AKE customised enterprise deployment

Our core service. We deploy the Agentic Knowledge Engine inside your environment and shape it around your knowledge, your systems and your approval points, then help your team run it.

Start a conversation Our approach

What AKE is

AKE, the Agentic Knowledge Engine, is the shared technology platform developed by Magellan Global with MTRI for AI applications in knowledge-intensive work. It follows one reference model, adapted to each deployment.

Authorised knowledge sources

Documents, systems of record, tickets, policies — each with an owner and permissions.

Retrieval and context

The right passages, records and terminology for the task, with provenance kept.

Agent-assisted tasks

Drafting, classification, checking, assembling cases, proposing next actions.

Human review

Approval points where a person sees the proposal, the evidence and the consequence.

Workflow integration

Approved outcomes flow into your ERP, CRM, DMS or service desk — not a separate tool.

AwePlus implements the model; Magellan Global stewards the platform and its IP; MTRI supports research and evaluation. Read about AKE on the Magellan Global site.

What “customised” means

A deployment is not a licence key. Six things are shaped to your enterprise before anything goes live.

Your knowledge

Sources, taxonomies and terminology from your organisation, mapped with owners and currency rules.

Your permissions

Agents inherit the access limits of the people they assist. No source is read that the user could not read.

Your systems

Integration with the ERP, CRM, document, identity and ticketing systems you already run.

Your approval points

Drafts and approved actions are separate objects. You decide which steps a person must confirm.

Your evaluation set

Tests built from your real cases, including difficult and incomplete ones, re-run on every change.

Your languages and jurisdictions

Traditional Chinese, Malay, Japanese and English workflows, with rules kept per country where they differ.

Who delivers it

Every deployment is led by a forward deployed engineer who works inside your team. The rest of the team stands behind them.

Forward deployed engineer (FDE)

Embedded with your process owners for the length of the deployment. Configures AKE, builds the first integrations, runs the evaluation set and trains your users. One named person, accountable for the outcome.

Solution and integration engineers

Taipei-based engineers who design the knowledge model, build connectors to your ERP, CRM, document and identity systems, and keep the platform current.

Delivery manager

Owns scope, acceptance criteria, decision points and change control, and is your escalation path. Works from the office nearest to you.

Deploy where your data lives

AKE runs where your governance requires. Availability of specific components in each environment is confirmed during Diagnose.

Your cloud tenant

Deployed inside your own AWS, Azure or Google Cloud subscription, under your identity and logging.

Private cloud or on-premises

For data that must not leave your estate. Our Taipei engineering team supports both.

Regulated environments

Deployment patterns for financial, healthcare and public-sector controls, agreed with your compliance owner.

How a deployment runs

Four stages with a decision point between each. Durations are indicative for a bounded first workflow.

Diagnose

2–4 weeks

Interview process owners and map the workflow.
Review data, systems, permissions and constraints.
Rank use cases and define a bounded pilot with acceptance criteria.

Design

3–6 weeks

Map knowledge sources and permissions.
Specify agent tasks, approval points and integration boundaries.
Build the evaluation set from your real cases.

Deploy

6–12 weeks

Configure AKE in your environment.
Build integrations and interfaces; test against the evaluation set.
Train users and go live on the agreed scope.

Operate

Ongoing

Named service owner, monitoring and support.
Change control when models, sources or rules change.
Improvement backlog reviewed with you.

What you receive

Knowledge-source and permission map
Agent workflow specification with approval points
Architecture and integration specification
Configured AKE environment with versioned releases
Evaluation set, results and acceptance records
Operating playbook, user guidance and training
Support and change-control plan

Engagement models

Pilot

A fixed-scope deployment of one workflow with agreed acceptance criteria. The fastest way to find out what AKE does in your environment.

Deployment programme

Several workflows deployed in sequence, each with its own decision point, under one delivery manager and one evaluation discipline.

Operations and improvement

A retainer that keeps deployed workflows dependable: monitoring, change control, support hours and a reviewed improvement backlog.

Questions

Is AKE a product we buy, or a service?

Both, and the service is what makes the product useful. AKE is the platform, stewarded by Magellan Global. AwePlus is the engineering and delivery team that configures it for your workflows, integrates it with your systems and supports it in operation.

Which AI models does it use?

AKE is model-agnostic. We configure it with the model providers your policy permits, including options hosted in your own region or tenant, and the evaluation set tells us how each performs on your cases.

Does our data train anyone else’s model?

No. Each engagement defines what information is used, who may access it and how it is retained. Client information is not pooled across clients or across Magellan Global brands.

Is every component of the reference model available today?

The reference model is adapted to each use case. The Diagnose stage confirms which components are available in your environment and what engineering is required, so the plan reflects what will actually be built.

How long before something is in use?

A bounded pilot typically goes live within three to five months of the first interview, with Diagnose and Design taking the first six to ten weeks. Larger programmes are sequenced so that something is in use early.

Can our own engineers run it after handover?

Yes, and that is the goal. Handover includes documentation, training and, where you want it, AI DX capability building for a nominated owner inside your enterprise.

Start with one workflow.

Tell us the process, the people who run it and what you would like to change. We will suggest the right first conversation, with no obligation.

Start a conversation hello@aweplus.com