Custom ERP Software for Singapore SMEs

Off-the-shelf ERP doesn't fit

Generic ERP software forces your team to change how they work. You end up paying for modules you don't use and still relying on spreadsheets on the side.

Data lives in 5 different tools

Sales in one system, inventory in another, finance in a third. No single source of truth means slow reporting and constant reconciliation.

Approvals happen in WhatsApp

Quotes, POs, and expense approvals get lost in chats. There's no audit trail and no clear visibility on what's stuck where.

AI feels out of reach

Everyone talks about AI agents and automation — but your processes aren't structured enough yet for AI to act on them safely.

Implementation horror stories

Traditional ERP rollouts take 12–18 months and cost six figures. You need something shippable in weeks, not a multi-year overhaul.

No vendor truly understands SG

GST, IRAS, MOM payroll, multi-currency for ASEAN — most generic ERP vendors get the local context wrong.

One record instead of five copies

A customer, a job or a stock item exists once. Sales, operations and finance read the same record, so nobody is working from a version that was accurate last Tuesday.

• A customer's address is changed once, not in three systems
• Stock levels reflect what was actually picked this morning
• Finance and operations argue about priorities, not about numbers

The handoffs become the system

Most SME pain is not in the work, it is in the gaps between people: quote to job, job to delivery, delivery to invoice. An ERP is mostly a way of making those handoffs explicit and recorded.

• Quote accepted automatically becomes a job with the same line items
• Job completion triggers the invoice rather than a reminder to raise one
• Nothing waits in someone's inbox to be forwarded

It is not one big replacement

The failed version of this project is replacing everything at once. The version that works starts with the single workflow costing you the most time, ships it, and then extends.

• Start with the workflow costing the most hours per week
• Ship it, let people use it, then extend from evidence
• Each module funds the argument for the next one

Reporting stops being a monthly exercise

When the operational records are already structured, management reporting is a view rather than a reconstruction job someone does in the first week of every month.

• Revenue, margin and ageing computed from live records
• No month-end week spent rebuilding the same spreadsheet
• Management sees the same numbers the team is working from

It survives staff turnover

A process encoded in software outlasts the person who invented it. A process encoded in one person's spreadsheet leaves when they do.

• The process is documented by being executed
• A new hire follows the system instead of shadowing someone
• Nobody's resignation takes a critical spreadsheet with it

The AI layer needs it first

Automations and AI agents can only act on structured, trustworthy data. Clean operations are the prerequisite, which is why we sequence the workflow work before the AI work.

• Agents can only act on records that are structured and current
• Automations fail loudly on messy data and quietly on stale data
• Clean operations first, then the copilots are worth switching on

A focused first module

From about S$3,500, live in roughly four to eight weeks, on a fixed quote issued after the workflow review. This solves one real workflow end to end rather than sketching twelve.

• Fixed quote issued after the workflow review, not before
• One workflow solved end to end, not twelve sketched
• Live on your own domain, owned by your team

A departmental system

Several linked workflows with a few roles and one or two integrations runs into the tens of thousands. This is where most Singapore SMEs land.

• Several linked workflows with a few roles
• One or two integrations to what you already run
• Where most Singapore SMEs actually land

A full multi-module operation

Sales, inventory, finance, approvals and reporting together is a six-figure conversation, and it should be phased rather than attempted in one go.

What drives the number

Scope, the number of distinct roles, how many integrations are required, how much historical data has to be migrated, and how much of your process is genuinely non-standard.

• Number of distinct user roles and permission levels
• How many external systems must be integrated
• How much historical data has to be migrated and cleaned
• How much of your process is genuinely non-standard

The costs people forget

Data migration, training, the parallel-run period where both systems are live, and the internal time your own team spends being interviewed and testing.

• Data migration and the cleaning that precedes it
• Training, and the productivity dip during it
• The parallel-run period where both systems are live
• Your own team's time being interviewed and testing

Licensed alternatives

Per-user monthly pricing looks small until you multiply it by headcount and sixty months, then add implementation and customisation on top. Compare five-year totals, not monthly figures.

• Multiply the monthly seat price by headcount and sixty months
• Add implementation, customisation and upgrade effort
• Then compare five-year totals, not monthly figures

Grants, honestly

PSG funds pre-approved packaged solutions, so a custom build does not qualify for it. EDG is the relevant path for custom work and supports up to 50% of qualifying costs for eligible local SMEs on a reimbursement basis, with the project not started before applying. Confirm current criteria with Enterprise Singapore.

• PSG covers pre-approved packaged solutions only
• EDG is the path for custom work, up to 50% of qualifying cost
• Reimbursement basis, and the project must not start before applying
• Confirm current criteria with Enterprise Singapore

What we will tell you to do instead

If your process is conventional and an off-the-shelf product fits, we will say so rather than sell you a build. That advice costs nothing and it is the fastest way to know which route you are on.

Scope set by a wish list

A requirements document assembled from every department produces a system nobody wanted. We scope from one workflow that is measurably costing you time.

• Scope from a workflow with a measurable cost
• Say no to features nobody can name a user for
• Revisit the list after the first module is live

No single owner on the client side

Projects stall when nobody internally can make a decision. One person with authority to say yes is worth more than a steering committee.

Big bang go-live

Switching everything on one Monday guarantees the first bad week becomes the story. Shipping one module and letting people use it makes the next module easier to sell internally.

• One module live beats twelve modules in testing
• The first good week is what sells the second module internally
• Rollback is possible when only one process moved

The spreadsheet that never dies

If the system does not do the one thing the QS or the storeman actually needs, they will keep their sheet, and their sheet becomes the real source of truth again.

• Interview the person who maintains it before designing
• If their edge case is unsupported, they will not switch
• Adoption is measured by that spreadsheet being deleted

Training treated as an afterthought

Adoption is the project. A system that is technically complete and unused has failed regardless of how well it was built.

Nobody owns the data migration

Historical data is always messier than anyone remembers. Deciding early what actually needs to come across, and what can stay in the old system as an archive, removes most of the pain.

Construction and contracting

BOQ-driven quoting, progress claims with retention, subcontractor certification and site-to-office capture.

• Claims roll forward from the last certified cycle
• Retention tracked as a ledger, not per project
• Variation orders dated and costed before the work

Dental and medical clinics

Appointments, practitioner rosters, procedure libraries and recall cycles, with patient data handled carefully.

• Chair and practitioner utilisation visible daily
• Recalls generated rather than remembered
• Procedure pricing consistent across clinics

Interior design and renovation

Quote by scope, track cost against budget per project, and manage a long supplier tail.

• Design fee and build cost separated cleanly
• Supplier lead times against the site programme
• Client sign-off recorded at each stage

Trading and distribution

Multi-location stock, tiered customer pricing and purchasing across currencies.

• Per-customer price tiers applied automatically
• Landed cost including freight and duty
• Reorder points that reflect real lead times

F&B and catering

Recipe costing, event orders, delivery scheduling and counter sales in one place.

• Food cost per dish computed from purchases
• Event orders separated from daily counter sales
• Delivery runs planned against capacity

Aesthetics, spa and wellness

Packages, session balances, therapist commissions and membership tracking.

• Prepaid packages drawn down accurately
• Therapist commission from the same sale record
• Membership renewals surfaced before lapse

Insurance agencies

Client books, policy renewals, commission reconciliation and compliance records.

• Renewal dates surfaced weeks ahead
• Commission reconciled against insurer statements
• Advisory records retained and retrievable

Specialty retail

POS tied to the same stock and customer records, with loyalty and multi-outlet reporting.

• Stock truth across outlets, not per till
• Loyalty tied to the customer, not the receipt
• Outlet performance compared on one screen

Phase one, the proving module

Budget S$3,500 to S$12,000 for one workflow solved end to end, live in 4 to 8 weeks on a fixed quote.

• S$3,500 entry, fixed after review
• 4 to 8 weeks to live
• One workflow, fully solved

Phase two, the department

Once phase one is in daily use, a departmental build covering 3 to 4 linked workflows typically runs S$25,000 to S$60,000.

• 3 to 4 linked workflows
• S$25,000 to S$60,000 typical
• Scoped from evidence, not a wish list

Phase three, the operation

Sales, stock, finance, approvals and reporting together is a six-figure programme and should run across 12 to 18 months, not one quarter.

• 12 to 18 months, phased
• Each phase funds the next
• No single big-bang go-live

Internal time to reserve

Budget 8 to 15 hours of your own team's time per module for interviews, testing and training. This is the line most plans omit.

• 8 to 15 hours per module
• Spread across 6 to 8 weeks
• Mostly the person who does the work

Running cost after go-live

Hosting for an SME system commonly runs S$50 to S$300 a month billed at cost to your own account, plus optional care quoted by scale.

• S$600 to S$3,600 a year hosting
• Billed to you at cost
• Care optional, quoted by scale

The comparison to run

Against a 20-seat licence at S$120 a month, five years is S$144,000 before implementation. Run both numbers over 5 years before deciding.

• 20 x S$120 x 12 = S$28,800 a year
• S$144,000 over 5 years, licence only
• Compare 5-year totals, not monthly

Grant treatment

EDG can support up to 50% of qualifying costs for eligible local SMEs on a custom build, on reimbursement, and the project must not start before applying. PSG does not cover custom work.

• Up to 50% of qualifying cost
• Reimbursement, not upfront
• Apply before starting
• Confirm with Enterprise Singapore

ERP

Enterprise resource planning: one system holding the records that several departments share, so a customer, job or stock item exists once rather than in five files.

• One record, many departments
• Replaces reconciliation with a shared source
• Scope is a choice, not all-or-nothing

Module

A functional area such as quoting, stock or payroll. A custom build means choosing three or four rather than buying twelve.

• Buy or build only what you use
• Modules share one data model
• Added when the last one is adopted

Master data

The reference records everything else points at: customers, suppliers, products, staff, cost codes.

• Cleaned once, maintained forever
• Duplicates here corrupt every report
• Owned by a named person

Transactional data

The events: quotes, orders, receipts, invoices, payments, timesheets. Volume grows daily; master data does not.

• Grows with trading volume
• Immutable once posted
• Corrections by reversal, not edit

Source of truth

The system everyone agrees to believe when two numbers disagree. Naming it is a management decision, not a technical one.

• Named explicitly, in writing
• Other systems reconcile to it
• Ambiguity here causes most disputes

Integration

A connection letting two systems exchange records automatically instead of by export and import.

• API or scheduled file exchange
• Direction matters: push, pull or both
• Failure handling is the hard part

API

An interface letting software talk to software. If a vendor has none, every future connection is manual.

• Ask before you buy
• REST is the common standard
• Rate limits and auth are the details

Audit trail

The record of who changed what and when. The difference between an explanation and an argument.

• Every change attributed
• Retained, not overwritten
• Required for most audits

Role-based access

Permissions granted by job function rather than by trust, so leavers are offboarded in one action.

• Permission by role, not person
• Sensitive screens hidden
• Revoked centrally

Cost code

The label that puts a cost against the job, department or project it belongs to.

• Applied at purchase, not month end
• Drives job profitability
• Kept short and stable

Committed cost

Money promised but not yet invoiced, usually at purchase order. The figure spreadsheets almost never capture.

• Visible at PO, not at invoice
• Early warning on overruns
• Distinct from actual cost

Reconciliation

Checking two records agree. Most of it exists because data was entered twice.

• Symptom of duplicate entry
• Removed by integration
• Never fully eliminated

Data migration

Moving existing records into the new system, usually the longest and least glamorous task.

• Cleaning is the larger half
• Archive rather than import everything
• Agree a cutoff date

Parallel run

Operating old and new together for one cycle so outputs can be compared before switching.

• One full cycle minimum
• Compare, do not assume
• Rollback stays available

Go-live

The point the new system becomes the record. Best done one module at a time.

• Never on a Friday
• One module, not twelve
• Support present for week one

User acceptance testing

The client testing that the system does what was agreed, using real scenarios rather than a script.

• Real cases, not happy paths
• Done by the people who will use it
• Sign-off before go-live

Change request

Work outside the agreed scope. Named and priced openly so it does not become an argument.

• Written, not verbal
• Priced before it is built
• Scope baseline retained

Technical debt

Shortcuts that make the next change slower. Sometimes worth taking deliberately, never worth taking silently.

• Recorded when taken
• Repaid before it compounds
• Invisible debt is the dangerous kind

Uptime

The share of time the system is available. Ask what it is measured against and who reports it.

• Measured, not promised
• Excludes planned maintenance
• Meaningless without monitoring

Backup and restore

Copies of your data, and the tested ability to bring them back. Only the second one matters.

• Automated and frequent
• Restore actually tested
• Recovery time measured

SSO

Single sign-on: one login across systems, so access is granted and revoked in one place.

• Central control of access
• Faster offboarding
• Fewer shared passwords

Two-factor authentication

A second proof of identity beyond the password. The cheapest security improvement available.

• Enabled for all admin users
• App-based beats SMS
• Recovery codes stored safely

PDPA

Singapore's Personal Data Protection Act, governing how personal data is collected, used and kept.

• Applies to staff and customer data
• Consent and purpose matter
• Breach obligations exist

GST

Goods and services tax, applied per line with the correct category. Errors here are a filing problem.

• Category per product, not per invoice
• Tax invoice timing matters
• Credit notes must reverse cleanly

Peppol

The international network standard behind structured e-invoicing, used by InvoiceNow in Singapore.

• Documents routed, not emailed
• Validated before acceptance
• Each participant has an identifier

CPF

Central Provident Fund: statutory contributions varying by age band and residency status.

• Rates change; confirm with CPF Board
• PR rates step up over two years
• Computed by the payroll run

SDL

Skills Development Levy, payable on top of CPF and computed in the same run.

• Applies alongside CPF
• Part of the payroll computation
• Not an optional extra

EDG

Enterprise Development Grant: supports up to 50% of qualifying costs for eligible local SMEs, on reimbursement.

• Custom builds may qualify
• Apply before the project starts
• Confirm with Enterprise Singapore

PSG

Productivity Solutions Grant, which funds pre-approved packaged solutions rather than custom work.

• Packaged solutions only
• Custom builds do not qualify
• Check the approved list first

SaaS

Software as a service: rented software, priced per user per month, hosted by the vendor.

• Recurring, not one-off
• Cost scales with headcount
• Vendor owns the roadmap

On-premise

Software running on a server you own and maintain, with the costs and control that implies.

• Hardware refresh every 4 to 5 years
• You own patching and backups
• Full control of the data

Total cost of ownership

Licence plus implementation plus migration plus training plus upgrades, over five years rather than one.

• Five years, not monthly
• Include internal time
• The only fair comparison

Vendor lock-in

The cost of leaving. Measured by whether you can export your data and who owns the code.

• Ask about export before buying
• Code ownership in writing
• Proprietary formats are the trap

Scalability

Whether the system copes as volume grows, and what it costs when it does.

• Load versus headcount pricing
• Test at 3x current volume
• Ask what breaks first

Workflow

The sequence of steps and approvals a piece of work passes through, and who owns each.

• Mapped before it is built
• Handoffs are where cost hides
• Encoded once, applied always

Approval threshold

The value above which something needs a second person to agree. Enforced by the system rather than by habit.

• Set by role and by value
• Auditable
• Prevents surprise discounts

Dashboard

A live view of the numbers management asks for, computed from records rather than assembled monthly.

• Live, not monthly
• Same data the team uses
• Few numbers, chosen deliberately

Reporting period

The window a report covers. Agreeing it prevents two people producing different truths.

• Defined and consistent
• Cutoff time stated
• Comparable period on period

Prerendering

Serving fully-formed HTML so crawlers and AI assistants read the content without running JavaScript.

• Content visible without JS
• Faster first paint
• Essential for search visibility

Agentic automation

Software that acts on records rather than only reporting on them, with human approval on anything consequential.

• Acts, not just reports
• Approval gates on money and messages
• Only as good as the data

Ask for the five-year total

Licence times sixty months plus implementation, migration and training. One number, comparable across vendors.

• 60 months, not one
• All one-off costs included
• Written, not verbal

Ask who owns the code

This single answer determines whether you can ever leave.

• Ownership in writing
• Repository access
• Another developer could take over

Ask about data export

A system you cannot export from is a system you cannot leave.

• Full export on demand
• Readable formats
• Scheduled exports you control

Ask what happens in week one

A vendor who starts with observation is scoping from your problem, not their product.

• Who they interview
• Whether they walk a real job
• What they hand back

Ask what they will refuse to build

A vendor who says yes to everything has not understood the problem.

• Where they recommend off-the-shelf
• What they push back on
• Out of scope on principle

Ask to see it running

Screenshots are not a system. Ask for a populated live environment.

• Live, not slides
• Populated, not empty states
• Someone using it

Ask about the second module

The first is a project; the second tests whether the architecture extends.

• Shared data model
• What clients added later
• How long it took

Ask for the support commitment in hours

Response time in hours, business days defined, escalation path named.

• Hours, not adjectives
• Business days defined
• Named escalation

Ask who maintains statutory logic

CPF, GST and e-invoicing rules change. Someone must own keeping them current.

• Named owner
• In the contract
• Update mechanism stated

Ask about the parallel run

One full cycle with both systems live is what turns confidence into evidence.

• One cycle minimum
• Outputs compared
• Rollback available

Ask what training is included

Two to five days by role, with a follow-up after real use begins.

• Days stated
• Per role
• Follow-up included

Ask how change requests are priced

Scope will change. How that is handled is worth agreeing before it happens.

• Priced openly
• Baseline retained
• Written approval

Ask for a reference you choose

Not the reference they offer. One you pick from their portfolio.

• You pick, not them
• Same size and sector
• Ask what went wrong

Ask what the system will not do

Every system has limits. A vendor who names them is a vendor who understands theirs.

• Named limits
• Workarounds explained
• No magic claims

Ask about hosting and who is billed

Hosting billed at cost to your account avoids a reseller margin you cannot see.

• Billed to you at cost
• Your account, your access
• No hidden margin

Do we have to replace everything at once

No, and you should not. One module, adopted, beats four half-built. Everything else keeps running.

• One workflow first
• Existing tools stay
• Integrate rather than replace

What if we outgrow it

A build extends at the cost of the work. That is the argument for owning the code rather than renting seats.

• Extended, not replaced
• Your priority order
• No tier upgrade pricing

How do you handle our data

It stays in a database you own, in a region you choose, exportable in full at any time.

• Your database, your region
• Full export on demand
• Access documented and auditable

What if the project stalls

Fixed scope, fixed quote and weekly working builds mean a stall is visible in week three, not month six.

• Weekly working build
• Visible early
• Fixed scope baseline

Can our own developer take it over

Yes. Standard stack, documented handover, and the repository is yours.

• Standard technologies
• Handover documentation
• Repository ownership

What ongoing cost should we budget

Hosting at cost plus optional care quoted by scale. No per-seat component at all.

• Hosting S$600 to S$3,600 a year
• Care optional
• No per-seat fee

How do you price changes after go-live

Openly, against the written scope baseline, with approval before anything is built.

• Priced before built
• Baseline retained
• Written approval

What happens if we stop working with you

You keep the code, the data and the hosting accounts. That is the point of ownership.

• Code and data yours
• Hosting on your accounts
• Another developer can continue

Do you work with our existing accountant

Yes, and we usually ask to. They know where the numbers have to land.

• Accountant involved early
• Statutory treatment confirmed
• Reports built to their format

How do we measure whether it worked

Against the hours you measured before we started. That is why we measure them in week one.

• Baseline measured first
• Same measure after
• Module two argued from it

Quote turnaround

Manual assembly commonly takes 30 to 60 minutes per quote. A configured quote is 5 to 10 minutes.

• 30 to 60 minutes manual
• 5 to 10 minutes configured
• 40 quotes a month = 30 hours saved

Invoice handling

Manual invoice handling runs about 2 minutes each. At 300 a month that is 10 hours, or roughly S$4,800 a year.

• 2 minutes per invoice
• 300 a month = 10 hours
• About S$4,800 a year

Payroll cycle

2 to 3 days a month for 30 staff is common, roughly 20 hours or S$7,200 a year at a loaded S$30 an hour.

• 20 hours a month
• 240 hours a year
• About S$7,200

Stock accuracy

Annual-count businesses commonly sit at 85% to 92%. Cycle counting with disciplined receiving reaches 98% or better.

• 85% to 92% typical annually
• 98%+ achievable
• Within one quarter

Stock shrinkage

A 2% variance on S$400,000 of stock is S$8,000 a year written off with no explanation attached.

• 2% on S$400,000 = S$8,000
• Discovered 12 months late
• No trail to investigate

Warehouse picking

Poor slotting adds roughly 30 seconds per pick. At 500 lines a day that is 4 hours daily, or about S$18,000 a year.

• 30 seconds per pick
• 500 lines = 4.2 hours a day
• About S$18,000 a year

Mispick rate

1% on 500 lines a day is 5 errors daily, each carrying redelivery and admin cost plus the customer trust cost.

• 1% = 5 errors a day
• Redelivery plus admin
• Trust cost on top

Implementation time

A focused first module is 4 to 8 weeks. A departmental rollout is 3 to 6 months, phased.

• 4 to 8 weeks first module
• 3 to 6 months departmental
• 12 to 18 months full programme

Adoption

Systems used daily within two weeks of training tend to stick. Beyond a month, adoption rarely recovers without intervention.

• Daily use within 2 weeks
• Beyond a month is a warning
• Measured, not assumed

Training load

2 to 5 days total, delivered as 30 to 60 minute sessions per role rather than one long workshop.

• 2 to 5 days total
• 30 to 60 minutes per role
• Follow-up after week one

Support response

4 business hours is a reasonable commitment for a non-critical issue; 1 hour for anything blocking trading.

• 4 hours non-critical
• 1 hour if trading is blocked
• Escalation path named

Five-year cost, 20 seats

Licence at S$120 a seat is S$144,000 over five years. A departmental build is S$25,000 to S$60,000 once.

• S$144,000 licence over 5 years
• S$25,000 to S$60,000 built
• Before implementation on licence

Against a per-seat licence

20 seats at S$120 a month is S$28,800 a year and S$144,000 over five, before implementation of S$20,000 to S$50,000.

• S$28,800 a year at 20 seats
• S$144,000 over 5 years
• Plus S$20,000 to S$50,000 implementation
• Plus S$1,440 a year per extra hire

Against staying manual

A workflow costing 20 hours a month is 240 hours and about S$7,200 a year, rising with volume and never repaid.

• 240 hours a year
• About S$7,200
• Grows with volume
• No asset at the end

Against hiring instead

A back-office hire is S$36,000 to S$54,000 a year before CPF, recruitment and cover.

• S$36,000 to S$54,000 a year
• Plus CPF and overheads
• Plus recruitment and cover
• Leaves when they leave

Against a partial build

Half-building three modules costs more than fully building one, and delivers nothing usable.

• One module fully beats three halves
• Adoption needs completeness
• Partial builds get abandoned

The five-year picture

Licence route roughly S$165,000 to S$195,000 at 20 seats. Build route S$25,000 to S$60,000 plus S$600 to S$3,600 a year hosting.

• Licence: S$165,000 to S$195,000
• Build: S$25,000 to S$60,000
• Hosting S$600 to S$3,600 a year
• EDG may cover up to 50% of the build

What tips it either way

Headcount growth and process specificity. At 5 seats and a standard process, buy. At 40 seats and a specific process, build.

• Under 10 seats, usually buy
• Over 25 seats, usually build
• Specificity decides the middle
• We will tell you which you are

ERP means replacing everything

It does not, and the version that works never does. One workflow, adopted, then the next one.

• One module first
• Existing tools stay
• Replace only what fails

We are too small for ERP

At 5 to 10 staff a first module from S$3,500 often repays inside 12 months against a S$11,520 annual licence.

• S$3,500 first module
• 8 seats at S$120 = S$11,520 a year
• Repays inside 12 months

Custom means expensive

Custom means scoped. Twelve modules is expensive; two or three that matter is not.

• Scope drives cost
• 2 to 3 workflows typical
• Fixed quote after review

We will lose the vendor expertise

You keep it where it matters. We recommend off-the-shelf accounting and integrate rather than rebuild it.

• Keep Xero or similar
• Integrate, do not replace
• We say when to buy

AI will fix our processes

It will not. Agents act on structured data; messy operations produce confident nonsense.

• Clean operations first
• Then automate
• Then add agents

Implementation always overruns

It overruns when scope comes from a wish list. Scoped from one observed workflow, 4 to 8 weeks holds.

• Scope from observation
• Weekly working builds
• Slippage visible by week three

We cannot afford the disruption

One module, one parallel cycle, one team trained. The disruption budget is 8 to 15 hours of your time.

• 8 to 15 hours per module
• One parallel cycle
• Rollback available

Our staff will not use it

They will if it is faster than what they do now. That is the design constraint, and why we watch them work first.

• Faster than current method
• Interview the sceptic first
• Adoption measured weekly

We will be locked in

You own the code, the data and the hosting accounts. Any competent developer can continue it.

• Repository yours
• Hosting in your name
• Documented handover

Grants make packaged software cheaper

PSG funds packaged solutions and EDG funds custom work at up to 50%. Which is cheaper depends on your five-year total.

• PSG: packaged only
• EDG: up to 50% on custom
• Compare 5-year totals
• Confirm with Enterprise Singapore