Start with the disclaimer, because it matters

Grant schemes change. Support levels change, qualifying cost definitions change, approved-vendor lists change, and some enhancements are time-limited. Nothing in this article is an eligibility determination, a guarantee of funding, or a substitute for the official criteria published by the administering agency.

What this article can do is orient you: which schemes are commonly relevant to an AI or automation project, what the eligibility conditions generally look like, and — most usefully — the assumptions that cause applications to fail.

Verify current details at the source before you plan around any number.

The two schemes that come up most

The Productivity Solutions Grant (PSG). Designed for the adoption of pre-approved IT solutions and equipment. Support has commonly been set at up to 50% of qualifying costs, subject to a cap per applicant. The defining characteristic is the pre-approval: the solution and the vendor must appear on the official list. If they do not, the scheme does not apply, however good the solution is.

The Enterprise Development Grant (EDG). Designed for larger projects that build capability, typically across core capabilities, innovation and productivity, or market access. Support for eligible SMEs has commonly been up to 50% of qualifying project costs, with higher levels for certain project types and periods. EDG projects are assessed individually — there is no pre-approved list — which makes it the more usual route for a bespoke automation build.

Which one fits depends on the shape of your project. Buying a listed solution points toward PSG. Building something specific to your operation points toward EDG.

The scheme funds a project with a defined outcome. It does not fund "we would like to use AI".

The eligibility conditions people underestimate

Local shareholding. Schemes of this type generally require a minimum level of local equity — commonly at least 30% held directly or indirectly by Singapore citizens or permanent residents. This single condition disqualifies a large number of otherwise viable applicants, particularly foreign-owned subsidiaries operating in Singapore. Check your cap table before you check anything else.

Registration and operation in Singapore. The entity must be registered and operating here, and the solution generally must be deployed for use in Singapore.

Size criteria. SME definitions typically involve group-level thresholds on annual turnover or employee headcount. Group-level is the trap: a small Singapore entity within a large international group may not qualify as an SME.

Financial readiness. The applicant is generally expected to be financially able to start and complete the project. Agencies do assess this.

Approval before commitment. This is the most expensive assumption of all. These schemes generally do not fund work you have already committed to or paid for. Signing the contract, paying the deposit, or starting the build before approval typically disqualifies the cost. If you take nothing else from this article: apply before you commit.

What tends to be fundable, and what does not

Grant assessors are looking at a project with a defined scope, a defined outcome and defensible costs. Broadly:

Usually within scope: third-party implementation and consultancy costs for the project, qualifying software costs for a defined period, and equipment or systems directly required by the project.

Usually outside scope: your own staff's time, general operating expenses, costs incurred before approval, and open-ended retainers with no defined deliverable.

The practical consequence is that a well-defined project — "automate expense capture and coding for entity X, delivered in twelve weeks, with these deliverables and this measured outcome" — presents far better than an ambition.

What a credible application looks like

Whatever the scheme, the underlying question an assessor is answering is the same: is this a real project that will produce a real productivity improvement, and can this business execute it?

That means your application should contain:

A specific process. Named, scoped, bounded.

A baseline. What it costs or takes today, in hours, errors, or lost revenue.

A target outcome. What improves, by how much, measured how.

A defined scope of work. Deliverables, milestones, timeline.

Credible costs. Quotations from the implementing vendor, itemised.

Evidence you can execute. Who owns it internally, and how it will be sustained after go-live.

Notice how closely that overlaps with the discipline of running a good automation project. That is not a coincidence. Businesses that scope projects properly write better applications, because the application is mostly a description of a properly scoped project.

Five assumptions that cost businesses money

"We'll apply after we start." Costs committed before approval are generally not claimable. This is the most common and most expensive error.

"Any AI vendor's work qualifies." For PSG, the solution and vendor must be pre-approved. For EDG, the project must meet the assessment criteria. Neither is automatic.

"The grant covers everything." Co-funding means you fund the rest. Plan the cash flow for the full amount, because claims are generally reimbursed after the fact.

"We're Singapore-based, so we qualify." Local shareholding, group-level size thresholds and financial-readiness criteria all apply independently of where your office is.

"It'll be approved in a fortnight." Assessment takes time — commonly measured in weeks, sometimes longer for larger projects. Build the timeline into your plan rather than discovering it.

A sensible sequence

Scope the project properly first. One process, baseline, target, deliverables, timeline. You need this regardless of funding.

Check the disqualifiers early. Local shareholding, group size, entity status. Ten minutes, and it prevents wasted effort.

Identify the likely scheme. Pre-approved solution, or bespoke project? That determines the route.

Get itemised quotations from the implementing partner, matched to the scope of work.

Apply through the official channel — the government business grants portal, using corporate digital identity credentials.

Wait for approval before committing. Then execute, document as you go, and claim against the agreed milestones.

Where to verify

Go to the source. The administering agency's own pages and the government business grants portal carry the current support levels, eligibility criteria, qualifying cost definitions and approved-solution lists. Intermediaries — including us — can help you interpret and prepare, but the criteria that count are the published ones.

If you want a second read on whether your project is likely to present well, that is a reasonable thing to ask an implementation partner for. What no honest partner will do is promise you a grant.

The summary

For a Singapore-registered, majority-locally-held SME, co-funding of a properly scoped digitalisation project is a realistic possibility and worth investigating before you fund the work from cash flow. The disqualifiers are usually structural — shareholding, group size, or having already committed the spend — and all three can be checked in an afternoon.

Scope the project first. Check the disqualifiers second. Apply before you commit. And confirm every figure at the source, because the details in this article will eventually be out of date, and the ones on the agency's website will not.

Start with a free AI Readiness Assessment

Normally valued at SGD 1,500, currently free. You receive an opportunity report, a prioritised roadmap and honest ROI estimates for your own processes — with no obligation. Book yours at aigentify.tech/assessment.

AIgentify — Singapore-based AI implementation specialists. We design, build and support AI agents, workflow automations and intelligent business applications with measurable ROI. Live in weeks, not months. Your data stays yours.

This article is general information, not advice. Grant schemes, platform rules and regulatory requirements change; confirm current details with the relevant authority or provider before relying on them. Any figures shown are illustrative unless a source is stated.