See every step. Control every run.
Open-source execution control for data pipelines — the full product, self-hosted. Build, preview, run, and debug locally with a visual canvas, 90+ connectors, SQL transforms, AI assistance, and Steward reliability checks. Built for SQL-first data engineers, analysts, and small teams; when you need team governance, F-Pulse+ is the natural upgrade.
| id | name | status | last_active | |
|---|---|---|---|---|
| 1 | Sarah Chen | s.chen@acme.co | active | 2h ago |
| 2 | Marcus Reid | m.reid@corp.io | active | 5m ago |
| 3 | Anika Patel | a.patel@tech.co | idle | 1d ago |
Full visibility. Full control.
Everything SQL-first data engineers and analysts need to design, run, and ship a pipeline — visual builder, live data preview, 90+ connectors, scheduling, and an AI assistant. Open source, all free.
Visual Pipeline Builder
Drag-and-drop workflow canvas. Connect sources, transforms, and destinations with a schema-aware expression editor.
90+ Connectors, Honest Maturity
Databases, warehouses, streaming, 40+ SaaS APIs, and vector DBs. Most are SDK-validated, a handful integration-verified against live endpoints, none yet production-certified — we label maturity honestly. Need one we don't ship? Paste OpenAPI → working connector in 90 seconds.
SQL Transforms
Full expression editor with schema awareness, auto-complete, and AI-assisted SQL generation.
Smart Scheduling
Cron-based with dependency awareness. Backfill, retry policies, and SLA monitoring out of the box.
Live Monitoring
Live execution status, run history, data quality metrics, and alerting.
Built-in AI Assistant
Local-first via Ollama, cloud-optional. Generates SQL, explains failures, drafts pipelines from natural language.
Beyond logs. The layer above pipeline execution.
Other ETL tools tell you a pipeline failed. F-Pulse Steward notices that two of your pipelines are reading the same source into different warehouses — and remembers your fix the next time someone re-introduces the pattern. Read-only by construction. Ships in OSS, not paywalled.
Detects across pipelines
Duplicate sources, duplicate pipelines, orphaned tables, schema drift, join explosions — workspace-level patterns no per-job monitor catches. < 50ms per scan.
Learns from your fixes
Memory Layer turns operator dismissals + resolutions into typed, human-approved lessons. Propose → approve → revalidate workflow.
Escalates what you ignore
Persistent-occurrence counter promotes P3 → P2 → P1 across scans. What you keep ignoring gets louder, not quieter.
Read-only by construction
Never mutates pipelines. Never runs fixes. Never changes execution policy. Advisor, not actor — enforced at the runtime layer.
Detection is pure code — no LLM in the decision path, no hallucinated findings. And it learns: a Memory Layer turns your fixes into lessons and escalates what you keep ignoring. We'd rather show "0 live" on a level than soft-label everything "coming soon."
Production execution control with F-Pulse+
When one laptop becomes a team, F-Pulse+ adds the controls production demands — without changing how you build. Govern who runs what, where, and when: vault-backed AES-256 credentials, 5-tier RBAC, two-gate approvals (sandbox → deploy), drift detection, lineage, OIDC/SAML SSO, audit retention, and 4-hour SLA support.
Stop guessing. See everything.
Every pipeline run is fully visible. Preview data at each node, control retries, approve promotions.
Live Node Preview
Click any node mid-run. See the actual data flowing through it — Table, Schema, or JSON view.
Execution Queue
Worker pools with P1-P5 priority. Control which pipelines run first when resources are constrained.
Retry with Visibility
Failed step? See exactly what data caused the failure. One-click retry from the failed node, not the beginning.
Approval Gates
Require approval before promoting to production. Dev builds, lead reviews, admin approves. (F-Pulse+)
From idea to production in three steps
Design
Drag sources, transforms, and destinations onto the canvas. Use templates to start fast.
Run Locally
Run pipelines on your laptop, inspect outputs, debug with real data. Schedule, monitor, ship.
Add a Team
Move to F-Pulse+ when you need governance. Encrypted credentials, 5-tier RBAC, audit trail, and security controls.
What developers build with F-Pulse
Real workflows running on F-Pulse today, built by solo developers on their own machines.
Postgres → Snowflake CDC
Stream changes from operational Postgres into a Snowflake warehouse with schema drift handling and automatic backfill on first run.
API → Lakehouse Ingest
Pull from Stripe, HubSpot, Shopify, and other SaaS APIs into S3 or MinIO as Parquet, partitioned by ingest date.
Bronze → Silver → Gold ETL
Medallion architecture template: raw landing, cleaned/typed silver, business-ready gold marts. Run on a schedule or trigger.
Reverse ETL to SaaS
Push warehouse-modeled customer attributes back into Salesforce, HubSpot, or Intercom so sales and marketing see the same numbers.
Data Quality Gates
Run row-count checks, null-rate thresholds, and uniqueness assertions between pipeline stages. Fail loudly, retry safely.
Scheduled Reports
Build a daily report query, render to CSV/Parquet, drop in S3, and notify Slack — without writing a single line of orchestrator code.
Common questions
Is F-Pulse really free?
Yes. F-Pulse v1.0.0 is open source under Apache 2.0. Self-host it, fork it, modify it, ship it. Telemetry is off by default (opt-in only), no usage caps, no feature gating. F-Pulse+ is a separate commercial tier for teams that adds workspace RBAC, two-gate approvals, vault-backed credentials, drift detection, lineage, SSO, and 4-hour SLA support. F-Pulse alone is fully usable for solo developers forever.
How do I install F-Pulse?
F-Pulse OSS isn't public yet — it launches July 20, 2026. Join the waitlist on the access page and we'll email you the install steps the moment it's live. It'll be free and open-source under Apache 2.0, installable via pip, Docker Compose, native desktop installers, or from source. Need it for a team now? F-Pulse+ is available today.
When should I upgrade from F-Pulse to F-Pulse+?
Upgrade when more than one person needs workspace RBAC, when you need DEV→PROD promotion with two-gate approvals, when you need vault-backed credentials with rotation, when you need lineage or drift detection, when you need OIDC/SAML SSO, or when you need 4-hour SLA support. If you're a solo developer running on your laptop, F-Pulse alone is the full product.
How does F-Pulse compare to other pipeline tools?
F-Pulse is visual-first. You build pipelines on a drag-and-drop canvas with a schema-aware expression editor instead of writing Python DAGs. It's faster to start with for SQL-first developers; code-first tools give you more flexibility for purely programmatic workflows. See the compare page for a side-by-side breakdown.
What connectors does F-Pulse support?
Around 90 connector types ship in the registry — relational databases (Postgres, MySQL, SQL Server, Oracle, SQLite, MariaDB, CockroachDB), NoSQL (MongoDB, Cassandra, DynamoDB), cloud warehouses (Snowflake, BigQuery, Redshift, Databricks, ClickHouse), object storage (S3, GCS, Azure Blob, MinIO), streaming (Kafka, Kinesis, Pub/Sub, RabbitMQ), 40+ SaaS REST manifests, and vector databases (Pinecone, Weaviate, Qdrant, pgvector, Milvus). We're honest about maturity instead of slapping 'Certified' on everything: most connectors are SDK-validated (code-complete and schema-correct), a handful are integration-verified against live endpoints, and none are yet production-certified under our strict rubric. Enterprise connectors (SAP, NetSuite, Workday, Dynamics 365, ServiceNow, Salesforce) are F-Pulse+. Anything we don't ship, you can add by pasting an OpenAPI spec.
What if you don't ship the connector I need?
You have four first-class paths, none requiring a vendor build cycle: (1) Paste an OpenAPI URL — working connector in 90 seconds via Insights → Author Connector → from OpenAPI. (2) Paste 1–5 sample API responses — same UI, 'from samples' mode, for vendors without a public spec. (3) Derive from an existing Apache-2.0 OSS project — read auth + pagination + error handling, port into an F-Pulse manifest, add a NOTICE entry. (4) Open a PR or file a request — templates pre-fill what we need to act on it. The OSS bet: we ship the framework, the community ships the long tail.
Can I run F-Pulse on Kubernetes or in the cloud?
Docker Compose is the fastest path for self-hosting. Every service is a standard container, so any K8s install works. Cloud deploy templates ship in the repo root: fly.toml (Fly.io), render.yaml (Render), railway.toml (Railway) — all mount /data to a managed volume so SQLite + configuration survive restarts. A first-party Helm chart is part of F-Pulse+.
Why loopback by default?
F-Pulse binds to 127.0.0.1 — invisible to your LAN. No accidental port exposure to coworkers, hotel WiFi, or conference networks. If you genuinely need LAN-visible binding for an on-prem multi-user install, set FPULSE_ALLOW_LAN=1 or pass --host 0.0.0.0 explicitly. Sovereignty-first by construction.
F-Pulse OSS is almost here
Free and open-source under Apache 2.0, self-hosted on your own machine. Install commands and the GitHub repo go live on July 20, 2026. Join the list and we'll email you the moment it's available.
Get notified at launchNeed it for your team sooner? F-Pulse+ is available today.
See it work. Control the run.
Start free with F-Pulse. Control production with F-Pulse+. Scale to D-Pulse for enterprise.