Who it’s for: All users (depending on access)
Platform: AMI / Sense Workplace
Available on: All plans (feature availability may vary. See availability)
Overview
AMI works by interpreting natural language instructions (prompts), whether entered as text or voice, and using them to analyse data, generate outputs, or perform actions.
The quality of your prompt directly affects the quality of the result.
Clear, specific prompts help AMI understand:
what you want to do
what data to use
what output you expect
What makes a good prompt
An effective prompt usually includes three things:
1. The task
What you want AMI to do
Examples:
“Analyse…”
“Show…”
“Create…”
“Send…”
"Summarise..."
"Build..."
2. The context
Relevant details that help define the scope
Examples:
department, role, or location
time period
specific employees or records
document or report names
3. The output (optional)
How you want the result presented
Examples:
“as a table”
“as a chart”
"as an image"
“with a summary”
“for download as an Excel file”
Example prompts
Basic prompt:
“Show me employee absences”
Result: Likely too broad and less useful
Improved prompt:
“Show me employees with more than 3 absences in the last 6 months”
Result: More targeted and useful
Advanced prompt:
“Show me employees with more than 3 absences in the last 6 months, grouped by department, and present the results as a table”
Result: Clear, structured output
Prompt tips for better results
Be specific
Avoid vague instructions.
Instead of:
“Analyse this”
Use:
“Analyse attendance patterns for the London office over the last 3 months”
Add relevant filters
Include criteria such as:
department
role
location
date range
This helps AMI narrow down the results.
Ask follow-up questions
You don’t need to get everything right in one prompt.
You can refine results by asking:
“Can you break that down by department?”
“Show this as a chart”
“Filter this to managers only”
Use natural language
You can create prompts as if you were asking a person.
There is no need for technical syntax.
Check and refine
If the result is not what you expected:
rephrase your prompt
add more detail
clarify your intent
Using prompts with different agents
Different AMI agents respond to different types of prompts based on their role and data.
Examples:
Doc Sender → “Send the updated health and safety policy to all warehouse employees”
Payroll → “Process this payslip file and match by NI number”
HR Calendar → “Show late arrivals for last week”
Quizzy → “Create interview questions for a sales manager role”
💡 Tip:
If you’re unsure what an agent can do, ask it directly. Most agents can explain their capabilities and guide you.
Reusing successful prompts
When you achieve the result you need in a conversation, you can ask AMI to provide the full prompt or instructions used to generate that output, then copy and save it to your own files for future use.
While you can revisit previous conversations using the sidebar, saving prompts separately makes them easier to reuse, standardise, and share with other administrators or managers.
This is useful for:
recurring or scheduled tasks
repeated analysis or reporting
standardised workflows
💡 Tip: Saving and reusing prompts helps ensure consistent results and reduces the time and token usage required for repetitive tasks.
Common mistakes to avoid
Prompts that are too vague
Example: “Show me data”
✓ Add detail about what data and why
Missing context
Example: “Send the document to employees”
✓ Specify which employees and which document
Unclear intent
Example: “Fix this”
✓ Explain what needs to be changed or improved
Basic troubleshooting
Issue | Possible cause |
Results are too broad | Prompt lacks detail or filters |
Results are incorrect | Missing or unclear context |
No useful output | Task, context, or output rules not clearly defined |
💡 Tip:
If you’re unsure how to phrase your request, ask AMI for help — it can guide you step by step.
Summary
Writing effective prompts helps you get the most out of AMI.
By being clear, specific, and iterative, you can:
improve accuracy
reduce the need for rework
generate more useful outputs
complete tasks more efficiently
Good prompts lead to better results — and better use of AMI.