Deep Legal Analytics
AI-Powered Legal Research

What is "AI-First" Thinking?

"AI-First" does not mean "let AI do everything." It means beginning with data rather than intuition.

When a traditional lawyer encounters a legal question, they typically start from a feeling — an instinct about which statute or provision is likely to apply — and then go looking for evidence to confirm that hunch. This is a form of confirmation bias that is deeply embedded in the legal profession.

An AI-First lawyer does the opposite: collect all available data first, then let patterns surface the answer. The approach mirrors that of a scientist who forms hypotheses from evidence rather than drawing conclusions first and retrofitting data to support them.

In the context of Thai law, this distinction carries particular weight. The Thai legal system is complex — hundreds of statutes, tens of thousands of Supreme Court decisions (Dika), continuously updated ministerial regulations. No single practitioner can hold all of this in memory. AI can process all of it.

The 3-Layer Thinking Model

At the core of Deep Legal Analytics is a three-layer model (Three-Layer Model) in which each layer feeds the next:

Case Study: 311 Ordinances → 74 Overlapping → Consolidation Proposal

The clearest illustration of the 3-Layer Model in action is a project analysing 311 Bangkok Metropolitan Administration ordinances:

Layer 1 (Data): AI read and categorised all 311 ordinances, extracting the key attributes of each — date of promulgation, scope of application, substantive content, current status.

Layer 2 (Pattern): AI identified 74 ordinances containing overlapping provisions. Some were enacted in different eras yet covered identical subject matter. Others contained conflicting provisions without any formal repeal of the earlier text — a pattern that would be invisible to any lawyer reading them one by one.

Layer 3 (Insight): These findings formed the evidentiary basis for a policy recommendation: consolidate the overlapping ordinances, reduce their total number, modernise the surviving text, and eliminate internal contradictions. Such a recommendation is only credible when grounded in data from all 311 sources — not informed guesswork drawn from partial experience.

Traditional Lawyer vs. AI-First Lawyer

The Right Question

In the era of AI, the most valuable lawyering skills are no longer about memory. They are:

• Questioning — Ask the right question and AI returns the right answer. Ask the wrong question and no AI can compensate. Framing the question correctly is a distinctly human, distinctly expert skill.
• Interpretation — AI surfaces data and patterns. Determining what those patterns mean in the specific factual and legal context of a given matter — that is the lawyer's responsibility.
• Strategic Judgment — When there are twenty viable options, selecting the most appropriate course for a particular client in particular circumstances is something AI cannot do alone.
• Communication — Translating complex analytical outputs into terms a client can understand and act upon quickly is a capability that remains irreducibly human.

This is why "AI-First" does not mean "AI does everything." It means: use AI for what AI does better, so the lawyer has capacity for what the lawyer does better.

How to Become an AI-First Lawyer

The transition from a traditional to an AI-First lawyer does not require writing code or studying computer science. It requires changing three habits of mind:

1. From "I know" to "The data shows"
   Before offering any opinion, ask yourself: do I have data to support this? If not — find the data first. Replace confident assertion with evidence-based analysis.
2. From "Search until I find something" to "Search everything first"
   Do not stop at the first Dika decision that supports your position. Search the full body of relevant decisions and examine whether there is a contrary line of authority. Completeness precedes conclusion.
3. From "Do it myself" to "Design the system"
   Rather than reading a contract clause by clause with unaided eyes, design the checklist and analytical framework that AI uses to review it. You become the decision-maker at the final layer, not the first reader at the raw-data layer.