What Clients Say After Working With Tensaro
Feedback from engineering leads and operations directors at mid-sized organisations across Malaysia who engaged Tensaro for AI compute advisory work.
Back to HomeEngagements completed
Average client satisfaction
Repeat engagement rate
Years advising in Malaysia
Client Testimonials
We came in not quite knowing what questions to ask. The Readiness Assessment gave us a very concrete picture — five dimensions, each scored, each with a next action attached. The scorecard format meant I could take it straight to the CTO without having to translate anything.
April 2025
The GPU Workload Planning Workshop was exactly the session our team needed before we started talking to vendors. We walked through memory budgeting and batch size tradeoffs with real workload examples. The capacity templates have been used since in at least three internal planning documents. I'd have liked one more hour on inference vs training trade-offs, but that's a minor point.
May 2025
We ran the Infrastructure Retainer across Q1 of this year. The shared decision log was genuinely useful — not just during the engagement but afterwards. When the next procurement cycle came up three months later, we had documentation of every trade-off we had already reasoned through. Syafiqah's input on vendor negotiation timing was particularly practical.
March 2025
What distinguished Tensaro from the alternatives we considered was the vendor-neutral stance. We had spoken to two other advisory firms both of which had obvious hardware affiliations. Tensaro's analysis started from our workload profile, not a product sheet. The assessment confirmed some things we suspected and surfaced two issues we had genuinely missed.
April 2025
The workshop was well structured. We went in expecting a fairly high-level session and came out with actual templates we could use. Ahmad walked through the memory bandwidth calculations in a way that made sense even for team members who hadn't dealt with GPU specs before. Fixed price and clear scope meant no budget anxiety going in.
May 2025
We started with the Readiness Assessment and that gave us enough clarity to go straight into the Workload Workshop four weeks later. Tensaro's sequential engagement model makes sense — each step builds on what the previous one uncovered. The written outputs at each stage meant we had a clear paper trail for internal approval processes.
April 2025
Engagement Outcomes
Three documented engagement stories showing how the advisory process moved from initial question to practical outcome.
Financial Services Firm, Kuala Lumpur — Readiness Assessment
Challenge
A 180-person financial services company had allocated budget for an AI compute cluster but had not yet determined whether their existing data pipelines or team could support the planned workloads. The engineering lead needed a structured answer before presenting to the board.
What We Did
Tensaro ran a three-week AI Readiness Assessment. Two workshops mapped data maturity, team skills, and the three candidate workloads the company had identified. We returned a scorecard and a summary report that prioritised two of the three workloads as viable for a first deployment.
Outcome
The board approved the first deployment based on the report. The third workload was deferred — a decision the team said saved approximately two months of rework. The company went on to complete a GPU Workload Workshop six weeks later.
"The scorecard format was exactly what we needed. It gave us a structured argument for the board rather than a list of opinions."
— Head of IT Operations, Kuala Lumpur
Manufacturing Company, Shah Alam — GPU Workload Workshop
Challenge
A mid-sized manufacturer was evaluating two GPU configurations for a quality inspection AI workload. The engineering team had different assumptions about memory requirements and throughput targets, and vendor quotes varied by RM 340,000 between options. They needed an independent framework for the decision.
What We Did
We facilitated a GPU Workload Planning Workshop with the engineering team. Using vendor-neutral benchmarks and the team's actual image resolution and inference frequency data, we modelled memory consumption and throughput for both configurations. The workshop produced a capacity template tailored to the workload.
Outcome
The team selected the lower-cost configuration with confidence, supported by documented reasoning from the workshop. The evaluation checklist we provided was used to verify vendor claims during the procurement process. The deployment went live eight weeks after the workshop.
"Having a neutral framework meant we could push back on vendor claims with actual numbers. The workshop paid for itself many times over."
— Director of Engineering, Shah Alam
Technology Company, Cyberjaya — Infrastructure Retainer
Challenge
A 300-person technology company was expanding its AI compute across three teams with different workload profiles. The engineering director needed a structured process for reviewing utilisation, making procurement decisions, and keeping a coherent architecture as the infrastructure scaled quarter by quarter.
What We Did
We ran a three-month Infrastructure Retainer with monthly review calls. A shared decision log was maintained from day one. Each review session covered utilisation data, upcoming procurement timing, and any architecture decisions in progress. Roadmap updates were produced at each cycle.
Outcome
By the end of the retainer, the company had a documented decision log covering fourteen architecture decisions, three procurement timing choices, and one configuration that was deferred based on utilisation data. The director described the log as the most useful thing they took from the engagement.
"The decision log was something we didn't know we needed until we had it. Going into the next quarter, we are not repeating discussions we already resolved."
— VP Technology, Cyberjaya
Reach Tensaro Directly
Phone
+60 3-2710 4683Office
Level 28, Menara Binjai
2 Jalan Binjai, KL
Hours
Mon–Fri: 9am–6pm
Sat: 9am–1pm
Add Your Organisation to the List
If you are working through AI compute planning decisions and would like a structured, vendor-neutral perspective, we are happy to start with a brief scoping conversation.
Get in Touch