Gaincue × Canvas Global
AI operations,
inside CGL.
A native AI layer that reads CGL's context, calls approved tools, and carries operational work through to a result — not another chatbot bolted on.
Quotation · September 2026 · RM40,000 fixed · valid 30 days · add-on to the RM80,000 CGL application
02 · Result
What the AI layer will do.
Read the context
Starts from the job, shipment, or customer in front of the user, not a blank chat box.
Retrieve and search
Pulls operational records, customer context, rates, services, and documents through approved tools.
Summarise and flag
States where a job currently stands and names the information that is missing.
Prepare the work
Populates structured inputs, drafts entries, and assembles what a task needs.
Execute approved steps
Runs routine operations through guarded, permission-aware tools.
Chain and report
Coordinates several tools into one task and reports the completed outcome.
The objective is not to add an AI chatbot. It is to let users ask for an outcome, rather than navigate every screen required to produce it.
03 · Architecture
The AI decides. CGL authorises. Tools execute.
Layering
- User and CGL data
- AI operations layer
- Approved CGL tools
- Internal application functions · Carrier browser automation · Other services
The AI layer decides what needs to be done; specialized execution modules perform the actual operation.
Example tools
- Find shipment
- Retrieve shipment details
- Retrieve customer information
- Search operational records
- Retrieve rates or services
- Create or update approved records
- Prepare structured workflow data
- Retrieve documents or references
- Trigger permitted workflows
The model is not the security boundary. Every tool stays subject to CGL authorization and server-side validation.
04 · How it runs
Lightweight by design. Tool-first, not prompt-only.
Operating model
- • Reuses CGL's existing authentication and permissions.
- • Tools are explicit; the AI cannot reach past them.
- • Model calls run through a compatible gateway, so the provider can change.
- • Model chosen per workload: small models for classification and routing.
- • High-impact actions can require explicit human approval.
- • Structured data instead of uncontrolled prompt-only workflows.
Architecture principles
- • Minimal framework dependencies; no large standalone AI application framework.
- • Server-side authorization on every tool call.
- • Context and token sizes managed to contain cost.
- • Usage and tool execution logged for traceability.
- • New workflows added incrementally, without redesigning the core.
- • Sensitive-data-aware logging and safe error handling.
05 · Scope
What the RM40,000 covers.
| Component | Description | Amount |
|---|---|---|
| 1 · AI operations architecture & orchestration layer | Core AI runtime, model communication, tool execution lifecycle, structured responses, conversation orchestration. | RM7,000 |
| 2 · Native CGL AI user experience | Integrated AI interface, streaming responses, conversation handling, contextual state, action and result presentation. | RM4,000 |
| 3 · CGL tool & action framework | Secure framework allowing AI to retrieve data and invoke approved CGL business functions. | RM7,000 |
| 4 · Initial AI operational workflows | High-value workflows aimed at reducing repetitive manual work across existing CGL operations. | RM7,000 |
| 5 · Context, retrieval & business data integration | Application data retrieval, entity context, workflow state, structured business-context injection. | RM4,000 |
| 6 · Authentication, authorization & AI security controls | Permission-aware tools, server-side authorization, validation, execution boundaries, protection against unauthorized AI actions. | RM4,000 |
| 7 · Model gateway, usage & cost controls | Provider abstraction, compatible routing layer, model selection, token and context controls, usage visibility. | RM2,500 |
| 8 · Audit logging & observability | AI request visibility, tool execution logs, errors, operational diagnostics, traceability. | RM2,000 |
| 9 · Testing, UAT & technical handover | Functional testing, security-sensitive workflow validation, UAT support, documentation, handover. | RM2,500 |
| Total | RM40,000 | |
Additional scope beyond the original RM80,000 CGL application. Fees in MYR, exclusive of tax.
06 · Investment
RM40,000 fixed. An add-on to CGL.
Fixed price, ex-tax
RM40,000
billed 40 / 40 / 20
The AI operations layer, the tool framework, and the agreed initial operational workflows.
Payment milestones
- CommencementRM16,000
- Core platform & stagingRM16,000
- UAT & handoverRM8,000
In the fixed price
Nine components
see scope, slide 05
Built into your CGL application and repository, on the stack you already own.
Covers
- ✓AI runtime and orchestration
- ✓Tool and action framework
- ✓Initial operational workflows
- ✓Security controls and audit logging
- ✓Testing, UAT and handover
Third-party costs (you pay)
Pass-through
excluded from the fee
Billed directly to you, or separately approved.
May include
- ·AI inference / API usage
- ·Model gateway services
- ·Cloud infrastructure
- ·External vector or search services
- ·Observability services
A 30-day defect warranty applies after production acceptance. Carrier browser automation is separately quoted.
07 · Timeline
6 to 8 weeks. Working platform early.
- Wk 1–2 · FoundationAI runtime, model gateway, provider abstraction, conversation orchestration, structured responses.
- Wk 3–4 · ToolsTool and action framework with authorization, plus the first prioritized workflows demonstrating into staging.
- Wk 5–6 · ContextEntity context, retrieval, business-context injection, and the remaining agreed workflows.
- Wk 7–8 · HandoverSecurity controls, audit logging, cost controls, testing, UAT support, documentation and handover.
Subject to final priority-workflow selection, existing API/service readiness, staging data availability, and UAT response time.
08 · Next steps
To start, confirm two things.
1 · The priority workflows
Chosen for manual effort, frequency, repetitive navigation and data entry, operational delay, and error risk.
2 · Gateway and staging access
The model provider or gateway you prefer, and access to staging data and services.
Then we build the platform foundation, and the first workflows on top of it.
Find us at +60174421238 · info@gaincue.com
gaincue.com · 2026