01
Trace — see what your database is doing
Arm the activity monitor and it gains a memory: record the last minutes of everything running, then scrub the timeline back to the moment things slowed down. Queries that finish in under a second slip past the live view, so a Top tab reads pg_stat_statements ranked by total time. See a live tree of who's blocking whom and kill any session from the graph — and Lily's local AI explains the root cause of a slow query in plain language, entirely offline.
02
Speed up a slow query in one click
Point Lily at a slow query and it proposes one index, applies it with CREATE INDEX CONCURRENTLY so writes never block, and shows the estimated cost before and after so you can see it worked. The plan and the query stay on your machine — the AI only ever sees the redacted statement, never a single row of data.
03
Understand any database — as a living map
Lily's AI explains the business behind your schema — mapping cryptic legacy names (tb_usr → Customers), detecting the domain, and drawing it as a living constellation where business sub-systems become clusters you click to explore. The map is written in your language, and it remembers what it documented: change the schema and it re-documents only what changed.
04
Recipes for every entity
For each business entity Lily writes the everyday queries you actually run — "recent orders for a customer", "find by email" — parameterized and ready to open in the editor. Read-only and EXPLAIN-validated before they're ever offered; the core entities pre-warm in the background so they're waiting when you look.
05
Change your schema with AI
Describe a change in plain language — "add soft-delete", "make email unique" — and Lily generates the forward DDL plus its exact rollback, grounded in your real columns, keys and indexes. Nothing runs until you open it in the editor and confirm; destructive changes are flagged and pass the same write guardrails as any statement.
06
Semantic schema search
Ask "where do we store refunds?" and Lily ranks the real tables and columns that hold the concept — each with a one-line reason — searching your schema's meaning, not just its names. Click a result to jump straight to the table. It searches metadata only, never your rows.
07
See the blast radius before you write
Before any UPDATE, DELETE, TRUNCATE, or DROP, Lily estimates how many rows it will touch and which tables it reaches by foreign key — without ever running the write. The riskiest ones (no WHERE, TRUNCATE, DROP) make you type the table name to confirm.
08
PII scan, redacted by default
Lily flags columns holding personal data — CPF/CNPJ, cards, emails, phones — using local format fingerprints, and automatically redacts them before any AI feature ever samples a row. Your data never leaves your machine for the classification.
09
Eight engines, one app
PostgreSQL, MySQL/MariaDB, SQLite, SQL Server, Redis, MongoDB, Trino, and ClickHouse — each with a UI tuned to its data model. SSH tunnels support agent, key-passphrase, and password auth for corporate bastions.
10
Results you can work
Sort any column (numbers, dates, and text handled right) and filter with real operators — no re-query. Freeze the first columns while the rest scroll, click a header for instant count/sum/avg/min/max, chart the result, or let local AI summarize the patterns.
11
One-click Health Check
Audit any database with ready-to-run fixes plus an AI architecture review — schema, index, and security rules (RLS, PUBLIC grants, unused indexes) for relational, TTLs and big keys for Redis, indexes and document size for MongoDB. Jump from a value to the row it references, kill stuck queries, and render any plan as a visual EXPLAIN cost-bar tree.
