From training judgment to fostering creativity: How AI is helping staff from these SMEs grow on the job

The inaugural SME AI Impact Awards 2026 recognized 30 winners from 290 nominations across 21 industries, highlighting the transformative power of AI in SMEs. At Accountingbpo, an accounting firm, AI tools confidently prepared tax analysis but were prone to errors, as demonstrated when the tool cited non-existent provisions in tax agreements.

An accountant's expertise was crucial to catching and correcting this mistake, emphasizing the importance of human oversight in AI adoption. Accountingbpo, founded in Singapore in 1997, serves approximately 400 SME clients with a focus on financial document processing, assignment coordination, and report generation. The firm developed its own AI solution in 2025, winning the SME Innovation Excellence Award at the awards ceremony.

IMDA's CEO, Ng Cher Pong, emphasized that successful AI integration lies in its seamless embedding within business workflows and its positive impact on operations, employees, and customers. Accountingbpo's management understands that younger or less experienced employees require further development to challenge AI-generated outputs effectively.

To enhance their knowledge, junior accountants use AI to research technical questions within authoritative sources, such as the Inland Revenue Authority of Singapore's IRAS e-Tax Guide and the firm's resources. This practice, coupled with mentorship from senior colleagues and increased client interactions, enables juniors to question AI-generated answers and understand the reasoning behind transaction recording.

GetGo Technologies, Singapore's largest carsharing business, employs AI to streamline fleet maintenance processes. The company received thousands of photos daily for damage assessment, a task that was manually intensive. Their AI solution, Project Desmond, employs computer vision and neural networks to identify potential damage in photos, significantly reducing the time required for review.

Project Desmond, which earned GetGo the SME Innovation Excellence Award, still requires human input to ensure accuracy, as it may mistake raindrops or reflections for damage or miss real damage reported by users. The system receives continuous guidance from human reviewers to improve its accuracy over time.