— Why Kilau Neural
What we do differently,
and why it matters.
Our approach is shaped by what consistently goes wrong in ML engagements. We have built our practice around the failure modes we have observed, and we are direct about the trade-offs.
Back to Home— Competitive Position
Six things we do that shape the outcome
Problem-first scoping
Before any technical work begins, we produce and agree a written problem statement. This document defines what success looks like and what is out of scope — and it constrains our work throughout the engagement.
Evaluation as a deliverable
We build evaluation infrastructure before we build models. This includes test sets drawn from real production data, written evaluation protocols, and integration with your deployment process.
Documentation written for your team
Our documentation is written for the technical team that will maintain the system — not for a hypothetical reader with our specific background. We test this by reviewing documentation with your team before the engagement closes.
Malaysian regulatory grounding
We understand PDPA obligations, ASEAN AI governance directions, and the data residency preferences of Malaysian financial and public-sector clients. These are not addons — they are part of how we scope and build.
Team engagement, not isolated delivery
We work alongside your team throughout the engagement. Knowledge transfer is structural — your engineers participate in design decisions, code reviews, and evaluation sessions so they understand the system being built.
Transparent, fixed-scope pricing
Our engagement prices are stated clearly. Scope changes follow a documented amendment process with agreed implications on cost and timeline. We do not operate on open-ended retainers or time-and-materials arrangements.
— In Depth
Each advantage examined
01 — Expertise
Practitioners, not generalists
The principals at Kilau Neural have built and maintained production ML systems in Malaysian operating conditions — not in academic settings or as side projects at large technology firms. We understand the specific constraints: smaller data volumes than the benchmarks assume, regulatory requirements that US and European tooling does not handle by default, and internal teams with varying levels of ML familiarity.
We have worked on credit scoring, document processing, churn prediction, inventory forecasting, and fraud detection problems — across financial services, retail, manufacturing, and logistics sectors in Malaysia.
Expertise indicators
- Production ML experience in Malaysian financial services, retail, and logistics
- Direct knowledge of PDPA and ASEAN governance frameworks as they apply to data-driven systems
- Experience with small-to-medium data volumes — not calibrated to hyperscale benchmarks
- Prior advisory work with boards on technology governance and procurement decisions
Technical approach
- Tools chosen for maintainability, not for novelty or vendor alignment
- Evaluation infrastructure built before model training begins
- Data pipelines designed to run in the client's existing infrastructure where possible
- Monitoring and drift detection included as standard, not as optional additions
02 — Technology
Tools chosen for your team's maintenance burden
We do not default to the most sophisticated or most recent approaches. We select tools and architectures based on what your team can understand, maintain, and extend after the engagement ends. A simpler model that your team can reason about is usually preferable to a more performant model that is opaque to them.
Where we make technical choices that carry trade-offs, we document those choices and their implications. Your team should be able to revisit any architectural decision we made and understand why it was made.
03 — Service
Direct communication throughout
We communicate directly with the people doing the work — not through account management layers. The engineers and practitioners who scope your engagement are the same people who build it and present results. This reduces interpretation errors and speeds up decision-making when problems arise.
We respond to client enquiries within one working day. We produce written progress notes at agreed intervals, and we raise concerns in writing before they become problems.
Service commitments
- Direct access to practitioners — no account management layer
- Written progress notes at agreed intervals throughout the engagement
- One working day response time on client communications
- Structured handover session with documentation review at engagement close
Pricing structure
Fixed-scope pricing. Amendments documented in writing.
04 — Value
Stated prices, written scope, no surprises
Our engagement prices are published on our Solutions page. The scope of each engagement is agreed in writing before any work begins. Additional scope requests are handled through a written amendment process that identifies the cost and timeline implications before any additional work is authorised.
We do not offer open-ended retainers or time-and-materials arrangements. This protects you from scope drift and budget unpredictability — and it keeps us focused on delivering defined outcomes rather than billing hours.
05 — Results
Outcomes defined before the engagement starts
We begin every engagement by defining what a successful outcome looks like — in terms that can be measured. For a Capability Build, this means agreeing the business metric the model will be evaluated against and the threshold required for deployment. For an Evaluation Framework engagement, it means agreeing what constitutes a functioning evaluation system.
We do not declare success based on our own assessment of quality. Outcomes are measured against the criteria agreed at the start of the engagement.
Results approach
- Success criteria defined and agreed in writing at engagement start
- Evaluation against real production data, not synthetic benchmarks
- Capability Build closures include a structured handover with your technical team
- Follow-up note delivered after Senior Briefing sessions for reference and sharing
— Comparison
Typical AI consulting versus our approach
These are patterns we observe frequently, not characterisations of any specific firm.
| Feature | Typical Consulting Approach | Kilau Neural |
|---|---|---|
| Scope definition |
Broad mandate, defined iteratively during engagement
|
Written problem statement agreed before work begins
|
| Evaluation |
Added at the end if time permits; often omitted
|
Built before model training; included as a core deliverable
|
| Documentation |
Written for the consulting team; opaque to client's engineers
|
Written explicitly for your internal team's maintenance needs
|
| Pricing model |
Time-and-materials or open-ended retainer
|
Fixed-scope, published prices; amendments by written agreement
|
| Local regulatory knowledge |
Often generic or EU/US-calibrated advice
|
PDPA and ASEAN framework awareness factored in by default
|
| Knowledge transfer |
Informal; creates dependency on continued external support
|
Structural throughout; your team participates in design and review
|
— What Sets Us Apart
Distinctive elements of how we work
Written problem statements before any code
We require a written and agreed problem statement before any technical work begins. This is not a formality — it is the document against which we evaluate success. It prevents scope drift and gives your leadership a clear record of what was commissioned.
Evaluation-first development sequence
Our development sequence requires that evaluation infrastructure — test sets, evaluation protocols, and measurement tooling — is in place before model training begins. Most practices treat evaluation as a late-stage step. We treat it as a precondition.
In-person senior briefings, with written follow-up
The Senior Briefing is conducted at your premises, in person. We do not offer this as a remote session. Following the briefing, we deliver a written note the next working day — a document suitable for distribution to board members who were not present.
Deliberate engagement volume
We take a small number of engagements at any one time. This is not a commercial constraint — it is a quality decision. ML systems built under divided attention degrade faster and document less thoroughly. Our engagement limit protects the quality of your work.
— Milestones
Practice milestones and recognitions
Engagements completed
Years in practice
Engagements with written documentation
Industries served
MDEC Digital Transformation Partner
Listed under the Malaysia Digital Economy Corporation's network of AI practice partners supporting Malaysian enterprise digitalisation.
2024 ASEAN AI Practitioner Recognition
Recognised by the ASEAN AI Practitioners Forum for responsible AI deployment practice in Southeast Asian enterprise contexts.
Malaysian Institute of Accountants — Technology Panel
Contributing member of the MIA Technology Advisory Panel, providing input on AI governance standards for accounting and finance professionals.
— Take the Next Step
Ready to discuss a specific problem?
We respond to every enquiry with a brief note explaining whether and how we can help. No obligation, no extended discovery process.
Get in Touch