Client Perspectives
What clients say
about working with us.
Perspectives from organizations across Hong Kong that have worked with Synthwave on AI design and data projects.
← Back to HomeAvg. satisfaction
Clients served
Years in HK
Repeat engagements
What People Say
In their own words.
"We engaged Synthwave for a conversational AI project — an internal Q&A tool for our operations team. What stood out was how much time they spent understanding our actual workflows before building anything. The end result reflected how our team really communicates, not a generic bot template."
"The data pipeline work they did for us took about three months, which was longer than initially estimated — but they were transparent about why at each stage. The quality of what was delivered, and the documentation, was genuinely impressive. Our internal team can now maintain it without needing us to call Synthwave every time something changes."
"We came to Synthwave with an idea we'd been thinking about for a while but weren't sure was viable. Their Proof of Concept service was exactly what we needed — two weeks, working prototype, honest feedback. The feedback wasn't entirely what we'd hoped, but it saved us from a much larger investment in the wrong direction."
"Marcus and his team have a rare combination — they understand the technical side deeply but can explain things in plain language to non-technical stakeholders. That made the internal communication around this project much easier than we expected."
"I appreciated that they were upfront when our initial data quality wasn't good enough for what we wanted to do. Instead of proceeding and billing us for something unlikely to work, they helped us understand what we needed to fix first. That kind of guidance is worth a lot."
"The conversational agent they built for our customer service function now handles around 60% of initial enquiries without escalation. More importantly, the ones that do escalate to our team are better prepared — the handoff includes context that our staff actually find useful."
Case Studies
Project snapshots
from recent work.
Internal Q&A System for a Financial Services Firm
A 200-person firm's staff spent significant time searching internal policy documents and escalating questions to a small compliance team. The team needed a way for staff to get reliable answers quickly without overloading compliance resources.
We designed a conversational system grounded in the firm's policy documents, with clear escalation paths to human reviewers for edge cases. Dialogue mapping was done collaboratively with compliance team members to capture their domain knowledge.
Approximately 70% of staff queries resolved without escalation within the first month of deployment. Compliance team reported a meaningful reduction in routine questions, freeing time for higher-complexity cases. Engagement duration: 7 weeks.
"The time savings for our compliance team were real and visible within weeks." — Operations Director
Data Infrastructure Rebuild for a Regional Retailer
Data from seven retail locations, an e-commerce platform, and a warehouse system was stored in incompatible formats. Reporting required manual reconciliation every week — a process that was error-prone and time-consuming.
We designed a unified pipeline architecture that ingested from all three source systems, applied standardized transformations, and loaded into a central data warehouse. Quality checks were built in at each stage with alerting for anomalies.
Weekly reporting process reduced from 12+ hours to under 2 hours. Data reconciliation errors eliminated. The retailer's analytics team now has a reliable foundation for further work. Engagement duration: 11 weeks.
"We can actually trust our numbers now. That sounds basic but it changes everything." — Head of Analytics
Document Classification PoC for a Legal Services Firm
The firm wanted to explore whether AI could automatically classify incoming client documents by type and priority, reducing manual triage time. They had no prior experience with AI and needed to understand the realistic potential before committing budget.
Two-week PoC using a sample of historical documents provided by the firm. We built a classification prototype and evaluated its performance against human categorization on a held-out test set. Results were analyzed and presented with full transparency about limitations.
Classification accuracy reached 84% on the test set — useful but below the threshold the firm needed for an autonomous system. We recommended a human-in-the-loop design instead, which the firm subsequently engaged us to build. Engagement duration: 2.5 weeks.
"The PoC gave us the clarity we needed. Honest findings, no padding." — Innovation Lead
Credentials & Recognition
A practice built on
verifiable track record.
Shortlisted · Best AI Application
Information security management across all projects
Hong Kong Computer Society professional membership
Post-project survey average, 2022–2025
Ready to Explore?
Interested in working
together?
We're always happy to have an initial conversation about your situation — without commitment or obligation on either side.
Get in Touch