Assessing AI Solutions under the Enhanced DTSPP
Before adopting AI, a business should set clear goals and success metrics. This article looks at how to assess a solution's fit with the workflow, the quality of its output, the time it saves and the management it needs afterwards.
The Hong Kong government has announced that it will enhance the Digital Transformation Support Pilot Programme (DTSPP), using matching grants to help more SMEs adopt AI and cybersecurity solutions. Which AI products or services will meet the requirements of the new round is still subject to formal guidelines.
For a business, assessing an AI solution should not stop at the feature list or a demo. The more practical approach is to confirm whether the solution can improve one specific piece of work, and how you will judge whether that improvement is real.
Start with a single workflow
Choose one repetitive, time-consuming or error-prone task as the starting point: sorting customer enquiries, organising documents, searching internal information, or drafting answers to common questions, for example.
First record how the work is handled today: how many hours a week it takes, how many staff are involved, and where errors or delays usually occur. This becomes your baseline for comparison.
Set measurable indicators
Different AI projects call for different indicators. Choose them according to the actual workflow, for example:
Average handling time per enquiry
Staff hours saved each week
The share of AI replies that need editing or handover to a person
Accuracy of document classification or information search
How often staff actually use the solution
Indicators should be simple, recordable and directly related to the problem you set out to improve. User numbers or the volume of generated content alone may not show whether the solution is genuinely helping.
Pilot on a small scale before expanding
Before full adoption, run a pilot with one department, one type of document or one kind of enquiry. During the pilot, record output quality, how staff use the tool, and where human review is still needed.
If the AI will handle customer data or internal documents, first confirm access permissions, storage arrangements and what staff are allowed to enter. Keep a human check and correction step for anything important.
Include ongoing management in the assessment
After an AI tool goes live, someone still has to handle account management, data updates, usage guidelines and follow-up on problems. When comparing solutions, ask about implementation time, training, support scope, subscription costs and possible extra charges.
The specific application conditions and the scope of qualifying solutions under the enhanced DTSPP are still to be announced. A business can assess solutions now using clear business goals and success metrics, then check a project against the formal guidelines when they are published.
Want to assess whether an AI solution fits your current workflow? Contact us to discuss suitable use cases and success metrics. Eligibility and funding arrangements are subject to the latest official announcement.