Perspectives from organisations in Singapore who have completed AI engagements with Tensora.
Back to HomeCompleted engagements
Average rating
Repeat clients
In practice
"We brought Tensora in to build a demand forecasting model for our distribution operation. What stood out was the assessment phase — they identified two data issues upfront that would have undermined the model. That kind of honesty saved us from a much longer and more expensive problem down the line."
Wei Cheng
Operations Director · Logistics, Singapore
February 2025
"The conversational AI Tensora built for our client-facing service has been running for five months now. The dialogue architecture handles edge cases far better than I expected. The documentation they provided made it straightforward for our internal team to extend it when our service scope changed."
Priya Ramasamy
Digital Services Lead · Financial Services
January 2025
"Our knowledge graph engagement covered twelve years of internal documentation. The ontology design sessions were genuinely collaborative — the team understood our domain quickly and pushed back constructively when our initial ideas about entity boundaries would have caused problems. The result is something we actually use daily."
Hideyuki Tanaka
Head of Knowledge Management · Professional Services
December 2024
"We were a small team at a business unit level, not a large enterprise, and the forecasting consultation was sized appropriately for what we needed. Three weeks in we had a working model and a clear understanding of what it could and couldn't tell us. That kind of calibrated expectation is genuinely useful."
Sarah Ng
Analytics Manager · Retail Group
January 2025
"The handover process was the most thorough I have experienced in an AI project. The sessions with our technical team were not perfunctory — they went until questions ran out. Six months later we have made several modifications ourselves, which says something about the quality of the documentation."
Aditya Mehta
CTO · Healthcare Technology, Singapore
February 2025
"I appreciated that the scoping conversation included a frank discussion of where the forecasting approach would have genuine value versus where it would add complexity without meaningfully improving our decisions. That kind of guidance at the start shaped the whole engagement positively."
Lin Kai
VP Strategy · Energy Sector
December 2024
Challenge
The organisation was using manual spreadsheet-based volume projections that required significant analyst time and were consistently underperforming against actuals during peak periods. The data existed but had not been systematically applied to forecasting.
Solution
Following a data quality assessment that identified two missing feeder tables, Tensora developed a time series model incorporating seasonal decomposition and external calendar factors. The model was designed to be retrained monthly by the in-house analytics team.
Outcome
Forecast error over a three-month post-deployment period averaged 8% below the baseline. Analyst time spent on projections dropped substantially. The model has been running in production for eight months without structural changes.
Challenge
A professional services firm handling a high volume of prospective client enquiries needed to qualify and route them more efficiently. The existing intake process required senior staff time for questions that could be answered systematically.
Solution
Tensora designed and built a domain-specific conversational assistant covering intake qualification, service-type identification, and routing. Failure handling was extensively tested to ensure graceful escalation for out-of-scope queries.
Outcome
Within three months of launch, a measurable portion of initial client contacts were handled without requiring senior staff involvement at the qualification stage. The management interface allowed the team to update content independently without engineering support.
Challenge
Years of internal research documents, regulatory guidance, and product documentation existed in disconnected repositories. Analysts spent significant time manually cross-referencing sources that had implicit relationships but no structured linkage.
Solution
Tensora designed an ontology covering financial products, regulatory entities, and procedural concepts. Entity extraction was applied to 6,000+ documents, with iterative validation sessions involving subject matter experts from the client team.
Outcome
The graph now serves as a foundation for the institution's internal search and compliance checking workflows. Analysts reported a measurable reduction in time required for cross-source research tasks. Three downstream AI applications have been built on the graph since delivery.
If reading these accounts has raised a question about whether your own situation might be a fit, we are glad to talk it through.
9 Battery Road, #28-01
Straits Trading Building, Singapore 049910
Monday – Friday, 9:00 AM – 6:00 PM SGT
Member organisation contributing to Singapore's applied AI practitioner community since 2019.
Data processing procedures independently reviewed against Singapore Personal Data Protection Act requirements.
Recognised under the Enterprise Development Board's capability development initiative for Singapore SMEs.
The first conversation carries no commitment and no sales pressure — just a candid discussion about what your data situation suggests is feasible.
Start a Conversation