AI can make a business faster, sharper and more consistent. Yet many teams experience the opposite: they spend time on long conversations with ChatGPT or Claude, receive generic answers and end up rewriting everything. The problem is rarely the tool alone. More often, the task, context and workflow have not been defined clearly enough.
The best starting point is not “Which AI should we buy?” Ask instead: “Where do we repeat work, where is knowledge lost, and where do customers or employees wait unnecessarily?”
A good prompt begins before the prompt
A useful instruction describes the role, goal, audience, source material, constraints and required format. It also explains what a successful result must achieve. AI should not have to guess your tone, rules or customer type when that information can be supplied from the start.
A simple structure is:
- Task: What must be produced or decided?
- Context: Which company, audience and situation are involved?
- Sources: Which documents, data or examples may be used?
- Requirements: What must be included or avoided?
- Format: Should the answer be a table, email, plan or finished draft?
- Quality control: What must a person verify before the result is used?
Find the right places to use AI
AI creates the most value in clear, repeatable processes with a known outcome. Examples include first drafts of proposals and newsletters, meeting summaries, sorting customer enquiries, product copy, research, concept development and internal knowledge search. In sales, AI can assist preparation, follow-up and prioritisation, but it should not send unchecked promises or impersonal bulk messages to customers.
A good pilot is small enough to measure and important enough to matter. Select one process, record the time it takes today, test a new workflow and compare quality, speed and errors after two to four weeks.
ChatGPT, Claude or a custom system?
ChatGPT and Claude are capable general-purpose tools. They can assist analysis, writing, structure, ideas and code when they receive the right context. But if employees must find the same files, repeat the same introduction and copy results between systems every time, the underlying process has not yet been improved.
The next level may be a shared prompt template, an internal assistant using approved sources or an automated workflow connecting forms, CRM, email and company data. Access, personal data, accountability and human approval should be designed into the process from the beginning.
A tailored four-hour AI workshop
BarnerDesign’s workshop starts with your real business. Before the session, you answer questions about your work, current AI use, growth goals and bottlenecks. The four hours can then focus on practical tasks instead of a generic technology presentation.
The session typically includes:
- mapping workflows and selecting the best first AI use cases;
- hands-on work with ChatGPT and Claude;
- prompt methods, iteration and quality control;
- ideas for content, service, sales and internal processes;
- prioritising what your team can do and what may be worth automating.
Explore the AI workshop for businesses, or learn about custom AI systems and automation.
Frequently asked questions
Do employees need previous AI experience?
No. The workshop can be adapted for beginners and experienced users. What matters most is that participants understand the company’s real work.
Can AI put sales on autopilot?
Parts of the sales process can be automated, including research, preparation and follow-up. Relationships, proposals, consent and important decisions should retain clear human responsibility.
Will we leave with something practical?
Yes. The aim is a prioritised list of opportunities plus usable workflows and prompt templates for the selected tasks.
