Legal AI for Evidence Management · Miguel Enes

Legal teams can spend substantial time organizing scanned documents, email exports, exhibits, and matter files before substantive review begins. illuma.law is a legal AI platform for evidence and matter workflows, designed to support this preparation with traceable processing rather than an ungrounded chat interface.

I am Miguel Enes, founder of illuma.law. My background includes payments and platform engineering at Lazada and NEW WORK SE; at illuma.law I work across ingestion pipelines and API delivery.

The problem with evidence review

Evidence review is high-volume, high-stakes, and unforgiving of shortcuts:

• Documents arrive in mixed formats — PDFs, scans, email threads, spreadsheets — with no consistent structure.
• Reviewers must classify and tag case documents before anyone can evaluate the case, so everything downstream waits on this step.
• Generic AI tools summarize text but cannot show where a conclusion came from, which is unacceptable when a tag influences case strategy.

The result is manual review effort that scales linearly with the case, and legal teams that serve paperwork instead of clients.

Delivery scope

illuma.law covers the evidence-to-evaluation pipeline end to end:

• Ingestion and retrieval. Python agents retrieve, parse, and summarize legal documents from internal databases, so matter files become queryable input instead of static attachments.
• Triage and classification. A rule-based and machine learning pipeline classifies and tags case documents, reducing manual review effort by half in the internal benchmark through automated first-pass categorization.
• Case evaluation support. Data pipelines feed precision case evaluation, so reviewers spend attention on judgment calls rather than sorting.
• API delivery. The platform exposes results through a FastAPI layer, integrating with the tools legal teams already use.

Deployment stays deliberate: your data, your workflow, and traceability for every automated decision.

Evidence from building illuma.law

This page describes a platform I build and operate, not a resold stack:

• I programmed the Python agent systems that retrieve, parse, and summarize legal documents from internal databases.
• I developed the rule-based and machine learning pipeline that classifies and tags case documents, reducing manual review effort by half in an internal benchmark during the build.
• I architected the full stack — Python with FastAPI, spaCy, and PostgreSQL — for data ingestion, NLP processing, and API delivery.

The full venture record, including the delivery ledger behind these points, is in the illuma.law dossier on the work page, with more product context in what illuma.law automates for legal workflows.

Who this is for

illuma.law fits teams whose bottleneck is evidence volume rather than legal headcount:

• Law firms with document-heavy practices — litigation, disputes, regulatory — where first-pass review delays every matter.
• In-house legal teams that inherit discovery bundles and need matters triaged before outside counsel hours start running.
• Legal operations leaders who need classification decisions that are auditable, not black-box suggestions.

If your review bottleneck is a different document domain — due diligence, claims, compliance — the same pipeline pattern applies; ask.

Work with me

Two ways to go deeper:

• Explore the product at illuma.law.
• Or discuss a build with Miguel directly — I answer for the architecture, the pipeline, and the delivery.

For the underlying concepts, start with multimodal legal evidence, explained; for the build philosophy behind owning your stack, read client-owned AI systems.