Helios vs PromptQL
When to use Helios or PromptQL
PromptQL answers questions over governed data with shared threads. Helios runs tasks that act on your tools. Many teams need both.
5,000 free credits to start.
PromptQL is Hasura's multiplayer AI agent. Its own site says it has "shared threads and a shared brain". It plans in a DSL, compiles the plan to Python and SQL, and runs it outside the model, so answers over data are repeatable and cited. Helios is a different category. Each run gets a computer, reaches tools through their full APIs, and starts on a schedule, a webhook, or an event. This page explains where each one fits.
At a glance
Side by side
Short facts for each product. The PromptQL column comes from public pages on the date shown at the bottom of this page.
| Topic | Helios | PromptQL |
|---|---|---|
| Computer per run | A fresh cloud sandbox for every run and chat, with shell, filesystem, and code execution. | Generated Python and SQL run in PromptQL's runtime outside the model. |
| Tools and APIs | 21 integrations reached through their full OpenAPI or GraphQL surface, plus Postgres, MySQL, and ClickHouse connections with an SSH tunnel option. Remote MCP servers can be attached. | Data sources and apps modeled in Hasura DDN. Cross-source joins and scoped access. |
| Triggers | Manual, schedule, signed webhook, and integration events from GitHub, Slack, and Plain. | Chat threads first. Automations are exposed as HTTP APIs. |
| Human review | Agents propose edits to shared Library files. A person approves or discards each file, with conflict detection and history. | Every answer carries citations and an audit trail. Answers are reviewed in shared threads.Library review depends on PRs 600 to 603. |
| Permissions | Cedar roles on every resource, viewer to admin. Share with a person, group, service account, or the whole organization, with expiry. | Scoped access on data sources, SSO on the enterprise plan. |
| CLI | helios CLI and terminal app with the same roles and sharing as the dashboard, JSON output when piped. | Hasura DDN CLI for metadata and project setup.CLI depends on PRs 619 and 624. |
| Pricing model | Credits. 5,000 free, then $0.002 per credit. No seats. | Per OLU. Starter $0.14 introductory, $0.20 standard. Pay as you go. Enterprise is custom. |
| Organization spend limits | Monthly spending limit per organization. Triggers are blocked when credits run out. | Free credit per project and per teammate. Pay as you go beyond that. |
| Memory and context | Shared Library files mounted into the sandbox. Changes are reviewed before they merge. | A self-seeding wiki built from connected sources, shared across threads.Library depends on PRs 600 to 603. |
| Best fit | Teams that want agents to do work on a trigger without a data engineering step first. | Data and analytics teams with engineering capacity for Hasura DDN metadata. |
Where PromptQL is strong
What PromptQL does well
- Deterministic plan then execute. The query plan compiles to Python and SQL and runs outside the model, so the same question gives the same answer with citations.
- Cross-database joins through Hasura DDN, with scoped access and a self-seeding team wiki built from Slack, Docs, Snowflake, PostHog, and Salesforce.
- Bring your own model, plus VPC and BYOC deployment on the enterprise plan.
Where Helios is different
How Helios works instead
- Helios runs any task, including questions over data. An agent can read a Stripe dispute, open a GitHub issue, and post to Slack in one run because it has a shell and the full API of each tool.
- Helios needs no metadata modeling to start. Connect an integration or paste a database connection string and describe the task. PromptQL's docs walk through DDN metadata setup first.
- Helios prices in credits with an organization spending limit. PromptQL prices per OLU, and the rate depends on the model used.
Pricing
What each one charges
Public list prices, checked on 7 September 2026. Prices change, so confirm on each vendor's page before you decide.
Helios
5,000 free credits, then $0.002 per credit. Usage based, no seats, with an organization spending limit and optional auto refill.
Helios pricing →PromptQL
As of September 7, 2026. Starter is $0.14 per OLU introductory, $0.20 standard, with up to $100 free credit per project plus $70 per teammate. Pay as you go. Enterprise is custom with VPC, BYOC, SSO, and bring your own LLM.
PromptQL pricing page →Checked on 7 September 2026
Using both
Helios and PromptQL together
Use PromptQL when the question is about governed data and the answer must be repeatable and cited. Use Helios when the work is a task that touches several tools or needs a computer. The two fit together. A Helios workflow can call a PromptQL automation over its HTTP API as one step, then act on the result.
FAQ
Helios and PromptQL, common questions
Run your first task.
5,000 free credits. Set an organization spending limit before your first workflow runs.
PromptQL information is from public pages on 7 September 2026. Marks belong to their owners. Helios is not affiliated with PromptQL. Tell us if something is out of date at [email protected].