News & Articles Hub
Beyond the Prompt: Making AI Actually Work for Your Business
HGA Takeaway
Beyond the Prompt: Making AI Actually Work for Your Business
AI can do far more than write the occasional email or social post. The real opportunity comes when you give it the right knowledge, clear instructions and repeatable jobs that take work off your team's plate.
Most of us have probably asked AI to write something by now. But if that's where your use of AI starts and ends, there's a lot more you could be getting from it.
Used well, AI can become much more than a blank box you type a question into. It can learn how your business communicates, work from your existing processes and information, carry out repeatable tasks and even connect with the tools your team already uses.
The AI Assistant & Workflow Essentials session explored how businesses can move from occasional prompting towards assistants, agents, skills and workflows that are actually useful in day to day work.
The biggest opportunity isn't asking AI to do more things. It's teaching it how your business already does things well.
A useful way to think about AI
Treat AI like a very enthusiastic junior employee
One of the simplest ideas from the session was to think of an AI model as a junior executive with infinite stamina.
It can process huge amounts of information, repeat tasks without getting bored and produce work incredibly quickly. But that doesn't mean you should give it a vague instruction and assume whatever comes back must be right.
Just as you would with a new member of staff, it needs context, instructions, examples, boundaries and somebody checking the quality of its work.
AI can do the heavy lifting. Humans still need to provide the judgement.
Takeaway 1
Think 10% setup, 70% AI, 20% quality control
The presentation applies a simple 10:70:20 effort ratio to AI. The idea is that humans put effort into setting the task up properly, AI carries out the bulk of the work and humans then review what comes back.
|
10% Set up & training Give AI the information, instructions and context it needs to do the job properly. |
70% AI does the work Let the technology handle the time consuming processing, drafting and repeatable activity. |
20% Quality assurance Check accuracy, sense check the output and make sure the finished work is genuinely good enough to use. |
Takeaway 2
Good AI starts with good foundations
Before worrying about agents, workflows and automation, there are three much simpler things to get right.
Start with:
Clear instructions: Tell AI what you want, give it context, explain your preferences and be clear about the expected output.
Documented processes: If your business already has a good way of doing something, capture it. That process can become part of the knowledge AI works from.
The right tool: Different platforms and models offer different capabilities. Start with the job you're trying to solve rather than choosing technology for the sake of it.
Takeaway 3
Give AI your business knowledge
This is where AI starts becoming much more useful.
Instead of explaining your business from scratch every time, assistants and agents can be given relevant company information to work from.
That might include your brand guidelines, tone of voice, previous product descriptions, company processes, photography guidance, policies, SEO information or examples of work you already know is good.
Think about the information sitting in your business already
Brand guidelines
Tone of voice documents
Product information
Previous examples of good work
Customer service processes
Internal policies
Standard operating procedures
Once AI has access to the right information, it can refer back to those documents when answering questions or producing things like reports, proposals, strategies and content.
Takeaway 4
Stop giving the same instructions over and over again
If you regularly find yourself typing the same long prompt, explaining the same tone of voice or uploading the same documents, that's a good sign that the task could be set up more permanently.
An AI agent can essentially become a reusable specialist. You give it a job, instructions, knowledge, tone and boundaries once, then call on it whenever that type of work comes up again.
For example
Imagine creating a Product Copy Assistant for your business.
You could give it your brand guidelines, previous product descriptions, SEO requirements and tone of voice. Instead of explaining all of that every time a new collection arrives, the assistant already understands how the job should be done.
Takeaway 5
Agents, skills, tools, workflows... what does any of it actually mean?
The terminology can make AI sound far more complicated than it needs to be. A simple way of thinking about it is:
|
Agent The specialist responsible for the overall job. It holds the relevant instructions, knowledge, tone and boundaries. |
Skill A specific repeatable task that specialist knows how to carry out. |
|
Tool The thing that allows AI to take an action outside the conversation, such as working with files, email or another business system. |
Workflow A repeatable sequence where several steps are carried out in the right order. |
Takeaway 6
Look for the jobs your team keeps repeating
The presentation gives a useful rule of thumb: build around work that happens repeatedly.
Across a Home & Gift business, there are plenty of potential examples.
Ecommerce: Product descriptions
Marketing: Campaign briefs
SEO: Optimising collection and category pages
Customer service: Drafting customer responses
Buying: Trend research
Merchandising: Cross sell recommendations
Social: Caption creation
Leadership: Weekly trading summaries
Takeaway 7
AI becomes even more useful when it can take action
Creating something is useful. Being able to do something with it is where another layer of opportunity opens up.
The session explains that tools and connectors can link AI with other systems, allowing it to move beyond simply responding in a chat.
That could mean an AI system that can:
Create a report and save it into the right shared folder
Draft or send an email
Create a calendar event
Post a notification into a team channel
Pass information into another stage of a business process
Takeaway 8
If it happens again and again, it might be a workflow
Think about the tasks in your business that follow roughly the same journey every time.
Someone does step one, passes something to somebody else, information gets checked, an email is sent, a document is created and somebody follows up. These repeatable processes are exactly where workflows can start becoming useful.
The simple rule
Repeatable task + multiple predictable steps = something worth exploring as a workflow.
Examples in the session range from employee onboarding and contract approval to expense processing, customer refunds, sales activity and event registration.
Takeaway 9
What does this look like in the real world?
Some of the most interesting parts of the session were the examples of assistants already being used by businesses.
|
LinkedIn content Turning supplied content into articles and posts while following the company's existing brand guidelines and tone of voice. |
Sales coaching Reviewing call transcripts and providing structured feedback based on the sales methodology used by the business. |
|
Email support Creating and reviewing emails while adapting them for different audiences, buyer stages and business requirements. |
Data cleaning Cleaning and matching information to make marketing attribution and reporting easier. |
The point isn't that every business needs these exact assistants. It's that the best opportunities often come from looking at the repetitive, time consuming work already happening inside your organisation.
Takeaway 10
You need to know how the job should be done first
There's an important point running through the whole session: AI still needs something good to learn from.
If your process isn't documented, your tone of voice isn't clear or nobody agrees what a good result looks like, giving the job to AI isn't going to magically solve that.
Before asking "Can AI do this?", ask "Do we know what good looks like when a human does it?"
Where to start
You don't need to automate your whole business tomorrow
Agents, skills, connectors and workflows can sound intimidating when they're all discussed at once.
In reality, the easiest starting point is much smaller. Find one job your team does regularly, particularly one that involves repeating the same instructions or working through the same steps.
Try this exercise with your team:
1. What do we do repeatedly?
2. Which of those jobs takes more time than it should?
3. Do we already have a clear process for doing it well?
4. What information would AI need to understand the job?
5. Which parts could AI reasonably handle?
6. Where does a human still need to check, approve or make a decision?
For Home & Gift businesses
Where could AI save you time?
For many businesses in our industry, teams are small and people wear several hats. The same person might be working on product launches in the morning, retailer emails after lunch and social content before they go home.
That makes the opportunity particularly interesting.
Imagine an assistant that already understands your brand voice and helps create product descriptions. Or one that turns product information into retailer emails, social posts and website copy. Perhaps your team spends hours preparing weekly sales reports, analysing marketing data, researching trends or answering similar customer questions.
The aim isn't to remove the human expertise behind those jobs. It's to stop talented people spending quite so much of their day doing the repetitive parts.
Use AI to remove the repetitive work, not the human thinking that makes the work valuable.
The HGA Takeaway
Start with the problem, not the technology
1. Treat AI like a capable junior member of the team. Give it context, instructions and oversight.
2. Don't skip quality control just because AI produced something quickly.
3. Document the processes your business already does well.
4. Give AI relevant company knowledge rather than starting from scratch every time.
5. Look for repeatable tasks where you're constantly giving the same instructions.
6. Use agents and skills to package knowledge and expertise you need repeatedly.
7. Explore workflows where a job involves several predictable steps.
8. Start small. Solve one genuine problem before trying to automate everything.
One final thought
The clever bit isn't the AI. It's knowing what to teach it.
The businesses that get the most value from AI won't necessarily be the ones using the most complicated technology.
They'll be the ones that understand their own processes, know what good work looks like and can identify the jobs where technology genuinely gives their people time back.
This HGA Takeaway is based on the AI Assistant & Workflow Essentials presentation. We've pulled out the points we felt were particularly useful and practical for businesses across the Home & Gift industry.
Home & Gift Association
Your Industry. Your Community. Your Growth Partner.