Held by the group. Inherited by every brand.
Specialists build the products. The group holds what makes them trustworthy: one standard, one way to rehearse decisions before they are made, and the engineering foundations underneath. Shared means reused where it fits. It does not mean one platform every brand must run on.
In shortEvery Owllys brand inherits four things from the group: the Owllys standard, Demo Forge (the enterprise rehearsal engine), decision-grade digital twins, and shared AI and engineering foundations.
The Owllys standard.
Six rules every brand ships to, whichever industry it serves. They are how a CFO can trust a recommendation, and how an operator keeps control of the floor.
- 01
Read-first.
Nothing changes in your systems on day one. We connect read-only to SAP, Oracle, Tally, WMS or MES first.
- 02
AI recommends. People govern.
Agents suggest; a named human approves every write. Autonomy is earned in simulation and can be revoked in one click.
- 03
Evidence on every answer.
Every recommendation ships as a record: decision, evidence, confidence, impact, approval. Audit trail and rollback built in.
- 04
Solvers for the maths.
Schedules, loads and allocations come from optimisation solvers. Language models explain; they don't invent the numbers.
- 05
Rehearse before you act.
New agents and decisions are proven in a simulated world before they touch yours. Demos use zero customer data.
- 06
A baseline in writing.
Value is measured against your numbers, agreed up front. If the number doesn't move, you don't scale.
The enterprise rehearsal engine.
A governed world engine where enterprises rehearse decisions before making them, and where AI agents learn, prove themselves and earn autonomy before they touch your business. Decision-grade by design: entities, flows, constraints and KPIs.
Business runs live. Every decision is take one.
- The demo that proved nothing.
- The pilot where the test environment was your business.
- The agent hired on faith.
Demo Forge exists so that consequential decisions get a rehearsal first.
Time is a control- Pause · Walk into any moment.
- Warp · A six-week pilot in one afternoon.
- Fork · Branch reality and compare futures.
- Replay · Same seed, same story: evidence reproduces.
Ledger: branch B ran 4.5 OTIF points above branch A on average after the event. Value = KPI(what you did) − KPI(what you almost did).
The world, in five layers
- L1Master data · Companies, sites, SKUs, BOMs, suppliers, machines: authored once, frozen, versioned.
- L2Behaviour · Seasonality, lead-time distributions and failure rates, calibrated to published industry bands.
- L3History · A deterministic simulator generates twelve months of transactions.
- L4Paperwork · Contracts, POs, quality certificates and emails, drafted against those exact numbers.
- L5Story seeds · Dormant risks planted in the world, waiting to be found.
Meridian: a conglomerate that doesn't exist, behaving like the ones that do. No customer data to leak or mask.
The academy
- 1Enroll · A curriculum per role, drawn from the scenario library.
- 2Train · Replay runs on forked worlds; failures cost nothing.
- 3Exam · Blind scenarios scored by a three-judge panel.
- 4Certify · Level, domain, date, and the evidence behind it.
- 5Watch · Scores tracked per domain; sustained regression proposes a demotion.
Autonomy is earned in simulation, granted by a human, revocable in one click.
The ledger
- Prove a pilot. Baseline signed in week 0, realised value against the do-nothing branch every week after.
- Defend a renewal. Twelve months of counterfactuals, with citations.
- Choose between futures. Fork, apply, run forward and compare before you commit capital.
The eval harness already gates SourceNest releases; the telemetry simulator descends from TransNest.
OpsNest in a simulated paper mill: a bearing fails at 02:04, and the plan heals by 02:07. The SAP emulator descends from OpsNest.
Semi3's Digital Twin Lab is powered by Demo Forge.
Your operation, running twice.
A digital twin mirrors the live operation: every order, machine, vehicle and pallet. Test the decision in the twin, compare the branches, then fire it in the real world. Autonomy moves at your pace.
Supply-chain twin: your operation, running twice. Test the plan first, then act, and compare the branches before anything is committed.
Plant twins: your factory, running twice, once in steel and once in software. Asset, process, line, product and plant twins take the risk first, so the plant only runs what already worked.
Digital Twin Lab: semiconductor twins, powered by Demo Forge, on semi3.com.
The parts no customer sees, and every customer relies on.
These are engineering patterns the brands reuse where they fit: how agents are governed, how answers are grounded, how numbers are computed and how systems are connected.
Governed agents
Approval gates, kill switches per agent and per tenant, autonomy that is earned and revocable.
Evidence or abstain
Answers grounded in your own documents and data, with citations, and the system says so when it doesn't know.
Optimisation solvers
Allocation, scheduling and routing done by solvers, so every number can be reproduced.
Operations graph & decision memory
One connected record of how your operation behaves: what was decided, on what evidence, and what happened.
Read-first connectors
SAP, Oracle, Tally, WMS, MES and historians, connected read-only first; writes go through approval.
Security & audit
Every action logged and reversible, data kept per tenant, and deletion on exit when a pilot ends.
What is shared, and what stays specialist.
The split is deliberate. Shared foundations make every brand stronger without making every brand look like the same product.
Held by the group
- The Owllys standard every brand ships to
- Demo Forge and digital-twin tooling
- Governance: approval gates, audit trail, kill switches
- Evaluation and release gates
- Read-first connector patterns
- Security and data handling
- The brand system and one front door
Owned by each brand
- Its industry's data model and rule set
- Its products, packs and roadmap
- Its subject-matter experts
- Its customer relationships and use cases
- Its website, demos and product detail
How we build, secure and govern it.
The engineering behind every brand, in detail: consulting, security, AI governance, the technology stack and the AI itself, plus the glossary, the case studies and the Insights library.
- 01Tech & AI ConsultingA seven-layer IT and process audit, a growth-aligned roadmap, and a partner accountable for the result.
- 02Application & Data SecurityProduction-grade security, not a checkbox. Every Owllys product is built secure by default — the same hardening that earned our live SaaS portfolio a 9.5/10 security rating.
- 03AI Security & GovernanceAI recommends. Business governs. Deploying agentic AI without losing control is a governance problem, and we solve it first.
- 04Technology StackBattle-tested engineering, shipped. A pragmatic, modern stack run by senior engineers — the same one behind our live, production SaaS products.
- 05AI & Machine LearningAgentic AI on a shared Operations Graph. The intelligent core under every division: Agentic AI grounded in RAG over your own data, reasoning on a shared Operations Graph, and optimisation where the maths matters.
- 06AI GlossaryPlain-language definitions of the AI terms used across the group.
- 07Case StudiesTwo production SaaS products, built end to end and live in the UK and the US.
- 08Insights library43 practical guides with infographics and process workflows, in seven topics.
See a decision rehearsed before it is made.
Bring one decision. We'll show you how it plays out in a simulated world, with zero customer data, before anything touches your systems.