Reviews from recent engagements
Dates reflect when the engagement concluded. Names used with permission.
We came in genuinely uncertain whether our data was good enough to support any AI work. The readiness assessment gave us a clear picture — some parts were better than we thought, a few areas needed work. That honesty was refreshing. We now have a prioritised plan and we're actually moving on it.
The custom model Synthrex built for document classification has saved our team a significant amount of manual processing time. The model performs consistently, the documentation is thorough, and our engineers were able to take it over without issues. The engagement stayed on timeline and on budget.
We needed help framing our AI approach for the board — not just a wish list, but a realistic plan with governance and investment rationale. The strategy engagement delivered that. The team was direct, they pushed back on assumptions in useful ways, and the final deck was something we could actually present with confidence.
What stood out was how well Synthrex understood the regulatory sensitivities around patient data before we even had to explain them. The anomaly detection model they built operates within a framework that our compliance team was comfortable signing off on. That kind of awareness is genuinely rare.
The readiness assessment was useful partly because it confirmed what we suspected — our forecasting data was messier than it needed to be. Rather than just flagging that, Synthrex gave us a specific remediation sequence. Six months on, we've addressed two of the four priority items and we're seeing better results from our existing tools.
We had a previous vendor who delivered a model that nobody in the organisation knew how to maintain after they left. Synthrex was visibly different — knowledge transfer was built into every phase, and our team left the engagement genuinely capable of working with what was built. That matters more than the model itself in the long run.
A closer look at three engagements
These are representative examples based on the types of work we do. Details have been generalised to protect client confidentiality.
Logistics firm — inventory forecasting
A regional logistics company was carrying 30% more buffer stock than the business needed, driven by unreliable demand predictions from a legacy spreadsheet model.
We started with an assessment to understand the data's shape and quality. After cleaning and restructuring two years of order history, we built a gradient boosting model with weekly retraining. Integration with their WMS took three weeks.
Buffer stock reduced by approximately 22% over the following quarter. Forecast accuracy improved from around 61% to 84% on weekly SKU-level predictions. The model has been running without intervention for six months.
Mid-size manufacturer — board AI strategy
A Selangor-based manufacturer's board had asked the exec team to present an AI strategy. The exec team had ideas but no framework to evaluate them or estimate investment requirements.
Five weeks of structured advisory: landscape review, use case prioritisation workshops with the exec team, vendor evaluation criteria, a governance framework, and a phased three-year roadmap with indicative costs.
The board presentation was well-received. The company approved Phase 1 of the roadmap (quality inspection automation) in the same meeting. Synthrex was subsequently engaged for the model development phase.
Insurance firm — AI readiness reality check
A regional insurer had received a vendor proposal for a claims processing AI system. Before committing, they wanted an independent view on whether their data and infrastructure could actually support it.
Two-week assessment covering data architecture, claims data quality, team capability, and a review of the vendor's technical proposal. We mapped what would need to change before implementation could succeed.
We identified two prerequisite data quality issues that would likely have caused the vendor engagement to fail. The insurer paused the procurement, addressed the data issues over four months, then proceeded. They avoided a costly failed implementation.
Talk to us about your situation
A short conversation is usually all it takes to understand whether we're the right fit for what you're working on.
Ready to have that conversation?
Send us a message with a little context about your organisation and what you're considering. We'll come back to you with our honest thinking.
Get in Touch