AI Automation

EDG Grant for AI Automation in Singapore

July 2026·11 min·VYR Team
An upward view of Singapore's downtown office towers with glass facades catching golden-hour sunlight, framed by overhanging tree branches.

EDG Grant for AI Automation in Singapore: Eligibility, Process, and Cost Recovery

Executive Summary

The Enterprise Development Grant (EDG) is a Singapore government co-funding scheme administered by Enterprise Singapore (ESG) for qualifying Singapore-registered enterprises undertaking capability-building and productivity-improvement projects. Custom AI workflow automation falls within scope under the Core Capabilities pillar, Innovation and Productivity category, and can attract co-funding of up to 50% of qualifying costs for eligible small and medium-sized enterprises (SMEs). This article covers eligibility, qualifying cost structure, the Business Grants Portal (BGP) application sequence, common causes of delay or rejection, and how EDG interacts with the Productivity Solutions Grant (PSG) and the Enterprise Innovation Scheme (EIS) 400% tax deduction, together with the announced consolidation of EDG, PSG, and Market Readiness Assistance (MRA) into a single EDGE scheme expected in the second half of 2026.


1. What the EDG Covers for AI Automation Projects

The EDG spans three pillars: Core Capabilities, Innovation and Productivity, and Market Access. AI automation projects generally sit under Core Capabilities, within Innovation and Productivity, which supports businesses adopting new processes, technologies, and operational capabilities. The grant is not restricted to technology companies. It is available to any qualifying Singapore enterprise investing in a productivity-improving project, regardless of industry.

Project types that have qualified include AI workflow agents handling multi-step operational processes, document-processing automation that reads, extracts, routes, and files records, customer-support automation covering triage and response drafting, lead-qualification and routing workflows, internal knowledge systems giving staff governed access to operating information, and back-office automation for reporting and reconciliation. The connecting requirement is a demonstrable productivity outcome — a reduction in manual steps, error rates, or processing time, evidenced against a measurable baseline. Exploratory technology trials without a deployment plan generally do not meet the bar.


2. Eligibility Criteria

Enterprise Singapore assesses eligibility at the level of the company and the project.

Company eligibility. The applicant must be a business entity registered and operating in Singapore (including sole proprietorships and partnerships) with at least 30% local shareholding, held by Singapore citizens or permanent residents. The entity must meet the SME definition: 200 employees or fewer, or annual group revenue of S$100 million or less. The applicant must also be the primary beneficiary of the project, meaning the automation must serve the applicant's own operations rather than being built primarily for resale or for a third party.

Project eligibility. The project must demonstrate a clear productivity, efficiency, or capability-building rationale, supported by a defined scope and a measurable outcome. Enterprise Singapore's assessors evaluate the operational case rather than the underlying technology, so a proposal built around a vague intention to "use AI," rather than a specific workflow, system, and outcome, is likely to be returned for clarification.


3. Qualifying Cost Categories and Co-Funding Levels

For SMEs under Core Capabilities, the EDG typically covers up to 50% of qualifying project costs, with the balance funded by the applicant. Qualifying categories include third-party vendor fees for development and implementation, project management costs directly attributable to the project, and staff training costs forming part of implementation.

No fixed cap per project is published, though a typical qualifying cost range for a single-workflow AI automation project falls between S$15,000 and S$40,000, depending on scope. A commonly cited practical ceiling for straightforward SME projects sits around S$30,000 in qualifying costs, corresponding to up to S$15,000 in grant support at the 50% rate. Larger, multi-workflow projects can attract proportionally higher amounts where the proposal justifies the additional scope.

One structural point matters throughout: the EDG operates as a reimbursement, not an upfront payment. Project costs are paid to the vendor first and claimed afterward against invoices and receipts, so the applicant needs the capital available before reimbursement.

AI Automation ROI: Before and After for Singapore SMEsComparison diagram showing the impact of AI workflow automation on Singapore SME operations — from manual processing to automated, governed workflowsBEFORE — MANUAL PROCESSWITH AI AUTOMATION3–5 hours/dayon repetitive tasks👤Staff time onlow-value workHuman errors& inconsistenciesAutomated in minutesnot hours🎯Staff focused onhigh-value workConsistent, governedoutputs every timeAverage payback period:3–6 monthsEDG grant:up to 50% subsidised

4. The Application Sequence (Business Grants Portal)

The EDG application follows a fixed sequence, and the single most consequential rule is that project costs incurred before written approval is granted are not eligible for reimbursement.

  1. Scope definition. The workflow, systems involved, measurable outcome, and productivity baseline should be defined before drafting begins. Vague scoping is the most common cause of clarification requests, and each round adds weeks to the timeline.

  2. Vendor quotation. A formal quotation from the selected vendor is required before submission, since it is what Enterprise Singapore uses to assess project cost and calculate the grant quantum. It should itemise implementation fees, project management, and training costs where applicable.

  3. Submission through the Business Grants Portal. Applications are submitted at businessgrants.gov.sg, which requires Corppass access — worth confirming early, since setup can itself take about a week. The applicant selects Enterprise Development Grant, then Core Capabilities as the pillar and Innovation and Productivity as the category, before completing the project details, company information, and document upload sections.

  4. Document attachment. A complete submission includes the two most recent years of financial statements (or management accounts for newer entities), a project proposal, the vendor quotation, and a description of the staff roles involved. Incomplete documentation is a common cause of delay.

  5. Assessment period. The standard processing window is six to eight weeks from submission, extending during high-volume periods or where clarification is requested. Work should not commence until a written letter of offer has been issued.

  6. Project commencement. Once the letter of offer is received, the project may begin, per the approved grant quantum, timeline, and any conditions attached.

  7. Reimbursement claim. After completion within the approved timeline, a claim is submitted through the portal, supported by vendor invoices, payment receipts, and evidence the project was delivered as scoped.

EDG Grant Application Process for AI Automation in SingaporeStep-by-step diagram showing the 6-stage EDG grant application process for Singapore SMEs seeking AI automation fundingEDG GRANT APPLICATION TIMELINE1Book Call30 min scopingWeek 12Project ProposalVYR preparesWeek 1–23BGP Applicationbusinessgrants.gov.sgWeek 24ESG Review6–8 weeks approval6–8 wks5Project KickoffAutomation build startsAfter approval6Claim ReimbursementUp to 50% backPost-deliveryVYR handles steps 1–2 at no extra cost

5. Structuring the Project Proposal

The proposal is typically the section on which applications succeed or fail. Assessors are evaluating whether the project has a clear productivity rationale, a defined scope, and a measurable outcome — not whether the underlying technology is sophisticated. A strong proposal answers three questions: what the operational problem is, what is being built to address it, and how the outcome will be measured.

A useful framing device is a capacity calculation: manual hours currently spent on the task, multiplied by staff involved and working weeks in a year, establishes the baseline capacity consumed by the problem. If two operations staff each spend six hours weekly on manual invoice processing, that represents roughly 600 hours of annual capacity — a figure the proposal can frame the automation as recovering. Outcomes should be expressed in measurable, baseline-referenced terms wherever possible — processing time, error rate, or response time reduced against a stated starting point — since specific figures are more persuasive to an assessor than general claims, and the technical description should stay operationally framed rather than heavy on implementation detail the assessor does not need.


6. Common Causes of Delay or Rejection

Retroactive applications. Costs incurred before a written letter of offer is issued are ineligible, regardless of submission date. That is the most costly and avoidable failure mode, typically arising when a project begins informally with a vendor before the timeline is understood.

Vague project descriptions. Proposals describing an intention to "use AI to improve operations," without specifying workflow, systems, or baseline, are routinely returned for clarification. Naming the workflow, systems, and outcome distinguishes an approvable proposal from a generic one.

Incomplete financials. Enterprise Singapore verifies both eligibility and capacity to co-fund the project, so missing or outdated financial statements are a frequent cause of delay.

Confusing the EDG with the PSG. The PSG covers pre-approved, off-the-shelf software from a curated vendor list; the EDG covers bespoke, customised projects, including custom AI automation — addressed further below.

A rejection is not necessarily final. Most fall into one of three categories: an unclear productivity rationale, a failure to meet a basic eligibility threshold such as the 30% local shareholding rule, or an incomplete application closed rather than held open for clarification. The first can generally be resubmitted with a tightened proposal; eligibility rejections require a structural change; incompleteness requires only that missing documentation be assembled. An appeal may be lodged within 14 days of an outcome notice, addressed directly to the stated reason for rejection.


7. EDG Compared with the Productivity Solutions Grant (PSG)

EDGPSG
StructureReimbursement grantReimbursement grant
Typical coverageUp to 50% of qualifying costsUp to 50–70% of pre-approved solution cost
Vendor restrictionNone — any qualified vendorPre-approved vendor list only
Project typeCustom, bespoke automation and integrationPre-approved, productised software
Application effortHigher — proposal, quotation, financialsLower — proof of purchase against listed solution

The PSG is faster to apply for, drawing on a pre-approved vendor list, but custom AI automation, agentic workflow builds, and bespoke integrations generally do not appear on that list and cannot be claimed under it. The EDG is the applicable programme for AI automation built around an entity's specific workflows and systems rather than a packaged product.


8. The Enterprise Innovation Scheme (EIS) 400% Tax Deduction

Budget 2026 enhanced the Enterprise Innovation Scheme with a 400% tax deduction on qualifying AI expenditure, capped at S$50,000 of qualifying spend per year, for Years of Assessment 2027 and 2028. Every dollar of qualifying AI spend can generate four dollars of taxable-income deduction, up to the annual cap.

The EIS deduction is designed to stack with the EDG grant on the same qualifying spend. As an illustrative example only: on a S$30,000 AI automation project qualifying under both schemes, the EDG would reimburse up to S$15,000, while the EIS deduction on the same spend would generate a S$120,000 deduction, worth approximately S$20,400 in tax savings at the 17% corporate tax rate. Actual outcomes depend on the entity's specific tax position and should be confirmed with a tax adviser.

The EIS deduction is administered through the annual IRAS corporate tax filing, separately from the Business Grants Portal process used for EDG, and requires clean records of qualifying spend, vendor invoices, and productivity rationale. Grants received under EDG are treated as taxable income under standard IRAS treatment, though the EIS deduction on the same spend typically offsets a meaningful portion of that exposure.


9. The EDGE Consolidation (Budget 2026)

Budget 2026 announced that EDG, PSG, and Market Readiness Assistance (MRA) will be consolidated into a single programme, EDGE, scheduled for the second half of 2026, intended to reduce administrative complexity and create one front door for capability-building, productivity, and market-expansion support.

For entities planning AI automation projects in 2026: the three schemes merge into one, though underlying support categories are expected to continue in some form; applications will continue through the unchanged Business Grants Portal; current EDG criteria and co-funding ratios remain in force until EDGE takes over; and projects already approved under EDG are expected to continue under their original letter of offer. Enterprise Singapore's official channels should be monitored for the confirmed launch date, since implementation details remain subject to change.


10. Data Governance Considerations for Grant-Funded AI Projects

Grant-funded AI automation projects typically process operational and, in some cases, personal data, bringing the Personal Data Protection Act 2012 (PDPA) into scope regardless of project size. Three obligations are particularly relevant during scoping: Purpose Limitation, requiring personal data be used only for notified, reasonable purposes; data minimisation, limiting the personal data an automated system accesses to what the specific task requires; and the Protection Obligation, requiring reasonable security arrangements against unauthorised access, modification, or disclosure.

For more substantial deployments, Singapore's Cyber Security Agency (CSA) has published the Guidelines on Securing AI Systems, together with an Addendum on Securing Agentic AI Systems addressing the risk surface of systems taking action across multiple connected applications. Referencing these guidelines during scoping tends to produce cleaner EDG proposal documentation, since assessors look favourably on projects treating data handling and system risk as part of the operational design rather than a compliance add-on.


Frequently Asked Questions

How long does the EDG application take to process? The standard window is six to eight weeks from submission. Complete, clearly scoped applications tend to clear closer to the lower end; those requiring clarification can extend well beyond it, since each round resets the queue position.

Does the EDG cover AI chatbots? Coverage depends on function rather than label. A standalone FAQ chatbot is unlikely to qualify alone, since it does not demonstrate a clear productivity outcome. A system that triages queries, routes them, drafts contextual responses, and updates operational records is evaluated on the operational outcome it produces.

What is EDGE, and when does it replace EDG? EDGE is the announced consolidated successor to EDG, PSG, and MRA, expected in the second half of 2026. Current EDG rules remain in force until that transition, and projects approved beforehand continue under their original terms.

Can the EDG grant be combined with the EIS tax deduction, and is EDG income taxable? Yes to both. EDG and EIS are structured for use together on the same qualifying spend: EDG as a reimbursement grant, EIS as a corporate tax deduction, claimed through separate processes. EDG amounts are treated as taxable income under standard IRAS treatment, though the EIS deduction on the same spend typically offsets a meaningful portion of that liability; confirmation from a tax adviser is recommended for the applicant's specific position.


Related Reading

Entities evaluating AI automation investment more broadly may find it useful to review the cost breakdown of AI workflow automation projects in Singapore, a practical guide to AI agent implementation, and a list of common SME workflows suited to automation. Orchestration architecture is covered under agentic workflow orchestration and operations automation, with the full service range on the services page; delivery structure and indicative pricing appear on how it works and pricing.


Schedule a Technical Scoping Call

Entities assessing whether a specific workflow is a suitable candidate for custom AI automation, and separately whether the associated project costs may qualify for EDG co-funding, can schedule a technical scoping call to review the workflow, the relevant systems, and the operational outcome being targeted.


Singapore enterprise entities embarking on custom development projects may evaluate eligibility for co-funding via the Enterprise Development Grant (EDG) administered by Enterprise Singapore.