Delphina is an enterprise data platform built as an AI-managed context layer for business analytics. It connects to data warehouses and internal documentation to build a canonical knowledge base of a company’s metrics, definitions, and rules. Users can then query that data in plain English, generate automated workflows, and build interactive data applications grounded in verified business logic.
The platform targets a specific problem: AI agents that answer confidently but incorrectly because they lack company context. Legacy BI tools solve this but require heavy analyst time, and direct database queries to an LLM miss the unwritten rules analysts hold in their heads. Delphina tries to close that gap by systematizing institutional knowledge so AI agents can produce accurate, checkable answers.
Delphina works through four stages: Ingest, Refine, Validate, and Evolve. It maps a company’s data and documentation, then runs answers through built-in evaluations and a critic agent that pressure-tests the reasoning before it reaches a user. The system updates as schemas change and priorities shift. Access happens through Delphina’s own Analytics Agent, Slack, or external agents like Claude and Cursor connected via the Model Context Protocol.
The core differentiator is that Delphina positions itself as a context layer rather than a standalone chatbot. Queries run in isolated, sandboxed environments with no internet access, and full SQL lineage supports org-wide observability. Its headless MCP server lets a company plug its existing AI tools into a verified layer of internal context, rather than replacing those tools outright.
Pricing
Delphina offers a free Community plan (no credit card required) limited to one user, public datasets, and file uploads, with no data warehouse connections. The Pro plan costs $250 per month, includes a 14-day free trial, and comes with $300 per month in credits covering compute and inference; usage beyond that is billed monthly, and Pro is capped at 10 users and one data warehouse connection. Enterprise pricing is custom, based on seats or consumption, and is required for single-tenant deployment, AWS VPC hosting, OIDC/SAML SSO, embedded analytics, and audit logs.
* Disclaimer: Please note that pricing information may not be up to date. For the most accurate and current pricing details, refer to the official website.
Key Features
- ✓
AI-managed context layer captures business definitions and unwritten rules
- ✓
Built-in critic agent pressure-tests reasoning before answers are delivered
- ✓
Natural language Analytics Agent generates SQL, charts, and written analysis
- ✓
Proactive agents monitor data for anomalies and metric shifts
- ✓
Automated scheduled workflows for weekly business reviews and KPI digests
- ✓
Headless MCP server connects Claude, Cursor, and ChatGPT to verified context
Use Cases
Automated Weekly Business Reviews
Data and sales teams schedule recurring metric digests instead of pulling charts manually each week. Delphina runs the workflow autonomously and delivers narrative findings and visuals directly to stakeholders.
Ad-Hoc Slack Data Queries
Non-technical staff ask business questions in plain English directly inside a Slack channel. Delphina pulls the relevant context, writes the SQL, and returns charts without routing the request through the data team.
Interactive Dashboard Generation
Operations teams need dashboards tied to newly tracked KPIs. Users prompt Delphina to build an interactive data app, then iterate on its styling and analysis through further chat.
Deep Data Science Investigations
Strategic teams need to understand complex patterns such as cohort retention or ARPU shifts. The Deep Research agent tests hypotheses, runs parallel investigations, and trains predictive models like churn-by-cohort into one report.
Proactive Anomaly Detection
Growth teams need immediate notice when a metric spikes or drops unexpectedly. Delphina’s always-on agents watch the data continuously and alert the team when anomalies or milestones occur.
Strengths & Weaknesses
Strengths
A built-in critic agent pressure-tests reasoning before answers reach the user.
The validated context layer works with external tools like Claude and Cursor via MCP.
Slack integration returns real-time data queries inside existing team channels.
Queries run in an isolated sandbox with read-only access and no internet egress.
The system adapts automatically as schemas update and business priorities shift.
Weaknesses
The free Community plan blocks warehouse connections, limiting testing to public or uploaded data.
The Pro plan caps access at 10 users and a single data warehouse connection.
SSO, audit logs, and embedded analytics are reserved for the custom-priced Enterprise plan.
Consumption-based billing on Pro could make monthly costs unpredictable for heavy query usage.
Who Is This For?
Data analysts and engineers who want routine SQL and dashboard requests automated so they can focus on deeper infrastructure work.
Business executives who want to query company data directly and get validated answers without waiting on a reporting queue.
Operations and growth teams that need always-on monitoring for anomalies, milestones, and metric shifts.
Enterprise IT and security teams that require SOC 2 Type II and HIPAA compliance plus single-tenant or AWS VPC deployment.
Frequently Asked Questions
How much does Delphina cost to get started?
The Community plan is free forever with no credit card required. The Pro plan starts at $250 per month plus consumption, with a 14-day free trial before billing begins.
How does consumption billing work on the Pro plan?
Pro includes $300 per month in credits covering compute and inference used by queries and workflows. Usage beyond that credit amount is billed month-to-month.
Does Delphina use my company’s data to train its models?
No. Delphina states it does not sell customer data or use it to train shared models, and customer data is isolated and encrypted.
How secure is my database connection to Delphina?
Delphina uses role-scoped, read-only warehouse credentials with no write paths. Queries execute in an isolated sandbox with no internet access.
Can Delphina be deployed on our own infrastructure?
Yes, Enterprise customers can deploy Delphina inside their own AWS VPC so raw data stays within their security perimeter.
What compliance certifications does Delphina hold?
Delphina is SOC 2 Type II certified and HIPAA compliant, covering the rigorous standards required for handling sensitive and medical data.
Which AI models and tools can connect to Delphina?
Delphina supports frontier models including Claude, GPT, and Gemini, and connects to external agents like Claude and Cursor through its headless MCP server.
What is the difference between the Community and Pro plans?
Community is limited to one user, public datasets, and file uploads with no warehouse connection. Pro supports up to 10 users and one data warehouse.
Do I need a credit card to try Delphina?
No. The Community plan requires no credit card, and the 14-day Pro trial does not require payment details until you continue past day 14.
What data sources and knowledge tools does Delphina connect to?
Delphina connects to warehouses like Snowflake, Databricks, BigQuery, Redshift, PostgreSQL, and Supabase, plus knowledge sources including GitHub, Notion, Google Workspace, dbt, Looker, and Tableau.
Delphina connects to data warehouses including Snowflake, Databricks, BigQuery, Redshift, PostgreSQL, and Supabase using read-only credentials. Knowledge sources include GitHub, Notion, Google Workspace, dbt, Looker, and Tableau. Claude, ChatGPT, and Cursor can connect as bring-your-own MCP clients, and Slack provides a chat interface for queries and alerts. Authentication supports Google SSO, SAML SSO, and OAuth 2.0.