LAW KG · DESIGN-PARTNER GATED

NO FIRM SIGNED YET · ACTIVELY SEEKING

Matter memory. Without sending a single matter to a public model.

Private knowledge graph for legal practice. Matters, parties, precedent, documents — indexed on infrastructure your firm controls. No public LLM. No data leakage. Full audit trail.

THE PROBLEM

Institutional memory walks out when partners retire.

DMS

Filing cabinet, not graph

You can find a doc by name. You cannot ask “what did we argue last time?” The question is unanswerable.

SCATTERED MEMORY

Email · Slack · Word · DMS

The longest-tenured partner is the search index. When they retire, the index walks out the door.

PUBLIC LLMS

Every keystroke a breach

In-house counsel won't approve them. Firms that need this most are structurally blocked from off-the-shelf.

WHAT YOU GET

Three layers. One private graph.

MATTER GRAPH

Linked · queryable · versioned

Every matter, party, opposing counsel, jurisdiction, outcome, billing record. Schema designed with practising legal-tech leads.

PRECEDENT + EVIDENCE

Provenance end-to-end

Documents, exhibits, expert reports, decisions linked to matters. Full chain of custody for ethics and discovery.

PRIVATE AI

Schema-only access

Natural-language queries. AI reads the schema; never sees raw documents unless permissioned. Every call audit-logged.

Full technical depth — schema, deployment shapes, audit-trail spec, privilege protection — at /kg/law.

PRIVACY · CONTROL · AUDIT

Why in-house counsel will sign.

Customer-controlled

Firm-owned infrastructure. Air-gapped option. No shared tenancy. No vendor-side index. Exit guaranteed.

No public-LLM exposure

Local or customer-deployed models. Matter content never leaves your perimeter. Schema goes to the model; bytes do not.

Audit trail per query

Every question, answer, document touched. Exportable for regulator and partner review.

WHO BUYS IT

Mid-size firms. 20–200 lawyers. Tech-forward leadership.

Managing partners

20–200 lawyer firms.

Watched a partner retire taking 25 years of precedent with them.

In-house counsel · CLOs

Tech-forward legal teams.

Rejected at least one generative-AI vendor on data-leakage grounds in the last 12 months.

Practice-group leads

Litigation · M&A · regulatory.

Junior associate spent three weeks reproducing a brief that already existed in the firm's files.

IT security · DPO

CISO · compliance leads.

DMS is the only system that knows where anything is — and you don't trust it to be complete.

DESIGN PARTNER PROGRAMME

First firm not yet signed. Looking for the right one.

Six-month engagement · signed scope · exclusive launch access · named technical contact · paid.

YOU COMMIT

6 months · 1 stakeholder

Signed scope on a single practice group. Weekly syncs. Honest feedback on schema fit.

CITHORUM COMMITS

Named CTO · exclusive slot

Spyros Argalias (CTO) as technical contact. Custom schema. No competing law-firm partner during the engagement.

YOU PAY

~⅓ of post-GA list

Single prepaid invoice. Funds the engineering — not margin extraction from the partner doing us a favour.

YOU GET AFTER

36-month price lock

First-mover position on GA. Locked SaaS pricing for 36 months. Co-marketing rights (opt-in). Graph is yours regardless.

First slot open. Air-gapped deploy where IT-security posture requires it.

Apply for design partnership →

PRICING

Design partner first. SaaS after.

Design partner: ~⅓ of post-GA SaaS list, prepaid. Post-GA: per-firm-per-month, base matter quota plus per-query metering.

DESIGN PARTNER

Six-month scoped engagement

Single prepaid invoice. Named CTO. Custom schema. Exclusive launch access.

Apply →

GA · POST-DESIGN-PARTNER

Per-firm / month

Base matter quota + per-query metering. Monthly invoice. 36-month price lock for design-partner alumni.

Talk to us →

ENTERPRISE / AIR-GAPPED

Custom

Air-gapped on firm-controlled hardware. Sovereign-jurisdiction options. Above ~200 lawyers or special-category data.

Contact us →

First firm not yet signed. Six-month paid engagement.

Tech-forward 20–200 lawyer firms whose legal team has rejected at least one generative-AI vendor on data-leakage grounds.