You don’t need to understand prompt engineering, JSON schemas, or workflow orchestration to get value from JieGou. Here’s how to go from zero to a working AI recipe in about five minutes.
Step 1: Pick your department
When you first sign in, JieGou asks which department you work in. This isn’t just a label — it determines which recipes, workflows, and integrations you see first. Pick the one that matches your role. You can always change it later.
Step 2: Install your starter pack
After selecting a department, you’ll see a recommended starter pack. The Sales pack includes 10 recipes and 5 workflows. Marketing gets the same. Support, HR, Finance, Operations, Legal, Engineering, and Executive each have their own curated sets.
Click Install Pack and everything gets added to your account. These are fully editable copies, not read-only templates.
Step 3: Run an existing recipe
The fastest way to see what a recipe does is to run one. Let’s say you installed the Sales pack and want to try Prospect Research.
Open the recipe and you’ll see a form with input fields — company name, website URL, any specific questions you want answered. Fill in the fields and click Run.
Behind the scenes, JieGou sends your inputs to the AI model (Claude, GPT, or Gemini — whichever you’ve configured), structured by the recipe’s prompt template and output schema. The result comes back as structured data: company overview, key contacts, recent news, competitive landscape.
That’s it. You just ran your first recipe.
Step 4: Create your own recipe
Now try building one from scratch. Click New Recipe and you’ll see three sections to fill in:
Input schema — What data does the recipe need? Define fields like customer_name (text), issue_description (text), or priority (select: high/medium/low). Each field gets a type and an optional description.
Prompt template — The instructions for the AI, with placeholders for your input fields. Something like:
Analyze the following customer issue and draft a professional response.
Customer: {{customer_name}} Issue: {{issue_description}} Priority: {{priority}}
Output schema — What structure should the result have? Define fields like root_cause (text), suggested_response (text), follow_up_needed (boolean).
Save the recipe, fill in the inputs, and run it. The AI returns structured output that matches your schema — not a wall of unformatted text.
What to try next
Once you’re comfortable with recipes, the natural next steps are:
-
Chain recipes into a workflow — Connect multiple recipes so the output of one feeds into the next. The Workflow Builder lets you add condition branches, loops, parallel execution, and approval gates.
-
Schedule recurring runs — Set up a cron schedule to run a recipe or workflow automatically. Pull inputs from Google Sheets, a webhook, or static values.
-
Try the conversational AI — Open the Chat page for a department-aware AI assistant that can suggest and run recipes during conversation.
You don’t need to start from scratch
The starter packs exist so you don’t have to build everything yourself. Run the pre-built recipes first, see how they work, then customize the prompts and schemas to match your team’s specific needs. Most teams are productive on day one.