Florentine.ai

Florentine.ai is a cloud service for turning natural-language questions into queries against MongoDB and MySQL. It is best suited to teams that want controlled database access in support, analysis, or internal search workflows. Its schema-aware generation and integration options are useful, but its two-database scope and request caps make fit and scale important considerations.

Overview

Florentine connects to MongoDB Atlas, MongoDB Community Edition, and MySQL using connection strings. It analyzes collection and table schemas to generate MongoDB aggregation pipelines or SQL suited to the database. That focus is valuable when a team’s data already sits in those systems, but it rules Florentine out for teams needing a broader range of databases.

Key features

Query generation and review

Users can receive a raw query, query results, or a natural-language answer. Read-only queries can run automatically or wait for review. This gives teams a practical choice between faster self-service and oversight before reads execute; it is not a general approval mechanism for other types of database operations.

AI, agents, and search

Florentine supports OpenAI, Deepseek, Google, and Anthropic through a bring-your-own-key model, so teams can connect a provider they already use but must supply its key. Its MCP server and API have guides for Claude, ChatGPT, and Cursor, making it useful for teams incorporating database queries into agent workflows.

MongoDB vector search with automated embeddings and MySQL semantic re-ranking using a self-hosted embedding model extend the product beyond straightforward question-to-query translation. Secure data separation adds mandatory user-based query filters without depending on the language model to apply them, a meaningful safeguard for multi-user applications.

Privacy and security

The application and its MongoDB Atlas databases are hosted in EU Frankfurt regions, with standard contractual clauses for possible US transfers. Stated technical measures include TLS, AES-256 encryption at rest, MFA, and role-based access controls. Customers can request a DPA by email.

Pricing

The Free plan costs nothing, billed forever. It allows one database, five active collections or tables, and 500 requests per month, and includes MongoDB, MySQL, the MCP Server, and API. It is a sensible way to try the core workflow, but the monthly quota and narrow schema allowance are likely to constrain regular team use.

Pro costs €39 a month, billed per month. It keeps the one-database limit while raising the allowance to 25 active collections or tables and 2,500 requests per month. That is the practical step up for a single-database team whose free quota or schema cap is too tight.

Enterprise costs €149 a month, billed per month, and provides five databases, 50 active collections or tables, and 10,000 requests per month. It is the best fit of the three for teams querying several databases or handling heavier use, though the published limits still cap monthly demand. All three plans include MongoDB, MySQL, MCP Server, and API. A free trial is offered.

Platforms

Florentine is a cloud deployment with web and API platforms. It supports MongoDB and MySQL; its documented connection options are MongoDB Atlas, MongoDB Community Edition, and MySQL connection strings.

Who it's for

Support teams, business analysts, and teams building internal search or recommendation engines are among the intended users. Florentine is a strong fit when those workflows need natural-language access to MongoDB or MySQL, structured query generation, and the option to review read-only queries. It is a weaker fit for organizations with other database types or workloads that exceed the plan quotas.

Pros and cons

Pros

Schema analysis informs query generation, which is useful when users cannot or should not write SQL or aggregation pipelines themselves.

Users can choose between raw queries, results, and natural-language answers, and can require review before read-only queries run.

MCP and API support, several BYO-key LLM providers, and agent guides suit teams integrating database access into existing workflows.

Mandatory user-based filters and stated encryption, MFA, and role-based controls address important concerns for shared database access.

Cons

Support is limited to MongoDB and MySQL, so teams on other database systems need another tool.

The Free plan's 500 monthly requests and five active collections or tables limit how far teams can take it without upgrading.

Pro still covers only one database; access to five databases starts with the €149 a month Enterprise plan.

BYO-key integrations require teams to provide their own LLM provider key.

Alternatives

Consider AI Database Assistants for a broader category comparison. Insight O' Mate is worth considering if a desktop app across Linux, macOS, and Windows is preferable: its Free plan allows 20 database analyses per day for one founder and one database, while Pro costs $29 a month for unlimited queries, priority queue, and advanced export.

Outerbase AI may suit teams wanting desktop or self-hosted options as well as web and API; its Free plan allows up to five users but is limited to one base, 10 EZQL queries per month, three saved queries, one dashboard, and transactional databases. Snowflake CoCo is another option to compare if its paid offering and platform mix are a better fit.

Wren AI is an alternative for teams considering an open-source context engine: its OSS offering is free and aimed at individual developers using CLI and MCP without a UI. BlazeSQL is an alternative with paid plans and a free trial; its Pro plan costs $150 a month and includes an AI Data Analyst, graphs and dashboards, a desktop private database connection, and a knowledge base.

DataLine is a free, open-source project under GPL-3.0. DataZen is a free desktop option for Linux, macOS, and Windows, with GPLv3, no account, and BYO AI provider. YourQL is a free desktop application described as a work in progress.

Verdict

Choose Florentine.ai if your team works with MongoDB or MySQL and needs natural-language querying with schema-aware generation, integration flexibility, and control over read-only execution. The clearest reason to look elsewhere is database coverage: its focus on two database families, combined with monthly request and collection limits, may not suit broader or higher-volume workloads.