage points more. Proven. Invited someone in week one: 81% retained vs. 64% of other users, 17 percentage points more. Early signal. Completed a task 3+ times in the first 3 days: 80% retained vs. 68% of other users, 12 percentage points more. Early signal. Created a custom field in week one: 83% retained vs. 75% of other users, 8 percentage points more. Worth testing. Set a due date in week one: 77% retained vs. 70% of other users, 7 percentage points more. Worth testing. Commented on a task on two or more days in week one: 79% retained vs. 74% of other users, 5 percentage points more. Worth testing. Uploaded a file in week one: 76% retained vs. 71% of other users, 5 percentage points more. Early hint.
Product suggestions
Ship what the data backs.
Suggestions come from tested retention signals. Your agent can accept one, build it and mark it shipped, and Databrush measures whether it worked.
Suggestion: Prompt “Invite someone” in the first week. It moves from Draft to Accepted to Shipped, then shows Awaiting data, then Too early, then Readout · invited someone in week one 34% → 52% · changed.
Draft
Databrush drafts it from a signal
Accepted
Your agent accepts it
Shipped
Your agent ships the change
Readout
Databrush measures the result
Know your customers.
Grow at your pace.
Every feature, on every plan. Choose how many customers you monitor.
Growth
$79/month
Up to 2,500 customers
- Daily churn and retention analysis
- At-risk customer warnings
- Retention Signals and suggestions
- Access for your team and agents
Scale
$149/month
Up to 10,000 customers
- Daily churn and retention analysis
- At-risk customer warnings
- Retention Signals and suggestions
- Access for your team and agents
Every plan starts with a 14-day free trial. Cancel before it ends and you won’t be charged. Customers are counted across your workspaces, including dormant and churned customers still being monitored. No automatic overage charges.
More than 10,000 customers?
FAQ
Before you connect.
Why not just ask my agent to query my database?
You can, but every chat starts from scratch. Your agent has to work out what counts as a customer, an active user and churn, so the numbers shift from one conversation to the next. Databrush agrees those definitions with you once, analyzes your data against them every night, and remembers who you’ve contacted and what you’ve shipped. Every agent, in every chat, gets the same numbers, and none of them needs access to your database.
Do I have to define churn myself?
No. Databrush reads your database and proposes who your customers are, which actions count as activity, and when a customer has churned. You confirm or change them, and every number follows them from then on.
Do I need to add tracking to my app?
No. Databrush works from the data already in your Supabase database, so there’s no SDK to install and no events to track.
Can Databrush change my database?
No. Databrush only reads it. When your agent marks a customer as contacted or a suggestion as shipped, that’s saved in Databrush, never in your database.
Can my agent see customer names and emails?
Only if you allow it. Agents see an anonymous label for each customer until you turn on names and emails in Settings › MCP.