AI for Database

Connect a PostgreSQL, MySQL, MongoDB, SQL Server, SQLite (beta), or Google Sheets source and ask questions in plain English; the product generates SQL, runs it read-only, and returns a table, chart, or number, with a "View SQL" option to inspect or edit the generated query (docs: https://www.aifordatabase.com/docs/first-query/). Answers can be pinned to a dashboard or turned into a scheduled workflow — a trigger, a query, a condition, and an action (email, Slack, or webhook) — described in plain sentences rather than a node-based builder (https://www.aifordatabase.com/docs/create-workflow/). New workflows stay as unpublished drafts with a dry-run/test step before anything live fires.

Beyond the chat UI, AI for Database publishes a documented OpenAPI 3.1 spec (6,800+ lines, https://app.aifordatabase.com/api/v1/openapi.json) covering connections, chat, dashboards, workflows, metrics, saved queries, and scoped API keys. Three features here are genuinely more than a wrapper around an LLM call: a schema-introspection endpoint that returns structured table/column/relationship metadata instead of making an agent guess; a semantic layer where a metric like "revenue" is defined once as SQL and every agent query then uses that same definition; and a query-approval mechanism that lets you require human sign-off for queries touching PII columns or cross-database joins before they execute (https://www.aifordatabase.com/agents/). Combined with read-only-by-default connections and destination-bound, encrypted credentials for any outbound action, this is a more considered answer to "give my agent a database" than a raw MCP-to-database bridge, which the vendor's own comparison table (correctly) says can DROP TABLE with no audit trail.

The main pricing page (https://www.aifordatabase.com/pricing/, verified 2026-09-15) lists Pro at $20/user/month and Team at $50/user/month, billed monthly, with 20% off annual ($192/user/yr Pro, $480/user/yr Team). This isn't a seat license in the usual sense: each payment adds that dollar amount as non-expiring per-user AI-usage credit, and usage pauses when the balance runs out — an unusual model that rewards light users and penalizes anyone who blows through credits mid-cycle. Top-ups come in $10/$25/$50/$100 increments. Enterprise is quote-based, as is the self-hosted option ("one-time setup fee plus annual license," no self-hosted number published: https://www.aifordatabase.com/self-hosted/). Separately, the agent/API page (https://www.aifordatabase.com/agents/) shows a different ladder entirely — a $0 free tier (3 workflows, 60 req/min) and a $19/month Pro tier (unlimited workflows, 300 req/min) — that doesn't match the $20/$50 credit-based structure on the pricing page. One of these two pages is stale; a buyer evaluating the API specifically should confirm the live number with sales rather than trusting either page in isolation.

The security page (https://www.aifordatabase.com/security/) states AES-256 encryption at rest, TLS 1.3 in transit, read-only-by-default access, and full audit logging — specific and plausible claims. But its compliance badges use deliberately soft language: "SOC 2 — Enterprise Ready," "GDPR — Designed For," "HIPAA — Enterprise Available," "ISO 27001 — Aligned." None of these is a claim of certification, and the page links no SOC 2 report, DPA, or subprocessor list. The Terms of Service (https://www.aifordatabase.com/terms/) reinforce this: the company caps its total liability at the greater of your prior 12 months' fees or $50, explicitly disclaims accuracy of AI-generated SQL, and puts the burden on the customer to decide whether the service meets HIPAA/GDPR/PCI-DSS requirements before connecting regulated data. For a team planning to route production database credentials through this, that combination — real encryption claims plus no actual compliance artifact — is the kind of gap worth asking about directly rather than assuming from the badge row.

The domain was registered 2025-09-25 (RDAP: https://rdap.verisign.com/com/v1/domain/aifordatabase.com), making the company under a year old as of this listing. No independent reviews were found: G2, Capterra, Trustpilot, and Product Hunt all returned access-blocked responses (403) rather than a page with reviews, and a Bing News search for "aifordatabase" returned zero results. The only customer testimonial on the site (a VP of Sales quote on the sales use-case page) is attributed to "Rachel M., B2B SaaS Company" with no company name given. There is no public GitHub organization for the project (api.github.com/orgs/aifordatabase returns 404), and the Docker image named in the self-hosting docs (aifordatabase/ai-database) is not found on Docker Hub, so self-hosting appears to require contacting sales for access rather than a public pull. None of this means the product doesn't work — the documentation and OpenAPI spec both show real engineering depth — but there is no third-party signal yet to corroborate vendor claims about scale or reliability.