PromptQL
PromptQL is a data access agent from Hasura that enables reliable AI interaction with structured, unstructured, or API data.
Publisher review
PromptQL is a data access agent from Hasura that enables reliable AI interaction with structured, unstructured, or API data. It is designed for developers and enterprises building AI assistants, copilots, or autonomous agents that need to query and manipulate business data using natural language with high accuracy. The platform is part of Hasura's Data Delivery Network (DDN), providing unified, real-time GraphQL API access across databases, REST servers, GraphQL endpoints, and third-party APIs like Stripe or Salesforce. PromptQL targets teams that require explainable, repeatable, and secure AI-driven data workflows, particularly in sales & marketing, healthcare, and financial services.
PromptQL uses a novel agent approach that composes tool calls and LLM tasks into dynamic query plans. It autonomously creates these plans, runs computations on data, intelligently retries failed queries, and modifies plans on the fly to reduce hallucinations. Each interaction may involve the creation and execution of one or more PromptQL programs; a simple data interaction might consume 2,000 tokens and execute 1 program, while a complex task could execute 10 programs and consume 10,000 tokens. The system supports semantic metadata with entity relationships and fine-grained entitlements for secure data handling. Developers interact via a Program API that allows invoking Python functions for reading/writing/searching data or interacting with external APIs, and can create memory artifacts from conversations to maintain context.
PromptQL competes directly with LangChain, Vertex AI Agent Builder, Azure Copilot Studio, AWS Bedrock AgentCore, Haystack, LlamaIndex, Flowise, Superagent, CrewAI, n8n, Zapier, and others. Its key differentiator is the emphasis on high-trust, accurate AI data access through dynamic query planning and built-in security features, rather than just a general-purpose agent framework. However, it is less mature in the open-source ecosystem compared to LangChain or LlamaIndex, and its pricing model is consumption-based rather than subscription or free-tier, which may deter smaller teams.
The honest trade-offs: PromptQL requires coding knowledge and expertise, making it less user-friendly than low-code alternatives like Flowise or Zapier. Performance issues can arise with extensive applications, especially when handling complex multi-step queries across many data sources. Its consumption-based pricing (programs at $0.042 each plus LLM token costs via Anthropic) can become expensive for high-volume usage, and there is no free tier or trial beyond the initial prepaid consumption dollars. The platform is tightly coupled to Hasura's DDN, so teams not already using Hasura may face integration overhead.
How it works
-
Dynamic query planning
Autonomously creates and modifies query plans on the fly, retrying failed queries and running computations to reduce hallucinations.
-
Unified data access
Connects to databases, REST servers, GraphQL endpoints, and third-party APIs like Stripe or Salesforce via Hasura's Data Delivery Network.
-
Semantic metadata support
Supports entity relationships and fine-grained entitlements for secure, context-aware data handling in enterprise environments.
-
Program API integration
Allows developers to invoke Python functions for reading/writing/searching data or interacting with external APIs seamlessly.
-
Memory artifacts
Creates memory artifacts from conversations or queries to maintain context over time for ongoing AI interactions.
-
Consumption-based billing
Pricing uses prepaid consumption dollars deducted per program ($0.042 each) and LLM tokens via Anthropic, with a usage dashboard.
-
Cross-team collaboration
Supports role-based access and a wiki for curating context, enabling team AI with organized information sharing.
Strengths and trade-offs
Strengths
- Provides high-trust LLM interaction with dynamic query plans that autonomously retry failed queries and modify plans on the fly to reduce hallucinations.
- Offers unified data access through Hasura's Data Delivery Network, connecting to databases, REST servers, GraphQL endpoints, and third-party APIs like Stripe or Salesforce in real time.
- Includes semantic metadata with entity relationships and fine-grained entitlements, ensuring secure handling of sensitive enterprise information.
- Demonstrated use cases in sales & marketing, healthcare, and financial services, including anti-money laundering transaction pattern analysis.
Trade-offs
- Requires coding knowledge and expertise, making it less accessible for non-developers compared to low-code alternatives like Flowise or Zapier.
- Performance issues can arise with extensive applications, especially when handling complex multi-step queries across many data sources.
- Consumption-based pricing (programs at $0.042 each plus LLM token costs via Anthropic) can become expensive for high-volume usage without a free tier.
- Tightly coupled to Hasura's Data Delivery Network, so teams not already using Hasura may face integration overhead and vendor lock-in.
Pricing context
Prepaid consumption system: $0.042 per program plus LLM token costs based on Anthropic pricing (input/output tokens). Customers purchase consumption dollars deducted via a usage dashboard. No free tier or trial beyond initial prepaid amount.
Getting started with PromptQL
-
Sign up for PromptQL
Go to the Hasura DDN console and create an account. Purchase prepaid consumption dollars to fund your usage. No free tier is available, so you must add payment details to get started.
-
Connect your data sources
In the Hasura DDN console, add your databases, REST servers, GraphQL endpoints, or third-party APIs like Stripe or Salesforce. PromptQL uses the Data Delivery Network to unify access to these sources in real time.
-
Define semantic metadata
Configure entity relationships and fine-grained entitlements for your data sources. This step ensures PromptQL understands your data model and enforces security policies during query planning.
-
Write your first program
Use the Program API to invoke Python functions for reading, writing, or searching data. Start with a simple query, such as fetching a customer record, to verify that PromptQL can access and return results accurately.
-
Deploy and monitor usage
Integrate the PromptQL agent into your AI assistant or copilot application. Monitor the usage dashboard to track program executions and LLM token costs, adjusting your consumption dollars as needed for production workloads.
Frequently Asked Questions
What is PromptQL and how does it work?
PromptQL is a data access agent from Hasura that enables reliable AI interaction with structured, unstructured, or API data. It uses dynamic query planning to autonomously create and modify query plans, retrying failed queries and running computations to reduce hallucinations.
How does PromptQL pricing work?
PromptQL uses a prepaid consumption system. Each program costs $0.042, plus LLM token costs based on Anthropic pricing. Customers purchase consumption dollars deducted via a usage dashboard. There is no free tier or trial beyond the initial prepaid amount.
What data sources can PromptQL connect to?
PromptQL connects to databases, REST servers, GraphQL endpoints, and third-party APIs like Stripe or Salesforce via Hasura's Data Delivery Network. This provides unified, real-time GraphQL API access across all these sources for AI-driven data workflows.
How does PromptQL compare to LangChain or LlamaIndex?
PromptQL competes with LangChain, LlamaIndex, and others but focuses on high-trust, accurate AI data access through dynamic query planning and built-in security. It is less mature in the open-source ecosystem and requires coding expertise, unlike some low-code alternatives.
What are the main weaknesses of PromptQL?
PromptQL requires coding knowledge, making it less accessible for non-developers. Performance issues can arise with complex multi-step queries across many data sources. Its consumption-based pricing can become expensive for high-volume usage, and it is tightly coupled to Hasura's DDN.
What industries use PromptQL for AI data access?
PromptQL is used in sales & marketing, healthcare, and financial services. For example, it supports anti-money laundering transaction pattern analysis. It targets teams needing explainable, repeatable, and secure AI-driven data workflows with high accuracy.
Alternatives
How PromptQL compares
Direct head-to-head against 3 competitors. Picked by 7wData.
PromptQL
- Pricing
- Prepaid consumption system: $0.042 per program plus LLM token costs based on Anthropic pricing (input/output tokens). Customers purchase consumption dollars deducted via a usage dashboard. No free tier or trial beyond initial prepaid amount.
- Target
- PromptQL is a data access agent from Hasura that enables reliable AI interaction with structured, unstructured, or API data.
- Strength
- Provides high-trust LLM interaction with dynamic query plans that autonomously retry failed queries and modify plans on the fly to reduce hallucinations.
- Watch for
- Requires coding knowledge and expertise, making it less accessible for non-developers compared to low-code alternatives like Flowise or Zapier.
CamelAI
- Pricing
- Custom/Contact sales; no free tier or self-service option
- Target
- Business teams needing fast setup and embeddable AI query tools
- Deployment
- Connects to warehouse in minutes
- Strength
- Learns business context automatically from usage, no semantic layer setup
- Watch for
- Pricing not publicly listed; may require sales engagement for access
Julius
- Pricing
- From $37/month billed annually
- Target
- Business teams asking data questions in plain English
- Deployment
- No setup required; natural language interface
- Strength
- Natural language interface with no upfront configuration needed
- Watch for
- Less deterministic than PromptQL for complex multi-step queries
Hex
- Pricing
- From $36/editor/month billed annually
- Target
- Data teams collaborating on SQL and Python analysis
- Deployment
- Cloud-based notebook with app builder
- Strength
- Real-time collaboration with notebook and app builder for sharing insights
- Watch for
- Requires SQL/Python skills; not a pure natural language tool
User reviews
No user reviews yet. Be the first to write one.
Sources
Reporting on this tool draws on these publicly available sources.