One Digital Employee. Six named agents.

The named Digital Employee. Justine is the visible identity the attorney supervises. She decomposes every matter into the stage-specialized sub-tasks the matter actually contains, dispatches each sub-task to the right agent below, composes the outputs into a single work product, and surfaces what needs the attorney’s decision. The supervisor pattern is what makes the architecture coherent — there is one named entity to supervise, not a constellation of separately-licensed AI tools.

Worked example — A new PI matter lands on Justine’s desk. She decomposes it into intake, medical chronology, valuation modeling, strategy posture, and (when in suit) deposition prep. Each sub-agent below works the slice; Justine assembles the dossier; the attorney reviews and signs.

Conversational and document-driven intake. Liability vector. Mechanism of injury. Jurisdiction. Statute-of-limitations clock. Coverage discovery (PIP / med-pay / UM/UIM identification). Client-conflict check. Per-field provenance to source document. Validator-council confirmation across sources. The intake agent is the entry point of the matter and the calibration point for the supervisor.

Worked example — You drop the police report, the ER intake form, and the registered-owner declaration page into the workspace. The intake agent extracts each field — CA jurisdiction, 2-year SOL clock, at-fault carrier, layered UM coverage — and shows the source page for every extracted value. A validator council confirms the SOL clock and surfaces a single cross-source conflict on the named driver, which you resolve before assembly.

Multi-provider medical chronology. Treatment-gap detection. ICD-10/CPT mapping. IME / DME response. Medical-bill audit. Permanent impairment rating modeling. Causation linkage to mechanism of injury. The Medical Agent reads the matter’s entire record set through Llama 4 Scout’s 10M-token context — chunked retrieval is not in the path.

Worked example — A 5,000-page record set across nine providers. The medical agent surfaces the 19-day treatment gap that the carrier’s IME will exploit, the two PT notes that contradict the IME’s functional-capacity opinion, and the three records that establish causation to the underlying accident — with the exact paragraph quoted in each finding.

Special damages tally. General damages comparable analysis. Future-medical cost projection. Policy-limit awareness. Lien total. Jury-verdict comparables for the venue. The Valuation Agent produces the demand range, the bottom-line authority, and the settlement-matrix model used in negotiation.

Worked example — Special damages $87K. General damages comparables for L4-L5 disc herniation in San Diego County run $180K–$320K. Lien total $42K. Policy limit $100K + $250K UM. Pre-suit demand: $385K. Bottom-line authority: $215K. The model is defensible to the carrier and to the partner reviewing the file.

Adjuster pattern recognition. Demand framing for the specific matter. Negotiation tree. Mediation positioning. Filing decision (settle vs sue). Discovery plan when the matter goes into suit.

Worked example — Allstate adjuster (this one specifically) settles soft-tissue under policy limit 88% of the time when the demand is framed around future-medical and includes a sworn provider affidavit. The Strategy Agent recommends including the affidavit in the demand and structuring the opening number 15% above the projected acceptance range.

Deposition preparation: outline construction from the matter, the opposing party’s prior testimony, and the produced documents. Live deposition analysis when the transcript is loaded. Opposing-expert Daubert pre-read. Motion-in-limine drafting. Witness-preparation script.

Worked example — You set a Thursday deposition of the carrier’s claim handler. The Deposition Agent assembles the prior testimony you have on this adjuster (across earlier matters where it’s public), the contradictions in the claim file produced in discovery, and the three reservation-of-rights letters that are inconsistent with each other. The deposition outline is forty questions — yours to revise and run.

A standalone advisory surface. The attorney asks Justine a strategic question — "we got a counter-offer of $185K, should we accept or counter; what’s the leverage path" — and the Case Advisor produces a structured strategic answer grounded in the matter’s record, the venue’s comparables, the carrier’s prior behavior, and the firm’s historical settlement patterns. Distinct from the in-matter Strategy Agent because the Case Advisor surface is the firm’s strategic counsel for the file, not the sub-agent that produces the in-matter strategy section.

Worked example — A demand-and-counter cycle has stalled at $185K against a $385K demand. The attorney asks the Case Advisor for a recommendation. The Advisor reviews the matter (Llama 4 Scout reads the whole file), benchmarks against the firm’s prior cases of this shape, considers the carrier’s tendencies, and recommends a final counter of $245K with a fourteen-day acceptance window plus a sealed Brandt-fees preservation letter — with the reasoning visible at every step.