Skip to content
Centrio

Technology consultancy · Kuala Lumpur

Software, AI and cloud,built to work, not to demo.

We design, build and operate custom applications, AI systems, cloud infrastructure and data platforms for organisations that need them working, not demonstrated.

At a glance

Who you are dealing with

A technology consultancy in Kuala Lumpur. We design, build and operate software, AI systems, cloud and data platforms for organisations that need the result to hold up under real load.

207

code repositories migrated in a single enterprise DevOps programme

projects
127
repositories
207
  • 6

    engineering practices, from custom software to legacy modernisation

  • 8

    industries delivered into, including government and capital markets

Capability

Six practices. Most engagements draw on two or three.

  • 8–16 weeks

    Custom Software Development

    Web and mobile applications, APIs and content platforms built around your business logic.

    • Next.js
    • React
    • TypeScript
    • PostgreSQL
  • 10–20 weeks

    AI & Intelligent Systems

    Retrieval-grounded assistants, agents with permissioned tools and document intelligence, with cost governance built in.

    • RAG
    • Tool calling
    • Azure AI
    • OpenAI
  • 6–14 weeks

    Cloud & DevOps

    Cloud-native architecture on Azure and multi-cloud, deployment pipelines, containers and infrastructure as code.

    • Azure DevOps
    • Kubernetes
    • Terraform
    • Docker
  • 6–12 weeks

    Data Management & Analytics

    Real-time feeds, ingestion pipelines and warehousing, turned into the reporting a business decides with.

    • BullMQ
    • Redis
    • SQL
    • Power BI
  • 10–18 weeks

    Digital Sales & Commerce

    Omni-channel commerce, customer platforms and the digital operations that connect a brand to its market.

    • Commerce
    • Entitlements
    • Payments
    • Campaigns
  • 12–24 weeks

    Legacy Modernisation

    Aging systems moved onto modern foundations, run in parallel until the new one has carried real traffic.

    • Migration
    • Parallel run
    • Cutover
    • Rollback

Our platform

Centrio One. Configured, not constructed.

Modules that share one core, one data layer and one control surface. Configured to your business and live in weeks rather than quarters.

Modules go live in 2 to 6 weeks

Centrio One/CRM

2,326

contacts managed

94

leads converted from automated follow-up

ContactStageScoreChannel
Operations leadQualifiedHotWhatsApp
ProcurementLeadWarmEmail
Store ownerCustomerHotWhatsApp
Finance managerLeadColdWeb form

Campaign delivery

Sent
Delivered
Read
Failed

Centrio Core · included with every module

  • Metering

    Every model call recorded with provider, feature, tokens and exact cost

  • Attribution

    Spend rolled up by workspace, feature and template

  • Tool governance

    Callable tools split between customer-facing and internal-only

  • Master switch

    Disables inference everywhere while keeping every setting

Shared across every module

  • Single contact graph
  • Shared knowledge index
  • One identity model
  • One cost ledger
  • One audit trail
  • One admin surface

A second module is configuration against the same core, not a second implementation project.

Evidence

What the systems handle.

Measured figures from systems we run for clients, with the time window named where it applies.

Customer engagement

96.0%

resolved with no human agent

messages handled in 30 days
18,903
customer contacts managed
2,326
leads converted from automated follow-up
94

Content production

US$0.31

measured cost per finished asset

finished assets in 30 days
3,124
brands run from a single seat
5
AI calls, each metered and costed
8,623
We can tell you what a single conversation or a single published post costs, because it is metered at the point the call is made.

Delivered work

What our team has delivered

  • 2024

    Government · Environment & forestry

    TAPAL 21

    Monitoring, evaluation and reporting for a regional forestry authority in Central Kalimantan, moved off spreadsheets onto role-based workflows with a full audit trail.

    TAPAL 21/Monitoring and evaluation

    Reporting period

    UnitIndicatorsStatus
    Unit AApproved
    Unit BIn review
    Unit CApproved
    Unit DSubmitted
    Unit EDraft
    • Public sector
    • Audit trail
    • Role-based access
  • 2024

    Mining

    Enterprise DevOps migration

    A fragmented engineering estate at a major mining corporation, mapped and moved onto one managed DevOps platform with templated pipelines.

    • 127 projects
    • 207 repositories
    • Azure DevOps
  • 2023

    Telecommunications

    Cloud infrastructure & CDN

    Storage and content delivery for a telecommunications holding company, defined as code, sized to real demand and costed from day one.

    • 200GB object storage
    • 500GB CDN
    • Cost baseline
  • 2019

    Finance & capital markets

    IPO platform & live market feed

    A bilingual corporate platform for a publicly listed manufacturer, with real-time stock data and a disclosure workflow built around the regulator.

    • Bilingual
    • Real-time feed
    • Regulator-compliant
  • 2019

    Direct sales

    Membership & commission platform

    A genealogy model and a deterministic commission engine, where the same inputs always produce the same payout.

    • Binary structure
    • Bonus engine
    • Multi-role
  • 2016–2019

    Multiple sectors

    Corporate platform programme

    Custom content and corporate platforms across manufacturing, logistics, automotive, education and legal, including national-scale event platforms.

    • Custom CMS
    • Event scale
    • Long-run maintenance

Trusted by

Clients and industries

Our team has delivered for national institutions, listed companies, government ministries and global brands.

Industries served

  • Government & public sector
  • Mining & energy
  • Finance & capital markets
  • Automotive
  • Education
  • E-commerce & retail
  • Legal & compliance
  • Environment & forestry

Audit trails, bilingual access, accessibility and public accountability are familiar ground, not a first attempt.

Certified team

Certified on the platforms we build with.

Our engineers hold vendor certifications across AI and cloud.

  • ClaudeClaude Certified
  • Microsoft AzureMicrosoft Certified: Azure
  • Amazon Web ServicesAWS Certified
  • Google Cloud Certified
  • Alibaba CloudAlibaba Cloud Certified

Engineering

The stack changes per engagement. The standards do not.

  1. 01

    Version control and review

    Every change through a reviewed pull request, with protected main.

  2. 02

    Automated delivery

    Build, test and deploy pipelines from day one. No manual production deploys.

  3. 03

    Environment parity

    Development, staging and production defined as code, so they do not drift.

  4. 04

    Observability by default

    Structured logs, error tracking and queue visibility ship with the product.

  5. 05

    Security posture

    Least privilege, secrets in a managed vault, dependency scanning in the pipeline.

Engagement

Five stages. A decision point after each.

You are never asked to commit to the whole thing to find out whether it will work.

  1. 1

    Stage 1: Discover

    2–4 weeks

    We assess your data, systems and processes, then produce a costed roadmap. You keep it whatever you decide next.

  2. 2

    Stage 2: Design

    2–3 weeks

    Solution architecture, integration, security and data handling, and a delivery plan with dates you can hold us to.

  3. 3

    Stage 3: Build

    6–16 weeks

    Two-week increments, with working software demonstrated at the end of each one.

  4. 4

    Stage 4: Deploy

    2–4 weeks

    Migration, pipelines, acceptance testing, training and handover documentation written to be used.

  5. 5

    Stage 5: Run

    Ongoing

    Monitoring, small enhancements and a named engineer who knows your system. Optional, and cancellable.

Platform modules: a one-off implementation plus a monthly fee tiered by volume. Engineering projects are quoted after discovery. Infrastructure and AI usage are recharged at cost.

Why the cost works

Pay for configuration, not construction.

Much of what Malaysian businesses are quoted for AI and custom software is construction cost, a learning curve and a margin on consumption. We remove all three.

The market reference

RM500,000 – RM2,000,000

Typical initial investment for an AI system in Malaysia, per the Malaysian Productivity Corporation. Most of it is construction.

  • Construction cost
  • Learning curve
  • Margin on consumption
Centrio One/Cost ledger

8,623

AI calls metered

US$0.31

Measured cost per asset

FeatureProviderTokensCost
Desk repliesOpenAIMeteredPer call
Intent classificationAzure AIMeteredPer call
Studio rendersOpenAIMeteredPer call
Scheduled analysisAzure AIMeteredPer call
  • We have built it before

    The systems we propose are ones our team has delivered and still operates.

  • We start small on purpose

    A short discovery gives you a costed roadmap before you commit to a build.

  • We show the cost

    Infrastructure and AI usage recharged at cost, with the handling fee visible.

  • We ship in weeks

    Platform modules go live in two to six weeks. Custom builds show working software every fortnight.

  • We keep humans in control

    Explicit handover to your team, approval steps where they matter, and a switch that stops it.

  • We are local where it counts

    Your contract, service agreement, account manager and first call are all in Kuala Lumpur.

If a project is not right for you, we will say so. A bad engagement costs us more than a lost one.

Next step

Start with a conversation.

Tell us the number you are trying to move and what it costs you today. We will tell you honestly whether we can help, and roughly what it would take.

What to bring

  • The process or system causing the problem
  • Roughly what it costs you today, in people or time
  • Who in your business would own the outcome
  • Any deadline the work has to land before