Decision guide
Build the business behind the AI.
A decision library for CyberG7: how the estate, Jarvis and WhatsApp become useful SME systems, repeatable delivery and defensible funding evidence. Refined edition · 2 October 2026; grant review 1 October; receptionist documentation update 2 October.
The operating loop
- 01Attract the right ownerA specific bottleneck and an honest demonstration
- 02Move a real transactionAccepted handover, persistent records, approvals
- 03Prove the benefitReal costs, useful capacity, contribution and adoption
- 04Retain the evidenceAwarded scope, delivery, invoices and full payment
Prove the operating chain, then expand the scope.
The recommended sequence
Start with one dependable answers-and-request product for appointment/service businesses. In parallel, qualify one operations-heavy ERP opportunity. Use approved partners when funding is essential now, and build your own repeatable client evidence over time.
| Iteration | Change in thinking | Business consequence |
|---|---|---|
| First pass | Funding map and illustrative proposals | Useful orientation; some budgets fail recurring economics. |
| This refinement | Transaction progression, operator acceptance and benefit realisation | A fast answer now has a measurable path to a business result. |
| Next client iteration | Replace hypotheses with a baseline, real systems and quotations | Choose a bounded package, partner or transformation with an explicit go/no-go. |
Choose a commercial path
| Path | Best situation | What must be true |
|---|---|---|
| Receptionist reference cohort | Repeat questions and enough recurring workload | A standard package can deliver useful requests profitably. |
| Approved-package partner | Client requires a verified grant route immediately | An actual approved principal and permitted delivery arrangement exist. |
| Operations-first ERP | Quote, work and invoice data are repeatedly reconstructed | A genuine three-function ERP is necessary; assess approved options first. |
| Secure casework transformation | Multiple systems block caseworker throughput | Integration and realised benefit justify the scope; otherwise start smaller. |
Explore the explainers
Every named estate group has a role; client projects select the necessary pieces.
The missing middleSuccess from enquiry to renewalStage-by-stage denominators, evidence, owners and stop rules.
Four commercial pathsStrategy variationsCompare speed, control, funding dependency and repeatability.
PlaybookAttract and qualify ownersHooks, demonstrations, destinations, funnel economics and proof.
PlaybookDeliver and evidence the claimSubmission boundaries, data readiness, UAT, evidence and settlement.
Proposal / Lynn BeautyAn enquiry becomes a usable requestLean, booking and multi-outlet variations with a lower-cost planning model.
Proposal / Kim PongA job becomes a billable recordThree-function ERP versus an existing-platform enhancement.
Proposal / GetSGPRMake the file ready for reviewTransformation conditions, whole cost and a defer-or-expand decision.
Alternative / GetSGPRA smaller AI adoption projectBounded scope, qualifying cost allocation and a tax illustration.
Application explainerWhat the assessor can verifyA claims story built from records, rather than three agent names.
The funding rules that shape the design
| Route | Design requirement | Planning ceiling / condition |
|---|---|---|
| Customer-engagement chatbot | Exact pre-approved package | SME: up to 50%; cost ceiling S$15,000, grant ceiling S$7,500. |
| Marketing-content solution | Separate exact pre-approved package | SME: up to 50%; cost ceiling S$5,000, grant ceiling S$2,500. |
| Integrated ERP | Single ERP with at least three core functions | SME: up to 50%; cost ceiling S$60,000, grant ceiling S$30,000. Other vendor only when suitable approved options lack required functions. |
| Large-scale transformation | Substantive operating change across systems | SME support up to 50%; scope and costs assessed. Isolated AI tasks do not establish eligibility. |
| EIS AI adoption | Qualifying AI cost allocation | 400% first S$50,000 for YA 2027/2028; no AI cash conversion. Net subsidy basis; actual tax usability matters. |
Source: GenAI customer engagement ↗ · GenAI marketing content ↗ · Integrated ERP activity ↗ · Large-scale digital transformation ↗ · IRAS EIS ↗
Source: EDGE FAQ ↗ · BCA Built Environment PSG ↗
Read this as a decision model
The real business names and public services are sourced. Their actual bottlenecks, applicant eligibility, proposed budgets and benefits are unverified. Current readiness notes are dated source observations, not a fresh production audit of the whole estate. No new EDGE approval, partnership or CyberG7 pre-approved package is asserted.
Planning prices are revisions for economic discussion, not live prices, binding quotations or vendor licence lists. Grant approval, delivery acceptance, tax eligibility and commercial success are four separate decisions. This edition supersedes the earlier broad budgets where explicitly stated.
Files and earlier report
Download all 11 explainers for offline reading · Read the earlier full report
The refined explainers supersede earlier assumptions where stated. Grant sources were reviewed 1 October 2026; receptionist readiness was refreshed from the 2 October handoff.