Starter
For small teams shipping their first AI pipelines.
- Up to 10 automated pipelines
- 100k events / month
- 7-day run history
- Community support
New · Helix 4 — autonomous pipelines
Helix is the AI runtime for production data — orchestrate pipelines, govern models, and observe every row with a platform engineered for the next decade of automation.
Trusted by data teams at
The platform
Seven primitives. Infinite pipelines. Helix collapses your data stack into a single, observable surface.
Detect, retry, and reroute failures in real time. Zero pager fatigue.
OpenAI, Anthropic, Mistral or your private weights — one runtime.
Lineage, cost and latency on every row, every run.
PII redaction, RBAC, residency controls and a complete audit ledger.
Warehouses, lakes, SaaS APIs, webhooks — bidirectional, idempotent.
Vector + lexical hybrid retrieval over your entire data surface.
Predictable cost curves to 100M+ events per day, region by region.
Pipeline uptime SLA
P50 step latency
Rows processed daily
Pricing
Every price below is computed live from a single configuration matrix — base rate, FX, regional tariff and billing cycle.
For small teams shipping their first AI pipelines.
Production workloads with team collaboration.
Enterprise observability and dedicated infrastructure.
Prices reflect a flat 20% annual discount and regional tariff adjustments. Taxes calculated at checkout.
Customers
“We replaced four orchestration tools with Helix in a quarter. Our data SLAs went from best-effort to contractually enforceable.
“The first pipeline platform that actually feels designed. Engineers shipped a fraud model end-to-end in nine days.
“Cost observability per row, per model, per region. It's the dashboard our CFO asked for two years ago.
FAQ
Helix unifies orchestration, model serving and governance into a single runtime. You write declarative pipelines; we handle retries, lineage, cost attribution and policy enforcement automatically.
Yes. Helix ships connectors for OpenAI, Anthropic, Mistral, AWS Bedrock and self-hosted vLLM endpoints. Swap providers per-step without rewriting pipelines.
Enterprise contracts move to committed throughput with regional discounts. Talk to sales for a custom matrix — every line item maps to the same engine you see on this page.
You pick the region per workspace: us-east, eu-west, ap-south, or your own VPC via PrivateLink. Data never leaves the chosen residency boundary.
Every plan starts with a 14-day trial including the full Growth feature set. No card required to start.
Fourteen-day trial. No credit card. Full Growth feature set, including SOC 2 audit logs.