Supabase + Kite

Kite reads the structure of your database first, so you don't have to explain your schema to it. It lists the tables and views in your project, then reads the columns, data types, and relationships in the ones that matter. From that it works out where your signups live, which column holds each account's plan, and which table records the actions that show someone is getting value from your product.

Say your app has a profiles table, a workspaces table, and an events table. Kite notices that a user's first workspace gets created a few minutes after signup and that people who create one tend to stay. It proposes that moment as your activation step and asks you to confirm it in Slack. Once you agree, Kite saves the definition to your team's knowledge base, so every later question starts from the same meaning of "activated."

Kite measures your funnel by querying your tables directly. It counts new signups by day or week, the share who reached your activation step, and how many moved to a paid plan. It can split any of those by a column you already store, such as plan, country, or the signup source your app records.

That's how a vague question gets a specific answer. You ask how last week's launch went. Kite finds 380 signups against a usual 140, but only 22 percent created a workspace, compared with 35 percent most weeks. The launch brought in plenty of people, and the onboarding lost more of them than usual. Kite points to that step and proposes a change to the welcome email or the first screen's copy.

Kite can also run the same queries on a schedule. Every Monday it can post last week's signups, activation rate, and paid conversions in Slack, with the change from the week before and a short explanation of what moved.

Kite builds email audiences from product behavior rather than from a static list. It queries for the people who match a situation worth acting on, such as users who signed up in the last 60 days but never created a workspace, or accounts that were active last quarter and haven't signed in for a month. For each person it returns the name, email, plan, signup date, and the last thing they did.

Then Kite writes the campaign around where each group stopped. Trial users who never invited a teammate get a message about what the product does for a team. Users who set up a project and left get a note about the next step in that project. Kite sends the emails from your Kite team address or a connected email product once you approve them. A week later it runs the same query again to see how many people came back.

Kite turns your customer table into research on who actually buys. It pulls the work email domains of your paying or most active users, then looks up those companies with its built-in research sources. That gives it each company's industry, headcount, location, funding, and the technology it uses.

The pattern is usually clearer than anyone expected. Kite might find that 40 percent of your paid accounts are agencies with 20 to 100 people, even though your site speaks mostly to in-house teams. It records that finding in your ideal customer profile and uses it the next time it builds a lead list or writes landing page copy.

Kite can write its finished work into Supabase when that's where your team keeps it. It can create a new table through a tracked migration and insert rows into it, or update rows that already exist, using SQL it writes against your schema.

For example, Kite builds a prospect list of 200 people at companies that look like your best customers. Each row carries the person's name, role, company, professional email, LinkedIn profile, a fit score from one to five, and the reason for that score. Kite saves the list to a prospects table that your internal sales dashboard already reads. It can also add a fit score to each row in your accounts table, so your team sees Kite's research next to the data they already use. Kite brings the table, the columns, and a sample of the rows to Slack as part of its plan before it writes.