Daniel runs Tilley & Patton, a Wellington-based engineering and fabrication business with 34 staff. The company designs and manufactures metal products for public infrastructure including street and park furniture. They work with councils, construction firms, architects and developers across New Zealand.
As the business grew, so did the amount of data it managed. It became harder to keep track of information across spreadsheets, planning tools, and customer relationship management (CRM) systems. This meant that decision-making took longer.
Daniel decided to test if AI could help save time and improve accuracy with tasks like forecasting, manufaturing process documentation, and organisation charts. Because he tested specific tasks, his team could check if the AI tool would work to speed up these tasks – but in a controlled way.
After a successful trial, he moved on to standardising templates, refining CRM data, and improving production planning.
The business chose paid licences for an AI assistant because the tool could work across multiple tasks and teams – including engineering, sales, finance and management. Teams used AI to tidy up and organise data, check spreadsheets, summarise documents, and help with reporting and planning. They also used it to plan sales and operations activities.
Daniel progressed to the next step and created simple AI tools and assistants to support specific tasks: One agent helps with product specifications, technical questions, and keeping information consistent for sales staff.
The other agent creates quick visual concepts for products with supporting proposals, so designers can test ideas before booking an engineer.
Agents automate tasks, but team members always review final outputs before using them.
The results for Daniel and the team were practical and clear. Time on admin dropped, reporting became quicker and more consistent, and CRM data was easier to find and use. This helped teams make good decisions faster and deliver more reliable technical responses to customers.
Like many businesses, Daniel quickly noticed that for AI, clear instructions are essential – vague inputs lead to mixed results. The team manages this risk by writing clear prompts and having a person validate all AI work – especially for engineering tasks where accuracy is important.
Daniel’s advice for other businesses is to start small on practical problems – like reporting, CRM setup or forecasting – and scale from there. Human oversight is important: “AI supports expertise – it does not replace it. It’s less about automation and more about improving clarity and speed.”