AI assistants for staff: set up for your business and rules
By Keith Hamilton · · 4 min read
AI assistants for staff are AI tools set up in advance with your tasks, templates and rules, so each person gets consistent, usable drafts instead of starting from a blank box. Many teams have access to an AI tool but use it for little more than the odd email, with everyone writing their own instructions and getting different results. A set-up assistant works differently: it already knows the task, your templates and your rules. This article explains what that looks like, which tasks suit it, and how to keep the results safe and checked.
What are AI assistants for staff?
An AI assistant for staff is a general AI tool with the briefing done in advance. A general AI chat tool knows nothing about your business, so every time someone uses it they have to explain the task, the format and the tone from scratch.
A set-up assistant already has:
- Clear instructions for one task or a small group of tasks.
- Your templates and examples of good work.
- Your tone of voice and house style.
- Rules on what it must not do, such as quote prices or make promises.
Most business AI tools let you build custom AI assistants like this without any coding. The skill is in deciding what goes into them.
This is different from an internal knowledge assistant, which answers staff questions from your documents. A task assistant helps people produce work.
Which tasks suit a set-up assistant?
The right tasks are ones people do often, that follow a pattern and that a person can check quickly. For example:
- Turning rough job notes into a clear update for a customer.
- Drafting a quote or proposal from your template and the details of the job.
- Summarising a long document, contract or report into the points that matter.
- Rewriting a technical explanation in plain English.
- Checking a document against your own checklist before it goes out.
- Drafting a first version of a standard letter or procedure.
Avoid tasks where a wrong answer is costly and hard to spot, or where the work depends on judgement about people, money or legal matters.
What goes into setting up AI assistants?
Three things go into the set-up: instructions, examples and rules.
Instructions
Write instructions as you would brief a capable new colleague. Say what the task is, who the output is for, what format it should take and what to do when information is missing. Tell the assistant to flag gaps rather than fill them with guesses.
Examples
A few examples of good finished work do more than long instructions. Use real examples with personal details removed.
Rules
Set clear limits. For example: never include prices unless they are given, never promise delivery dates, and always leave the final sign-off for a person.
Which AI tools and data should employees use?
Staff should use assistants on business accounts that you control, not personal accounts on free tools. Check what each tool does with the information you put in, where it is stored and whether it is used to train the supplier's models.
Be clear about what data can go in. Customer personal data, staff records, payment details and passwords need particular care, and some should never go into an AI tool at all. Our guide to writing an AI policy covers the rules to set, and our UK GDPR and AI checklist covers the data protection questions.
Does a person need to check every output?
Yes. An assistant produces drafts, and the person using it remains responsible for what goes out. That means reading the output, checking facts and figures against the source, and editing before it is sent or saved.
AI produces text that sounds confident whether it is right or wrong. The more important the document, the more carefully it needs checking.
How do you get the team to use them?
People use assistants that save them time on work they already dislike. Four things make the difference:
- Build the first assistants around tasks the team already finds tedious.
- Show people how to use them on their own real work, not generic examples.
- Keep a shared library of assistants and instructions, with an owner for each.
- Ask for feedback and improve the instructions when results are poor.
How much time do AI assistants save staff?
The saving is the time per task before, minus the time to check the draft, multiplied across the team. Five people who each spend 30 minutes a day turning job notes into customer updates spend 2.5 hours a day on it across the team.
With an assistant producing a good draft that takes ten minutes to check and send, each person spends 20 minutes less a day. Across five people that is 100 minutes, or 1 hour 40 minutes a day back. Measure your own figures before and after to see the effect in your business.
How do you measure the result?
Track four figures from before the assistant goes live:
- Time per task, before and after.
- How much editing each draft needs.
- How many people use each assistant regularly.
- Errors or complaints relating to the work produced.
Where Forwardcycle fits
Our efficiency and capacity service sets up AI assistants for your business and your rules. Our Team Enablement workshops then help staff use them on their real work, leaving you with a library of prompts and workflows. To see where to start, take the free AI Readiness Assessment or arrange a free call.