Build agentic workflows on a no-code canvas or describe them in chat. Agent steps, branches, retries and approvals run in one flow, on the same connections as the Agent Gateway.
Prompt to workflowDescribe the process in chat and Engini builds it on the canvas. Here, a won Salesforce deal creates a customer and a sales order in NetSuite.
01 / The no-code path
One flow. Agent steps next to fixed steps, branches and approvals.
Agentic WorkflowNo-code
Item createdMonday · trigger
AI agentClaude · decides
Get recordNetSuite
Approvalover threshold
Create recordSalesforce
retriesbranchesapprovalsaudit on every step
02 / What changes
Beyond if-then automation. Three differences that matter.
Decisions
Inside clear limits
Agent steps choose the next action from context, using only the tools in their set. Calls that cross your rules wait for a person.
Orchestration
Many agents, one record
Split a process across agents, each with its own toolset. They share context through the workflow and one audit trail.
Resilience
Recorded execution
Failed steps retry, errors branch, exceptions go to the right person. Every step stays on the record.
03 / In practice
Where teams start. Three common first workflows.
Sales
CRM update to proposal
An agent researches the account and drafts the proposal. Discounts above your threshold go to a manager first.
IT and support
Level-2 tickets
Agents read logs and apply approved fixes through ServiceNow. Anything outside the toolset becomes a brief for an engineer.
Strategy
Market intelligence
Agents track competitor pricing and news and write findings into your workspace, with every call on the record.
Prefer code? The same connections are available to your own agents through MCP, SDK, CLI and API.
A workflow in which AI agent steps decide the next action from context, alongside fixed steps, branches, retries and human approvals.
In Engini you build one on a no-code canvas or by describing it in chat. Agent steps can only use the tools in their toolset, and every step is recorded in the audit trail.
How is it different from traditional automation?
Traditional automation follows a fixed path. An agentic workflow can handle variation, such as reading an unstructured request and choosing the right next step, while staying inside the limits you set.
Fixed steps still run exactly as designed. Agent steps add judgment where rules alone are not enough, and approval rules hold sensitive actions for a person.
Do we need developers to build workflows?
No. Workflows are built on a visual canvas or described in chat, using the same connections your team already set up in Engini.
Developers can still extend them, and teams that prefer code can call the same tools directly through the Agent Gateway.
How do approvals work in a workflow?
You set approval rules by amount, system or action. When a step matches a rule, the workflow pauses, routes the request to the right approver with the details, and continues once they decide.
The decision and the approver are recorded in the audit trail.
Which systems can workflows use?
Every system connected to Engini, including ERP platforms such as SAP, Oracle and NetSuite, CRM and ITSM systems, and on-premises systems reached through the On-Prem Agent.
How is this different from the Agent Gateway?
The Agent Gateway is the code path: your own agents call Engini through MCP, the SDK, the CLI or the REST API. Agentic Workflows is the no-code path.
Both use the same connections, toolsets, approval rules and audit trail, so you can mix them as your needs change.
Item createdMonday · trigger
Get recordNetSuite
Create recordSalesforce