First Focus analysed 108 distinct project ideas submitted through its New Zealand AI Investment Fund. They reveal where small and medium businesses expect AI to create measurable value, from reducing manual work and turnaround times to improving consistency and giving skilled employees more capacity.
New Zealand small and medium businesses are not short of AI ideas.
They want tools that can prepare quotes from architectural plans, turn handwritten dispatch notes into sales orders, review contracts, find information in SharePoint and monitor project margins.
The NZ$100,000 AI Investment fund will support five client projects, with up to NZ$20,000 available for each.
Across the submissions, the clearest ideas started with work that already mattered and results the business could measure, such as time spent, processing delays, error rates, customer response times or work completed. They also used information the business already held and kept people accountable for important outcomes.
First Focus’s analysis found that almost half of the 108 ideas focused on workflow, administration, sales, quoting or customer experience.
These were not speculative uses of AI. They addressed processes that already affected employee capacity, customer response times, revenue or management attention.
Applicants proposed tools that could:
Practical AI opportunities may be hiding in work employees complete every day.
Finding information. Reviewing documents. Moving data between systems. Preparing a quote. Following up an approval. Checking that someone has completed a task.
These activities matter, but they can consume skilled employees without making full use of their judgement or experience.
Some submissions also reflected a familiar stage-of-growth issue. Processes that once relied on close coordination and individual knowledge had become harder to manage as volumes increased, teams expanded and businesses introduced more systems.
That was one pattern among several. Other businesses wanted to improve reporting, reduce risk, use existing information more effectively or give employees faster access to specialised knowledge.
Across the submissions, the clearest opportunities began with a defined job, a result the business could measure and a person responsible for the outcome.
Most applicants did not need help inventing an AI use case. They had already found work that was taking too long, relying on manual checks or tying up experienced employees. The real challenge was deciding which idea was worth pursuing first, whether the information was ready, how the tool would fit into day-to-day operations and what result would prove its value.
Philip Barton, Chief Customer Officer, First Focus
Source: First Focus analysis of 108 distinct New Zealand AI Investment Fund ideas, July 2026.
First Focus grouped each idea according to the main business problem it addressed. Some projects could also support other business functions.
The submissions covered dozens of industries and business processes. Three examples show what a well-defined opportunity can look like.
One applicant described a process that began with handwritten dispatch notes.
Employees had to read the notes, interpret the information and enter it into the sales system before the order could move forward.
The proposed tool would read the document, extract the required details and prepare the sales order for review.
The value went beyond faster data entry. A better process could reduce delays between dispatch and administration, create more consistent records and allow employees to focus on exceptions rather than retyping every transaction.
A person would still confirm the information before completing the order.
The business could measure the pilot by comparing processing time, correction rates and the delay between dispatch and order entry with the current process.
This is a useful example of a contained first project. The task happens frequently, the business can observe the current process, the required information is identifiable and an employee can check the output.
Another applicant proposed a system that could review architectural drawings, identify relevant spaces, match them with suitable products and prepare an initial quote.
The system would not make the final commercial decision. It would complete the first stage of analysis, allowing an experienced employee to spend more time reviewing the recommendation and less time assembling it.
The value goes beyond quote preparation time. A controlled process could help the business respond sooner, apply product rules more consistently and rely less on individual employees remembering every detail.
It could also help the business handle more quotes without adding the same amount of manual preparation.
The business could compare quote turnaround time, employee preparation time and the proportion of drafts approved with only minor changes.
A third applicant described valuable technical knowledge that sat with a small number of experienced employees.
When a fault occurred, other staff depended on those people to diagnose the issue or locate the right information.
The proposed system would make approved technical material easier to search, helping employees find likely causes, previous solutions and relevant procedures.
The goal was not to replace the experienced employee’s judgement. It was to reduce routine interruptions, support less-experienced staff and make the organisation less dependent on one person being available.
The business could measure how long employees take to find answers, how often they need help from senior staff and whether newer employees resolve common issues sooner.
These three ideas differ, but they share the same basic structure: a repeated problem, identifiable source information, a defined user, a clear point of human review and a result the business can compare with the current process.
First Focus found that nearly two-thirds of the ideas depended on business records such as emails, contracts, reports, invoices, drawings, forms or spreadsheets.
In many cases, the information already existed. Employees simply spent too long finding, reviewing, interpreting and acting on it.
Applicants wanted tools that could search previous projects, find answers in company policies, review contracts against approved commercial positions, extract requirements from drawings and prepare summaries from financial and operational information.
Small and medium business leaders should note this distinction.
The business may not need more information. It may need a better way to use the information it already owns.
However, an AI tool can only be as useful as its sources.
A knowledge assistant cannot correct outdated policies. A reporting tool cannot resolve inconsistent financial definitions. A quoting system cannot reliably apply pricing rules that exist only in the heads of several employees.
Before choosing a tool, the business needs to understand which information the system will use, whether that information is accurate and current, who owns it, who may access it and how employees can verify the answer.
AI can make information easier to use. It does not remove the need to manage that information properly.
More than half of the ideas mentioned Microsoft 365 or applications such as SharePoint, Outlook, Teams, Excel, Word, Power BI or OneDrive.
Almost half named at least two specific business applications.
That reflects how work happens inside a business. A process rarely begins and ends within one system.
Consider the proposed architectural quoting tool.
The AI component might interpret the plans and suggest a product selection. The complete process could also involve receiving documents through Outlook, retrieving customer details from a CRM, accessing product specifications in SharePoint, applying pricing from an accounting system and recording the approved result.
The demonstration may be the easiest part.
Putting the tool into daily use requires the business to address information quality, system access, permissions, process rules and ownership.
This is one reason promising trials can struggle to move beyond the pilot stage. The AI may work even when the surrounding business process does not.
Small and medium business leaders do not need to design the technical solution, but they should understand the wider process:
These may sound like operational and IT questions rather than AI questions.
In practice, they are both.
What turns an AI idea into a working business process
Several applicants proposed AI for work where an incorrect result could have financial, legal, safety or professional consequences. These ideas covered contract review, financial checks, engineering design, audit reporting, payroll remediation and quality control.
In these cases, applicants generally planned for AI to support a qualified person rather than make the final decision.
A system might identify an unusual contract clause, flag missing information, compare a design against defined rules or prepare a draft report. A lawyer, accountant, engineer, auditor or manager would remain responsible for the outcome.
Business leaders should define that boundary before development begins.
For every project, decide what the system may do independently, what it may recommend, what requires human approval and what it must never decide.
The level of review should reflect the consequence of a wrong answer. An internal meeting summary does not carry the same risk as a customer quote, safety decision or tax return.
Where personal information is involved, the organisation must also consider its obligations under the Privacy Act. The Office of the Privacy Commissioner recommends completing a Privacy Impact Assessment before using AI with personal information and reviewing it as the system’s use changes.
Applicants described almost two-thirds of their ideas as being at validation stage or later.
Many had already identified the process they wanted to improve, the employees who would use the tool, the information it would require and the expected business result.
Some applicants also estimated the time or financial value their ideas could create. These figures represent forecasts, not measured results. However, they show that many applicants were thinking about business value rather than novelty alone.
This reflects a wider shift in the market.
The Ministry of Business, Innovation and Employment commissioned research involving 500 New Zealand SMEs in April 2025. The study found that 94% knew about at least one AI tool, while confidence, capability, privacy, security and uncertainty about where to begin continued to affect adoption.
AI Awareness is no longer the main challenge.
The harder task is choosing a project that is worthwhile, practical to deliver and suitable for the organisation’s information and risk environment.
Acting without that discipline also carries a cost.
A business can spend money on licences, demonstrations and isolated pilots without changing how people complete the work. Employees may test several tools, but no one owns the source information, adoption or ongoing process.
The result is more software and another layer to manage.
Small and medium business leaders do not need to choose an AI model or write technical requirements.
They do need to establish whether the project has the right business conditions.
Define the intended business result before choosing a tool.
Establish a baseline by asking:
Without a baseline, the business may know that the tool works but still not know whether the project was worthwhile.
A frequent process with a moderate cost may make a better first project than a rare problem with a large theoretical return.
Identify the documents, records and systems the tool will need.
Confirm who owns the information, whether it is current, who may access it and how the business will maintain it.
Poor source information will limit the value of even a capable AI system.
Begin with one team, process, document type or customer group.
A contained pilot makes it easier to test accuracy, gather feedback and correct weaknesses before extending the tool.
The goal is to learn under controlled conditions, not to introduce AI across the entire organisation at once.
Assign a senior business sponsor and a process owner.
Someone must take responsibility for output quality, employee adoption, permissions, source information, feedback and measurement.
Without a clear owner, an AI project is likely to remain a demonstration.
Many submissions crossed the boundaries between Microsoft 365, SharePoint, business applications, security, data governance and employee training.
That creates an ownership problem.
A specialist may build the initial tool, but someone still needs to manage its information, access controls, system connections, users and ongoing changes.
This is the thinking behind CORE, First Focus’s Managed AI & IT Services model.
CORE supports New Zealand businesses with approximately 20 to 200 employees that need a fully managed technology function. It brings IT support, cyber security, data governance, SharePoint management, AI adoption, workflow improvement, training and ongoing guidance under one accountable partner.
The base service does not include every custom build or major project. Larger implementations, integrations and development work may require separate scope or consulting capacity.
The managed model gives the business continuity and accountability.
Instead of treating each AI idea as an isolated experiment, CORE gives the business one partner to consider the wider environment around the project and keep asking:
What should we improve next, how will we measure it and what needs to be in place for it to work?
The clearest lesson from the fund is not that New Zealand businesses need more AI ideas.
They already have them.
The opportunity is to choose a business process worth improving, understand the information and systems behind it, define where people remain responsible and agree on how the business will measure success.
For some organisations, this will mean updating processes that have not kept pace with growth. For others, it will mean improving reporting, knowledge access, customer service, compliance or operational planning.
In every case, the project should begin with the business result rather than the technology.
Start with one recurring process that regularly pulls capable employees away from customers, decisions or higher-value work.
Document how it works today and choose the result you want to improve. That may be turnaround time, processing cost, error rates, customer response time or the amount of skilled employee time the process consumes.
Then identify the information and systems involved, decide where a person must review the result and test the idea within a contained scope.
That approach is less dramatic than trying to introduce AI across the entire organisation.
It is also more likely to produce something employees use, the business can maintain and leaders can assess against a clear result.
First Focus can help assess the business problem, intended result, information, systems, risks and likely delivery path before you commit to a tool or project.
The conversation will help determine whether the opportunity is suitable, which foundations the business may need, what a sensible first scope could look like and how you could measure its value.
First Focus analysed 108 distinct project ideas submitted by its New Zealand clients through the AI Investment Fund in July 2026.
First Focus grouped each idea by its primary business theme and also reviewed project readiness, systems, information sources and expected impact.
Applicants supplied their expected outcomes and estimates. First Focus did not independently verify these figures as measured business results.
A business can define its measures of success before a pilot. Measured results only become available after implementation and comparison with an agreed baseline.
Because these organisations applied for AI funding, the findings reflect businesses already motivated to explore AI. This analysis does not represent every New Zealand small or medium business.
Sources
Ministry of Business, Innovation and Employment, AI adoption by New Zealand small and medium-sized businesses, July 2025.
Office of the Privacy Commissioner, guidance on artificial intelligence and privacy.
First Focus, New Zealand AI Investment Fund, 2026.
First Focus, CORE: Managed AI & IT Services.
IT Consulting
IT Strategy
Data Services
IT Infrastructure Projects
IT Procurement
Virtual CIO
Offshore Staff Outsourcing (BPO)
Managed AI & IT Services
IT Support Services
Managed Cloud Services
Managed SharePoint Services
End User Computing
Unified Communications
Connectivity
Managed Security Services
Cybersecurity Risk Assessment
Business Continuity Planning
Cloud Backup and Disaster Recovery
Automation
Digital Transformation
Power BI
Careers
Awards and Certifications
Management Team
Case Studies
Sell Your IT Business
Our Commitment to the Environment
CSAT Charity Donations
Media Kit
Auckland | Tāmaki Makaurau
