Pipelines runs 17 node types today.
Pipelines are the wire-driven half of the company. A pipeline is a DAG of typed nodes that fires from a Run-now button, a 5-field cron line, a webhook URL only you hold, an inbound Gmail message, or a new task in a Project, and it does the same thing on every pass. Put an AI Employee in the middle only where a step needs judgment.
Five trigger kinds start a Pipeline.
A visual DAG of 17 node types fired by cron, a webhook, inbound Gmail, or a new task, with 20 Integrations callable as nodes.
Drag a node, then type {{trigger.body.name}}
Drag nodes from a catalog palette, wire edges between handles, and configure each node in a side panel. Data moves between them as {{trigger.body.name}} templates, and a whole-token value keeps its type.
Five triggers, from a cron line to inbound Gmail
Fire a pipeline from a Run-now button, a 5-field cron schedule, an incoming webhook with its own secret URL, an inbound email on a connected Gmail mailbox, or a new task added to a Project. The scheduler's 30-second heartbeat advances before firing, so slow runs can't double-fire.
Six actions write into Bases, Projects, and channels
Six built-in actions write straight into the primitives your team already uses: post a channel message, add a todo, create a project, append a Base record, ask an AI Employee, or write a journal note.
Color-coded true/false edges and a full HTTP node
If/else branches carry color-coded true/false edges. Set-variable nodes and delays sit alongside a full HTTP request node with method, headers, and body, whose responses auto-parse as JSON.
One node calls Stripe, Postgres, or 18 more
One node invokes any tool on any connected Integration, from Stripe and Gmail to GitHub, Notion, Linear, Airtable, Postgres, Telegram and more, with the result captured for downstream nodes.
The Runs tab holds your last 50 runs
Every run records its status, which trigger fired, the payload, per-node outputs, and a step-by-step log. The Runs tab lists the last 50 and refreshes itself while one is in flight.
Stripe reported a payment over $1,000.
Pipelines and AI Employees are complements, not competitors. Keep the deterministic 90% on wires and drop a model in only where a decision is genuinely needed.
Ask AI Employee, mid-flow
One node sends a message to an employee and captures its reply for downstream nodes — a webhook arrives, the employee summarizes or decides, the pipeline carries on deterministically.
Routines can pull the trigger
An employee running a Routine can POST to a pipeline's webhook URL, so AI-driven work can kick off wire-driven work.
Employees build them, not just run in them
Ask an employee to stand up a receiver and it composes the steps, tests them, and hands you the webhook URL — through its genosyn tools, no canvas required.
An employee-built flow can't exceed its author
Pipelines run as the company, so every step an employee writes is checked against that employee's own Grants before it saves. A Base, private channel, restricted Project, or Connection it wasn't granted is refused, naming the step.
Readers ask 5 questions about Pipelines.
How are Pipelines different from Routines?
Routines are scheduled work performed by an AI Employee — a model is always in the loop. Pipelines are deterministic DAGs of typed nodes: same result every run, no LLM involved unless you explicitly add an Ask-AI-employee node. Routines are AI-driven; Pipelines are wire-driven.
What can trigger a pipeline?
Five trigger types: Manual (a Run-now button), Schedule (standard 5-field cron on a 30-second heartbeat), Webhook (each node gets a unique secret URL and the POST body becomes the trigger payload), Email received (a new inbound message in a connected Gmail inbox), and Task created (a new task added to any Project). The email trigger takes an optional comma-separated Mailboxes list — leave it empty to watch every connected mailbox — plus sender, subject, and attachment filters; the task trigger narrows by Project, priority, or title. A pipeline can carry multiple triggers, and each run records which one fired.
Can a pipeline talk to my other tools?
Yes. The Call-integration node invokes any tool on any connected Integration — 20 are registered, including Stripe, Gmail, GitHub, Notion, Linear, Airtable, Postgres, MySQL, ClickHouse, Telegram, X, Reddit, and LinkedIn. For everything else there's a generic HTTP node.
How do I debug a failed run?
Open the Runs tab: every execution keeps its status, trigger kind, payload, per-node outputs, any error, and a step-by-step log. Safety rails cap runs at 200 steps and delays at 60 seconds, so a wiring mistake can't loop forever.
Do I need to learn a DSL or write code?
No. You build on a visual canvas with a node palette and per-node config forms. Data flows between nodes with {{trigger.body.name}}-style templates, and whole-token templates preserve types like numbers and arrays.
4 more products share one database.
AI Employees
Teammates that carry a markdown Soul and work Routines on a schedule you pick, with every Run transcribed and costed.
Workspace
Channels, DMs, and 25 MB file uploads where AI Employees are members you can @mention.
Tasks
A task manager where any todo goes to a human or an AI Employee, and AI work stops at in_review until you sign it off.
Bases
Multi-table workspaces with 11 field types, saved views, comments, and attachments, plus 21 built-in tools for a granted AI Employee.
Tomorrow one thing is finished before 09:30.
Install Genosyn, register an AI Model, write one role, and put it on a schedule. The company grows from there.