Can AI untangle a century of Philippine foreclosure law?
Intellegal’s Visual Digest, Citation Map and Deep Synthesis take on a problem where one missed exception can change who gets possession—and whether a filing survives scrutiny.
Text size
Intellegal’s Visual Digest, Citation Map and Deep Synthesis take on a problem where one missed exception can change who gets possession—and whether a filing survives scrutiny.
Fluent AI prose is no longer impressive. For lawyers, the useful question is whether a system can find the controlling cases, separate the rule from its exceptions and let counsel verify every step.
Intellegal was put to that test.
The Philippine-focused legal AI platform was asked to identify the requirements for a writ of possession following an extrajudicial foreclosure under Act No. 3135 and determine when a third party already occupying the property could resist it.
Its report drew from statutes, procedural authorities, court issuances and 110 Supreme Court decisions spanning 1941 to 2023. More importantly, it organized those authorities around the facts that actually change the legal answer.
Lawyers now have several AI options.
ChatGPT is a general-purpose assistant, with OpenAI itself cautioning that it can produce incorrect or misleading information. Harvey targets enterprise legal and professional-services work globally.
Intellegal takes a narrower bet: Philippine law, researched against a dedicated legal corpus, with the underlying authorities kept open for inspection.
That distinction matters in legal work. A fabricated citation is more than an embarrassing AI mistake. It gives opposing counsel an opening, weakens a filing and places pressure on a lawyer’s duties of competence, diligence and due care.
A legal AI earns trust one link at a time, from the proposition in its report to the case, statute or rule supporting it.
The research turned first on timing.
During the redemption period, Section 7 of Act No. 3135 requires a sworn ex parte application in the proper court and a bond equivalent to 12 months’ use of the property.
After the redemption period expires without redemption and ownership is consolidated, the purchaser’s position changes. A bond is generally no longer required, while issuance of the writ becomes a ministerial duty upon proper application and proof of title.
Then comes the harder question: Who occupies the property, and under what right?
A mortgagor, agent or successor whose claim derives from the mortgagor generally cannot defeat the summary process merely by attacking the foreclosure.
A genuine third party asserting an independent right adverse to the mortgagor stands differently. The court may have to receive evidence, and the purchaser could be required to pursue an ordinary action for possession.
The cases therefore do not turn simply on whether an occupant objects. What matters is the source of that occupant’s claimed right.
That is the distinction litigators need: rule, exception, burden and remedy, rather than an eloquent version of “it depends.”
Intellegal builds that research process around three connected tools.
Visual Digest turns lengthy decisions into structured views of their timelines, arguments, legal concepts and relationships among parties and courts. Instead of spending an afternoon discovering that a case does not address the right procedural question, counsel can first see how the decision is built.
Citation Map follows the authorities behind a decision, linking cited cases, statutes and legal principles while allowing users to open the underlying decisions. The point is not merely visualization. It keeps a proposition connected to its authority as the research expands.
Deep Synthesis handles questions too broad for a single case. It breaks a request into researchable issues, searches Philippine legal sources and relevant current materials, produces an IRAC-structured report and checks citations against retrieved sources.
In the foreclosure exercise, the tool did more than summarize doctrine. It translated the law into an evidence plan.
A purchaser would need records of the sale, proof of notice and publication, title documents and evidence connecting the occupant to the mortgagor.
The occupant, meanwhile, would need evidence supporting an independent claim, such as a separate chain of title, proof of possession, contracts, tax records, receipts or barangay certifications.
The research retrieved 110 Supreme Court decisions covering 82 years. Of those, 22 were tagged as granted, four partly granted, 75 denied and nine as other or non-merits outcomes.
Those figures should not be mistaken for predictive analytics.
A petition marked “denied” does not necessarily reveal which party ultimately prevailed in the underlying property dispute. Nor does a retrieved collection of cases constitute a statistical model capable of forecasting the next judgment.
Its usefulness is more modest and arguably more practical: lawyers can see the shape of the jurisprudence, identify clusters and decide where close reading should begin.
Foreclosure is only the stress test.
The same research problem runs throughout Philippine legal practice: statutes sit in one place, rules and circulars in another, while decades of judicial decisions qualify, refine or sometimes narrow what came before.
AI can shorten that research path. It cannot decide whether a witness is credible, authenticate a deed or discover a missing fact nobody supplied.
Those judgments remain with counsel.
That may ultimately be the more useful measure of legal AI.
Fluent answers are cheap.
A research trail that survives inspection is not.
