Intelligence Without Limits
Ontilus designs and engineers bespoke AI operating systems for complex, multi-entity operators. We turn fragmented operational data into one live source of truth — and one engine that tells your leadership what to do next.
Engineering AI systems for multi-entity operators worldwide
Capabilities
Live operational streams — transactions, inventory, labour, finance — unified across every legal entity and reconciled against your own close reports.
Demand forecasting, pricing optimisation and variance detection that run before the day starts — not after it ends.
It answers the daily why and recommends the next best action. A strategic partner in the business, not a static report.
Role-based access, entity-level isolation and audit-ready logging — designed from day one for regulated, multi-company groups. Built in Malaysia, for groups running several entities on separate systems.
The Ontilus platform
We don't hand over a dashboard and leave. We build the layer that sits between your operations and your decisions — on your data, in your entity structure, under your governance.
Applications
Command centers · Agents · Workflows · Executive views
Ontilus AI engine
Reasoning · Memory · Forecasting · Recommendation
Data intelligence layer
Identity graph · Telemetry · Analytics · Reconciliation
Infrastructure
Isolated VPC · Encrypted · Audit-logged · Regional
The systems we build
Every engagement is bespoke, but almost all of them resolve into these two — and they're designed to fuse into a single engine. The Command Center below is shown answering four different sectors' questions; the figures are illustrative, the modules are what stay constant.
System 01
Real-time operations, AI decisioning and executive intelligence on one live surface — from the front line to the board pack.
Net revenue — today
RM 428,650
▲ 6.2% vs target
Orders
7,412
▲ 3.1% WoW
Average order value
RM 57.83
▲ 2.9% WoW
Stores above target
24 / 33
9 need attention
Gross margin by channel
Cost intelligence · target 42.0%
Margin & stock alerts
Detected in the last 7 days
Stock is uneven, not short — 6 of the 18 shortfalls exist in another store
Morning briefing
Decision engine · generated 06:00 · for the Retail Director
Revenue is 6.2% ahead of target, but the mix is moving toward your two lowest-margin channels. Two actions are worth taking today.
Rebalance stock before reordering
Six of the eighteen shortfalls are sitting in a store that is not selling them. Transferring costs freight; reordering costs freight and working capital.
Est. impact +RM 38,000 working capital · confidence high
Reprice the 42 slow-moving lines now
Sell-through has been under 60% for six weeks. Marking down in-season recovers more than clearing at end of season, and the model puts the break-even at week nine.
Est. impact +RM 61,400 margin · confidence medium
Appointments — today
1,284
▲ 4.4% vs plan
Room utilisation
78.4%
▼ 6.6 pts vs target
No-show rate
9.2%
▼ 1.4 pts WoW
Branches reporting
12 / 12
All feeds healthy
Utilisation by branch
Operations · rooms and practitioners · target 85%
Scheduling & governance alerts
Open now · across all entities
Built from booking and operational data — clinical systems are read, never replaced
Morning briefing
Decision engine · generated 06:00 · for the Group Operations Director
Demand is ahead of plan while utilisation is 6.6 points below target — the group is turning work away in one place and idle in another. Two actions are worth taking today.
Move two sessions from Branch 01 to Branch 07
Branch 01 is running at 88% with a three-week wait; Branch 07 sits at 66% and is eleven minutes away. The waitlist overlaps both catchments.
Est. impact +RM 24,000 / month · confidence high
Run the overdue recall list into the unfilled slots
128 patients are past their review date and 34 slots are empty this week. Consent is current on 119 of them; the rest are excluded automatically.
Est. impact fills 26 of 34 slots · confidence medium
Billable utilisation
71.2%
▼ 3.8 pts vs target
Realisation
88.6%
▲ 1.2 pts MoM
Work in progress
RM 1.42M
31 days average age
Engagements at risk
6
of 84 active
Utilisation by practice group
Operations · billable hours against target 75%
Engagement risk
Cost intelligence · flagged against agreed scope
Scope drift is flagged against the estimate mid-engagement, not discovered at invoicing
Morning briefing
Decision engine · generated 06:00 · for the Managing Partner
Realisation is improving, but utilisation is 3.8 points short and it is concentrated in two practice groups. Two actions are worth taking today.
Raise a change order on Engagement 2214
It is at 142% of estimate with six weeks still to run. The scope moved in week three and was never papered; every hour past this point is written off by default.
Est. impact +RM 96,000 recoverable · confidence high
Move four Tax staff onto the Advisory pipeline
Tax is at 64.8% between filing seasons while Advisory is turning down work at 79.4%. Three of the four have the required qualification already.
Est. impact +RM 47,000 / month · confidence medium
Net sales — today
RM 428,650
▲ 6.2% vs target
Covers
7,412
▲ 3.1% WoW
Average spend
RM 57.83
▲ 2.9% WoW
Outlets above target
24 / 33
9 need attention
Food cost % against target
Cost intelligence · by entity · target 29.3%
Supplier price drift
Detected in the last 7 days
Ingredient stock auto-debits on every sale — alerts fire before the shortfall
Morning briefing
Decision engine · generated 06:00 · for the Group Operations Director
Group is tracking 6.2% ahead of target on revenue, but gross margin is being eaten from two directions. Two actions are worth taking today.
Approve the poultry supplier switch
Chicken thigh is up 9.4% at Supplier 04 while Supplier 07 has held price for 11 weeks. Same spec, same lead time.
Est. impact +RM 4,180 / month · confidence high
Re-staff cold larder for dinner service
Cold larder is 45% over target ticket time and is the largest driver of dinner table-turn delay across six outlets.
Est. impact +RM 2,600 / week · confidence medium
System 02
Unified customer intelligence and an engine that runs engagement on its own — so retention stops depending on who remembered to send the campaign.
A single 360° profile per customer, patient, student or client — one identity across every entity you operate.
Journeys that fire themselves, on time, every time — with consent tracked per customer, per channel.
External signal measured against internal return — so marketing spend is judged on contribution, not impressions.
Fused with the Command Center, operational and customer data become one predictive engine.
Ask the system
Your team already asks these questions. Today the answer takes three days and four spreadsheets.
›
How we deploy
Four phases, in the same order every time. Nothing goes group-wide until it reconciles against numbers your finance team already trusts.
Weeks 1–3
Where the pain actually is
Weeks 4–6
Implementation and iteration
Weeks 7–10
Fix, tune, then switch on
Weeks 11–12
We stay on it after go-live
What makes it land
Standardise the taxonomy. One naming convention across every system before we integrate.
Appoint one champion. A single internal owner for data validation, not a committee.
Three metrics first. Teams learn performance against target, cost on the day, and critical alerts. Everything else follows.
Security & governance
Multi-entity groups don't get to be casual about data. Separation, consent and audit trails are architecture decisions here, not settings we add later.
Transparent running costs
You'll know your monthly operating cost before you sign — inference, infrastructure and integration, itemised and sized to your actual transaction volume. Confirmed at technical scoping, not estimated after go-live.
TLS 1.3 in transit, AES-256 at rest, across every service.
Scoped by entity, brand and site. People see their scope, nothing else.
Hard separation between legal entities — not a filter on a shared table.
Full change history on every record and every access event.
Consent tracked per person, per channel. PII minimised by default.
Private VPC, regular key rotation, regional data residency.
Where we work
We're strongest where there are many locations, several legal entities, and data that has never been in one place.
Multi-outlet, multi-brand groups with mixed POS estates and central kitchens.
Inventory across channels, margin per SKU, and demand that moves weekly.
Scheduling, utilisation and compliance reporting under strict data handling.
Enrolment pipelines, cohort performance and cross-campus consolidation.
Utilisation, realisation and pipeline intelligence across practice groups.
Group reporting where every subsidiary runs a different stack.
Selected work
Sixty-second films from systems we've designed, built and shipped. Deeper technical walkthroughs are available on request.
Lending · CRM
Outreach CRM with live pipeline telemetry, batch identity matching and a manager command view for a lending caller operation.
Media · Editorial OS
Three publications on one desk: planning board, AI-assisted drafting and publishing flow for a media group.
Fitness · Mobile
Coach-led programs, video training and a workout journal — shipped to Android with cloud media delivery.
Telco · Analytics
Campaign and content analytics for a national telco's TV product, consolidated into one live dashboard.
Automotive · Creative ops
Eighteen platform-perfect renditions across six platforms in one pass — fully client-side, no backend.
Questions
If yours isn't here, ask us directly. We answer scoping questions before there's a contract in sight.
Talk to usBespoke AI operating systems. In practice that means a live data layer over your existing systems, predictive engines on top of it, and an interface your team actually opens every morning — plus the automations that remove manual work underneath. We're not reselling a product; we architect and build for your entity structure.
A dashboard reports what happened. Our systems explain why it happened and recommend what to do about it, then execute the parts that can be automated. The difference shows up in the morning briefing: a ranked, costed list of actions instead of twelve charts and an interpretation problem.
Yes — including the awkward ones. Where a system has webhooks or an API we integrate directly. Where it's on-premise or legacy we deploy local sync agents and batch ETL. Most groups we work with run three or four different systems across their entities, and that's the normal starting point, not a blocker.
Discovery runs three weeks, and you see working software from week four, reviewed with you weekly from then on. Twelve weeks to a live group-wide system, followed by hypercare. We deliberately validate against numbers your finance team already reports before anything scales — it's slower on paper and far faster in reality, because nobody has to unpick a bad rollout.
You do. Your data stays in infrastructure scoped to you, with residency where you need it. Commercial terms on the codebase are agreed up front — we'll tell you exactly what you own before you commit, not after.
Monthly operating cost breaks into three lines: AI inference, cloud infrastructure, and integration middleware. We size all three against your actual transaction volume during technical scoping and give you the number before you sign — including how it grows as you add sites.
Tell us what's slowing your operation down. We'll show you what the system would look like running on your data — before you commit to anything.