AI Workflow Suggestions

Turn a sentence into a starting workflow

Staring at a blank workflow is the hardest part of automation. AI workflow suggestions let you describe what you want in ordinary language — "when a customer pays an invoice, thank them by email and tag them as a paying customer" — and get back a suggested workflow: a proposed name, a trigger, and an ordered list of steps, each with an action and a short detail. You then refine it, wire up the real mappings, validate, and simulate before it goes live.

Think of it as a fast first draft, not a finished automation. It gets you to a sensible structure in seconds so you can spend your time on the details that matter.

How to use it

1. Open the workflow builder and choose Suggest with AI (or start a new workflow from a description).
2. Type a clear, specific description of the automation you want. Include the trigger ("when …"), the actions ("send …", "create …", "tag …"), and any conditions ("only if …").
3. Submit. StretchAutomate returns a suggested workflow: a name, a proposed trigger, and ordered steps with an action and detail for each.
4. Review and adapt. Map the AI's generic actions onto real connectors and actions (e.g. its "send email" becomes StretchMail send_email), fill in configurations and data mappings, and adjust dependencies.
5. Validate & simulate the result before activating — the suggestion is a draft and must earn its way to Active like any other workflow.

Writing a good prompt

The quality of the suggestion tracks the clarity of your description. Strong prompts:

• name the trigger event explicitly ("when a form is submitted", "when an order is created");
• list the actions in order ("first tag the customer, then send a welcome email, then create a task");
• state any conditions as filters ("only for orders over $500");
• mention the apps involved when you know them ("in StretchCRM", "via StretchMail").

Realistic example

A marketer types: "When someone submits the contact form, add them to the newsletter list, wait a day, then email them a getting‑started guide." The suggestion comes back named "New contact → nurture," with a form‑submission trigger and three steps: add to list, a delay, and a send‑email. The marketer maps the first step to StretchMail add_to_list, keeps the delay, sets the email step to StretchMail send_email with the recipient pulled from the form submission, validates, simulates with a test submission, and publishes.

Availability

AI workflow suggestions are optional and gated — an administrator enables them for your platform. When suggestions are turned off (or the AI service is not ready), StretchAutomate simply reports that the feature is unavailable and you build the workflow manually; nothing else is affected. The feature is fail‑safe: if the model can't produce a suggestion for a given description, you get a clear "no suggestion available" response rather than an error, and you continue building by hand.

Tips

• Treat the suggestion as scaffolding. Always replace generic actions with the specific connector + action you need, and add real data mappings.
• Be specific about trigger, order, and conditions — vague prompts yield vague drafts.
• The AI won't send anything or change data — it only proposes a structure. You still validate and simulate before activating.
• If suggestions are unavailable, check with your administrator; the capability is a toggle, not a bug.

FAQ

Does the AI run my workflow? No. It only proposes a name, trigger, and steps. You build, validate, simulate, and activate it yourself.

Why did I get "no suggestion available"? Either the description was empty, the feature is disabled, or the model couldn't produce a structured result. Rephrase with a clearer trigger and ordered actions, or build manually.

Is my description sent outside the platform? Suggestions are generated by the platform's own AI service when your administrator has enabled it. If it isn't enabled, no suggestion is generated at all.