How to Choose an AI Consultant in Singapore: 8 Questions

To choose an AI consultant in Singapore, check three things before you sign: what they will measure in your business before they start, whether they will tell you which work should stay manual, and what you keep if you stop after the first step. A good consultant answers all three in writing. The eight questions below cover the rest, from who builds the work to what it costs to run each month.

Take an example: a 15-person freight forwarding company. Two coordinators copy shipment details from customer emails into the job system, and customers phone through the day to ask where their cargo is. The owner has heard that AI could handle both, and three consultants have offered to help. Each pitch sounds different, and the owner has no way yet to compare them.

Cortex Lab AI is a Singapore consultancy that advises SMEs on AI and builds the systems that come out of that advice.

What does an AI consultant do for an SME?

An AI consultant works out which parts of your business AI can handle reliably, what that would cost, and whether the saving justifies it. For an SME, that usually means the repetitive work between your tools and your customers: enquiries answered by hand, reports rebuilt every week, details copied from one system into another.

A good consultant also tells you where AI is the wrong answer. Some work is too irregular, too low in volume or too dependent on judgement to automate at a profit. A template, a checklist or a setting in software you already pay for often does the job for less.

Some consultants stop at the written advice. Others also train your staff and build the automations inside the tools you already pay for. Our guide to AI integration explains how that building work connects the tools you have.

In the example, a consultant would start by asking how many shipment emails the coordinators handle a day, how long each takes to copy, and how often a typing mistake causes a problem later. They would ask the same about the tracking calls, and how many of them an automatic status message would answer.

Do you need an AI consultant at all?

Not always. Three situations call for a different starting point.

• You already know the one task you want handled, and a ready-made tool does it well. Start with the tool.
• Your team has not yet used the AI features in software you already pay for. Start with training on the tools you have, which costs less than a project.
• The work is low in volume or changes every month. Automating it would cost more to maintain than it saves, so keep it manual for now.

A consultant earns the fee when you have several candidate tasks and cannot tell which would pay back, when the work crosses two or more systems, or when you want the result measured against a baseline before you spend more. Our guide to AI for Singapore SMEs covers where to start on your own.

In the example, the freight forwarder has two candidate tasks, the email copying runs across two systems, and the owner wants to know which pays back first. That is a job for a consultant.

How do AI consulting firms in Singapore differ?

Mostly by the size of client they are built for and what you are left with at the end. Large firms are set up for enterprise programmes. Small consultancies are set up for an SME that wants one working change before it commits to a strategy.

Global firms are the large strategy and technology consultancies. Integrators replace or connect core systems such as ERP, the main software that runs finance and operations, and AI is often one part of a larger system sale. Vendor partners set up the platform they resell. An independent consultant gives you one person's advice, which works until that person is unavailable.

In the example, the freight forwarder has 15 staff and one job system, so a global firm or an ERP integrator is built for a much larger project. If one of the three consultants resells a single logistics platform, ask whether they would recommend a different tool if it fitted better.

What should you ask an AI consultant before hiring?

Ask these eight questions before you sign, and get the answers in writing.

1. What will you measure before you start, and how will we see the result? A good answer names a figure from your business, such as hours a week on a task or errors a month, and says when it will be measured again.
2. Which of our processes would you advise us not to automate, and why? A consultant who finds nothing to leave manual has not looked closely.
3. Can you show us something you built that a business uses today? A live system, even a small one, tells you more than a slide of client logos.
4. Who builds it, and who do we contact when it breaks? Get names. The person in the pitch is not always the person who does the work.
5. Does it run in tools we already pay for, or in a platform we must keep paying you for? If it is the second, ask what happens to the work when you stop paying.
6. Is our data used to train any AI model, and will you put the answer in writing? Ask where your customers' data is stored as well. Our PDPA checklist for AI chatbots lists what to check under the PDPA, Singapore's privacy law.
7. What happens when it gets something wrong, and who on our side is told? AI tools get some answers wrong. A good answer describes how a person catches the mistake and who fixes its cause.
8. What will it cost to run each month after handover? Ask for the project fee and the monthly running cost as separate figures. Software subscriptions, AI usage and support often sit outside the project quote.

If the consultant is quoting a chatbot, add the nine chatbot vendor questions, which cover WhatsApp message fees and who keeps the phone number.

In the example, a good answer to question 1 is a week of the coordinators logging how long each shipment email takes, before anyone quotes for automating it.

What are the red flags when hiring an AI consultant?

• A long strategy deck before anything is built. The most expensive AI projects we see are the ones with the longest slide decks. Ask what will be working, and measured, within the first few weeks.
• A return promised before they have seen your figures. Nobody can price a saving without knowing how long the work takes today.
• AI for every process. A proposal with nothing left manual has not weighed what each automation costs to run.
• Work you cannot take with you. If the automations live in the consultant's own platform, you pay for them for as long as you use them. Ask who owns the work outright.
• No exit after the first step. You should be able to stop after the assessment or the first build without a penalty.
• An offer to handle your grant. Enterprise Singapore's EDGE grant FAQ says "Submissions must not be made by vendors, service providers, or any other third party, on behalf of the grant applicant." It also says businesses "must not have started work or made any payment or deposit to a supplier, vendor, or third party" before they submit. A consultant who asks for a deposit first, or offers to apply for you, can cost you the grant. Our guide to AI grants after PSG explains what EDGE covers.

Why do most organisations get no return from AI?

Usually because of how the project was run. MIT Project NANDA's report, The GenAI Divide: State of AI in Business, 2025, found that 95% of organisations were getting zero return from their investment in generative AI (AI that writes text or creates images). It drew on a review of over 300 publicly disclosed AI initiatives, interviews with representatives of 52 organisations and survey responses from 153 senior leaders. It found the same divide among small and medium-sized businesses as among large enterprises. The authors conclude that the divide "does not seem to be driven by model quality or regulation, but seems to be determined by approach." Of the enterprise AI systems organisations evaluated, they say most "fail due to brittle workflows, lack of contextual learning, and misalignment with day-to-day operations."

Three habits address those causes, whoever you work with.

• Write down the number first. Record a baseline before work begins, using your own figures, such as hours spent, volumes handled and error rates.
• Start small. Begin with the smallest change that pays for itself. Add the next piece only once the last one has held up in real use.
• Hand over properly. Name an owner on your side, write down who is told when something breaks, and keep the baseline figures, so any drop shows up as a number.

The first change can be small. In one of our builds, a building maintenance manager with a team of five technicians used to read fault reports across several chat groups and check each technician's schedule before assigning a job. We brought the reports into one list and built one dashboard showing who is free and qualified, so a job is assigned in one step. By the manager's own count, that saves about 1 hour a day, roughly 5 hours a week. The building maintenance job dispatch case study has the detail.

In the example, the smallest first change is the email copying on its own, measured against the coordinators' logged timings, before anyone touches the tracking calls.

What is an AI readiness assessment?

An AI readiness assessment scores how prepared an organisation is to use AI, across areas such as leadership, data and governance. In Singapore the best-known one is the AI Readiness Index (AIRI), first developed by AI Singapore. Its current version, v3.1, scores an organisation on five pillars: Leadership and Culture, Ethics and Governance, Business Value, Data Foundation, and Infrastructure and Standards. It then places the organisation on a scale from AI Unaware to AI Catalyst. AI Singapore's website offers the assessment free, with 15 questions that take about 20 minutes. That is useful if you want a benchmark.

A consultant who starts from the work asks narrower questions: which tasks, how often, how long they take, and what a mistake would cost. A business can score low on a readiness index and still have one task worth automating next month. A business that scores high can still have nothing that would pay back this year.

In the example, the freight forwarder has no data team and no written AI policy, which pulls its AIRI score down. Neither stops the email copying from being worth automating.

Cortex Lab AI's free 30-minute consultation starts from the work and ends with a written report on where AI would pay back in your business, and where it would not.

Frequently asked questions

How much does an AI consultant cost in Singapore?

It depends on the type of firm and the scope, and most firms quote after a first call. D'Marketing Agency, a Singapore agency, lists one-to-one AI consulting with a written plan from $1,500, and a half-day AI workshop at your office for up to 12 people from $2,400. AI Singapore's AIRI readiness assessment is free.

Two government schemes change the net cost. The IRAS e-Tax Guide on the Enterprise Innovation Scheme gives a 400% tax deduction on the first S$50,000 of qualifying AI expenditure each Year of Assessment, for YA 2027 and YA 2028. That includes a "consultancy and strategy service" used to develop, deploy, operate or maintain an AI system, but not general management consultancy. Larger firms can look at the Enterprise Compute Initiative, which supports 70% of consulting costs, capped at S$105,000, for companies with at least 10 staff in Singapore, a tech team of two or more, and experience building a custom AI solution. Whichever firm you choose, ask for a fixed quote and the monthly running cost in writing.

Do we need a technical team to work with an AI consultant?

No, for most SME projects. The consultant or their builders do the technical work. Your side needs someone who knows how the work runs today and can check the result against it.

How do we know it will work outside a demo?

Run the first version on your real work for a few weeks, then compare the figures with the baseline you recorded before any work began. If the consultant did not record a baseline, you have nothing to compare against.

Can an AI consultant apply for the EDGE grant for us?

No. Enterprise Singapore's EDGE grant FAQ says submissions "must not be made by vendors, service providers, or any other third party, on behalf of the grant applicant." You apply yourself, before any work starts and before you pay any deposit.

Sources checked on 18 October 2026, except MIT Project NANDA, checked on 5 October 2026: MIT Project NANDA, The GenAI Divide: State of AI in Business, 2025 (July 2025; PDF copy hosted by MLQ.ai); AI Singapore, AI Readiness Index (AIRI); IRAS e-Tax Guide, Enterprise Innovation Scheme (third edition, 31 August 2026, Annex F); Digital Industry Singapore, Enterprise Compute Initiative; Enterprise Singapore, EDGE grant FAQ; D'Marketing Agency, AI consulting and training.