The Option to Purchase on a S$2.1 million resale condominium unit was exercised on a Tuesday. Completion is ten weeks out, and before the associate drafts a single requisition, five searches have to come back and get read line by line: title search, bankruptcy search on the seller, CPF search, and a check against the management corporation’s outstanding levies.
AI for conveyancing in Singapore means using a system to read those search results as they land, flag the entries a requisition depends on, and draft the first cut of the requisitions letter and the completion account, before a conveyancing lawyer reviews and sends them. It does not replace the search, the lawyer’s read on materiality, or the Singapore Land Authority’s Digital Conveyancing Portal (DCP), which is rolling out through 2026 to move conveyancing steps online. It removes the reading time sitting between the searches landing and the requisitions going out, verified against SLA’s own DCP timeline and MinLaw’s Guide for Using Generative AI in the Legal Sector, published 6 March 2026 on mlaw.gov.sg.
This piece covers where the hours in a conveyancing file actually go, what an AI system changes and what the DCP does not, how to stay inside the MinLaw guide on a file full of personal and financial data, and the grant that offsets the cost.
Where the hours actually go in a conveyancing file
Open any resale or sale-and-purchase file and the work splits the same way it does in due diligence: work that needs a lawyer’s judgment, and work that does not. Conveyancing is mostly the second kind, repeated on a fixed pattern across dozens of files a year.
- Ordering and reading the title search, the bankruptcy search on the seller, and, where CPF or HDB is involved, the CPF search and the outstanding-levies check with the management corporation
- Cross-referencing every search result against the Option to Purchase and the draft Sale and Purchase Agreement, checking for encumbrances, caveats, and easements the OTP did not disclose
- Drafting requisitions on title and reading the replies against a checklist that repeats on almost every file: outstanding property tax, planning contraventions, unauthorised structures, subdivision consents
- Preparing the completion account: apportioning property tax and maintenance, checking the redemption sum against the seller’s bank discharge letter, reconciling the balance to the dollar
- Repeating parts of the exercise close to completion, because a search result or a bank’s redemption figure can move in the weeks between exercise and completion
None of that is legal analysis a client is paying a lawyer’s judgment for. It is reading a fixed set of documents against a checklist that barely changes from file to file, at a volume that scales with how many conveyancing matters a practice runs at once. A small practice running twenty files a month is reading the same five search types twenty times over, with the one clause that matters buried somewhere different in each result.
What AI changes in Singapore conveyancing, and what the Digital Conveyancing Portal does not
Parliament passed the Electronic Conveyancing and Other Matters Bill on 15 October 2025, clearing the way for SLA’s Digital Conveyancing Portal (DCP), which is rolling out in three phases: the Option to Purchase stage first, then pre-completion and completion for developer sale transactions, then the same stages for resale and sub-sale deals. An early-2026 pilot brought the OTP phase to resale private residential transactions specifically, with full coverage across public housing, private housing, and commercial and industrial property targeted for later in 2026, verified 2026-08-11 on sla.gov.sg and MinLaw’s second reading speech for the Bill.
phases SLA is rolling the Digital Conveyancing Portal out in, from the option-to-purchase stage through to full pre-completion and completion coverage by 2026
Singapore Land Authority, Digital Conveyancing Portal press release, verified 2026-08-11
The portal changes where a step happens: e-signing instead of a physical signature, remote witnessing over a video call instead of a meeting room, digitised document submission and e-payment instead of a courier. It does not change who decides whether a flagged caveat is routine or a problem. A search result filed through the DCP still needs someone who knows which line matters. A requisition still needs someone who has cross-checked the seller’s reply against the OTP. The portal moves the paperwork. The reading stays exactly where it was.
That is the gap an AI conveyancing system is built for: a system that reads every title, bankruptcy, and CPF search as it comes back, flags the entries the requisitions checklist depends on, and drafts the first cut of the letter and the completion account, whether the search itself arrived through the DCP or the old way.
Representative arithmetic, conservative throughout. A conveyancing associate spends roughly three hours a file reading the search results, cross-referencing them against the OTP, and drafting the first draft of the requisitions letter. A practice running twenty live files a month is spending around sixty associate-hours a month on that reading and drafting alone, before anyone reaches the completion account.
- Manual search review + requisitions drafting, per file
- 3.0 hours
- AI-assisted first pass, associate checks the flags and drafts
- −1.5 hours
- Associate time per file after the flagged review
- 1.5 hours
At twenty files a month, that is roughly thirty associate-hours returned every month. At a loaded associate cost of S$65 an hour, that is close to S$2,000 a month, or around S$23,000 across a year, on a workflow that repeats on every file the practice takes on rather than a one-off saving on a single deal.
Staying inside the MinLaw guide on a file full of personal data
A conveyancing file holds a client’s NRIC, bank account details, CPF balances, and the price they paid for their home. Speed does not loosen the standard the MinLaw guide sets, and if anything a conveyancing file is a stricter test of it than most other legal work, because of how much personal and financial data sits inside a single search result.
Three things carry the weight. Oversight matched to risk: a flagged encumbrance or an unexpected caveat has real financial consequence for the client, which is exactly the human-in-the-loop review the guide expects, a lawyer checking every flag before a requisition or completion figure goes out, not a sample of them. Confidentiality: the search results and the completion account are confidential client information under Rule 6 of the Legal Profession (Professional Conduct) Rules 2015, so the system runs on the firm’s own tenant, with the provider contractually barred from training on the firm’s documents. Transparency: if AI is used substantially in preparing the requisitions or the completion account, the guide points toward disclosing that in the engagement letter, with an opt-out.
Our explainer on what the MinLaw guide asks of law firms covers the full five-step framework, and the closest workflow parallel, first-pass review of a due diligence data room, is in AI document review for Singapore law firms. The mechanics carry over almost exactly: read the documents, flag what the checklist depends on, let the lawyer decide.
The EDG grant path for AI conveyancing in Singapore
Most Singapore conveyancing practices clear the SME test the Enterprise Development Grant uses: Singapore-registered, at least 30% local equity, group turnover of S$100 million or less, or group headcount of 200 or fewer. EDG covers up to 50% of a qualifying custom project, consultancy fees, software, and internal manpower, verified July 2026 on enterprisesg.gov.sg. A Build & Train engagement for a conveyancing system runs S$25,000 to S$45,000, fixed scope. At 50% support that is roughly half back, subject to EnterpriseSG approval, never guaranteed and never automatic.
File through the Business Grants Portal before paying a vendor anything. Processing takes 8 to 12 weeks, so the application has to lead the build, not follow it. EnterpriseSG is folding EDG, PSG, and Market Readiness Assistance into the EDGE grant in the second half of 2026, and the transition rules for applications still in the queue at that point are not published yet. Our EDG grant guide for law firms covers the consultant-certification rule and the full timeline in more detail.
Why the conveyancing secretary still has to run it
A system that only a vendor can retune does not survive the next document format, the DCP’s next rollout phase, or the day the vendor’s support contract lapses. It also fails the guide’s own transparency test the moment a lawyer cannot explain, in the firm’s own words, how a flagged entry was checked. Speed that depends on someone outside the firm is not speed the firm can rely on once the DCP changes how a search result arrives again.
The alternative is training the associates and conveyancing secretaries who already run these files to operate and adjust the system themselves: a new search format, a new requisitions checklist item, a junior joining the team. The capability stays inside the practice instead of being rented back on every file. The wider argument for why that matters more than the tool itself is in why AI projects fail, and the full pricing breakdown for an engagement like this is in what AI consulting costs in Singapore.
Start with one file type, not the whole conveyancing desk. An assessment maps where a specific team’s hours go, on that team’s own numbers, before anything is built. Our engagement outline walks through how that works.
Common questions
What does AI actually do on a Singapore conveyancing file?
It reads the title search, bankruptcy search, and CPF search results as they come back, flags the entries a requisition depends on, and drafts the first cut of the requisitions letter and completion account. A conveyancing lawyer checks every flagged entry and every draft before it goes to the other side or the client. The system reads. The lawyer verifies.
Does the Digital Conveyancing Portal replace the need for AI or a lawyer?
No. The DCP, the Singapore Land Authority's platform rolling out through 2026, moves conveyancing steps such as e-signing and document submission online. It does not read a search result or judge whether a flagged encumbrance matters. That reading and that judgment are what an AI system, and ultimately the lawyer, still do.
Is AI conveyancing review compliant with Singapore's MinLaw guide?
It can be, if the workflow follows the guide's three principles. Oversight matched to risk means human-in-the-loop review on anything with legal or financial consequence, such as a flagged encumbrance. The system runs on the firm's own tenant, with training on client data contractually barred. And the firm can explain to a client how an output was checked. The guide, published 6 March 2026, is non-binding but interprets Rule 5 and Rule 6 of the Legal Profession (Professional Conduct) Rules 2015.
Does the EDG grant cover an AI conveyancing system for a law firm?
For a qualifying firm, yes, up to 50% of a custom Build & Train engagement, verified July 2026, subject to EnterpriseSG approval and applied for through the Business Grants Portal before any vendor is paid. A S$40,000 project comes to roughly S$20,000 net. Our EDG grant guide for law firms covers the full process.
