Engineering discipline in AI

§ 01 — Why Choose Anggun Systems

AI systems you can inspect, evaluate, and maintain without us in the room

The quality of an AI engagement is not visible at the point of sale. It becomes apparent months later, when something needs to be explained to an auditor, when the original team has moved on, or when the system produces an output that no one can trace. These pages explain how we try to prevent those problems.

Return to Homepage

§ 02 — Core Advantages

What you receive that others do not commonly provide

Each item below reflects a specific practice we follow on every engagement. These are not aspirational commitments; they are conditions of the work. Where a project constraint prevents us applying one, we say so at scoping.

Written evaluation reports

Every system we deliver is accompanied by a written evaluation report. It includes the methodology, the test set used, the results, and a section on known failure modes. You receive this document as part of the engagement, not as an optional add-on.

You own the deliverable

The model, the retrieval system, or the policy document is yours on delivery. We do not retain rights to it, host it on our infrastructure as a dependency, or require ongoing payment to keep it operational.

Stated limitations

We document what the system does not do as carefully as we document what it does. If the system has a residual error rate that matters to your operations, that rate is measured, stated, and explained before you decide whether to deploy.

Fixed scope and timeline

All three service lines have defined timelines: four weeks for policy drafting, a defined period for retrieval systems, and nine to fourteen weeks for custom model development. The timeline is agreed in writing before work begins.

Malaysian regulatory familiarity

Our engineers understand the Personal Data Protection Act, sector-level guidance from Bank Negara Malaysia and the Securities Commission, and the practical compliance requirements that affect our clients' industries. This is incorporated into the work by default.

Knowledge transfer included

We work with your data team, not around them. By the end of the engagement, your team understands how the system was built, what it does, and what they need to monitor. This is not a paid extra; it is part of every engagement.

§ 03 — Expertise in Detail

Professional expertise

Senior engineers on every engagement

The engineers at Anggun Systems have accumulated practical experience in production AI systems across Malaysian financial services, manufacturing, and professional services contexts. This is not academic experience; it is experience gained from systems that ran in organisations with real compliance requirements and real operational constraints.

Every engagement is conducted by engineers who have worked with these systems before. We do not use the engagement as a learning exercise for junior staff. The engineers who assess your problem are the same engineers who build the solution.

Process and method

Defined methodology, stated at outset

Each engagement follows a structured process: initial scoping, problem definition, data review, build, evaluation, and handover. The stages are described to the client at the start of the engagement. If a stage reveals something that changes the direction of the work, we discuss the implications before proceeding, not after delivering an unexpected result.

The evaluation stage is not a formality. We use a held-out test set, report results on standard metrics relevant to the task, and include a section documenting the cases where the system made errors and why.

Technology and tooling

Appropriate tools, not the newest tools

We select tools based on what is appropriate for the problem, the client's existing infrastructure, and the maintenance burden we are leaving behind. We do not advocate for a particular platform or vendor because we have a commercial relationship with them.

For retrieval-augmented systems, we use open-source components where they are suitable and document the full dependency stack so that your team can maintain it. For custom model development, we work with the frameworks that have the best established tooling and the largest maintenance communities.

Value and pricing

Stated prices, no hidden phases

Our prices are stated on our solutions page: MYR 2,280 for custom model development, MYR 1,290 for the retrieval system, and MYR 510 for AI policy drafting. These are the prices for the full engagement as described. There are no phases that require additional purchase to complete the deliverable.

The retrieval system engagement includes ninety days of post-deployment tuning at the stated price. This is not an optional support package; it is included because the first ninety days of a deployed retrieval system are when the most useful tuning data becomes available.

§ 04 — How We Compare

Anggun Systems compared with typical AI providers

This comparison reflects the most common patterns we observe in the broader AI consulting market. It is not a characterisation of any specific firm.

Practice area Typical providers Anggun Systems
Evaluation methodology Informal or not disclosed Written, included in delivery
Error rate disclosure Not measured or not shared Measured and stated in writing
Handover documentation Minimal or absent Included as a core deliverable
Malaysian regulatory context Generic advice, not jurisdiction-specific PDPA and sector guidance incorporated
Post-delivery dependency Requires ongoing provider involvement System owned and operable by client
Knowledge transfer Available at additional cost Part of every engagement

§ 05 — What Sets Us Apart

Distinctive features of the Anggun Systems practice

Built for the Malaysian market

We are based in Kuala Lumpur, operate within Malaysian business norms, and understand the specific regulatory landscape our clients operate in. This is not a general-purpose AI practice retrofitted to a Malaysian audience.

Policy drafting as a first-class service

Few AI providers include policy work as a distinct service line. We offer AI policy drafting because we have observed that many Malaysian firms have operational AI use that has outpaced their written governance. A capable internal policy is not a compliance formality; it is a management tool.

Three service lines, not an open menu

We have deliberately limited our practice to three service lines because they are the ones we can deliver to a consistent standard. We do not accept every AI project that is offered to us. When a problem falls outside our practice, we say so at the first conversation.

Ninety-day tuning included in retrieval work

The most useful performance data for a retrieval system comes from real usage. We include ninety days of post-deployment tuning in the retrieval system engagement because this is when the system can be meaningfully improved, not as a sales lever after delivery.

§ 06 — Practice Milestones

How our practice has developed

38

Completed engagements

6

Industry sectors served

91%

Avg. retrieval citation accuracy

100%

Deliverables with documentation

Figures reflect engagements completed as of April 2025.

§ 07 — Next Step

Discuss whether your problem fits our practice

We spend the first conversation understanding your situation before discussing any solution. Contact us to arrange a scoping discussion.

Get in Touch