FirmOps.io
← Back to Managed Firm Brain

Conduit Law operating proof

Updated November 8, 2026. Aligned the case-launch numbers with the public live-lab proof and added how the approval gates map to the State Bar of California's 2026 AI guidance.

The live lab behind FirmOps is not one automation. It is an operating system of supervised loops.

Conduit Law gives FirmOps a real personal-injury environment to test what law-firm AI must do: read across messy systems, prepare useful work, show its reasoning, and stop before anything sensitive changes without approval. The Managed Firm Brain was built here, in the same live lab behind 2.5h→10m case launch, 1,500+ cases, and 47→1 click. Meet Jon and Elliot. Jon founded Vertana and is the non-attorney founder and Firm Manager of Conduit Law; Elliot is Managing Partner.

01

Intake & qualification

02

Case opening

03

E-sign & packets

04

Documents & OCR

05

Records & treatment

06

Client communications

07

Payments & trust

08

KPIs & dashboards

A public-safe map: category names only, no client facts, internal endpoints, credentials, or matter-level details.

1,500+

matters in the live lab

Patterns are pressure-tested in a working PI firm before becoming FirmOps playbooks.

47→1

case-launch steps compressed

A repeated intake-to-matter pattern informs practical Managed Firm Brain builds.

2.5h→10m

case launch

One case launch went from about 2.5 hours of manual setup to roughly 10 minutes of reviewed work.

10+

operation families mapped

Intake, documents, records, communications, approvals, KPIs, SEO, and more.

See it live

Watch a FirmOps-managed AI agent running in the wild.

jonny.bot is a public-safe, chronological log of what this agent actually does day by day. Shape only, human-approved. Real proof that this is running, not a roadmap.

Open jonny.bot →

Automation atlas

The scale is the number of operating categories FirmOps has had to make safe.

A single demo workflow is easy. Conduit’s value is breadth: the same approval-first pattern has to survive intake, documents, records, payments, reporting, marketing, and governance without turning private firm operations into public content.

01

Intake & qualification

  • • lead capture
  • • qualification scripts
  • • handoff summaries

Safety: human review before client commitments

02

Case opening

  • • matter creation
  • • folder setup
  • • task templates

Safety: field mapping and duplicate checks

03

E-sign & packets

  • • signed packet status
  • • missing-form flags
  • • case-opening checklists

Safety: no document editing after signature

04

Documents & OCR

  • • safe PDF handling
  • • source preservation
  • • review queues

Safety: restore paths and hash checks before retries

05

Records & treatment

  • • provider follow-up
  • • records status
  • • treatment referrals

Safety: staff approves external sends

06

Client communications

  • • draft updates
  • • SMS review cards
  • • inbox triage

Safety: no unsupervised client sends

07

Payments & trust

  • • payment requests
  • • paid-state review
  • • matter-linked notes

Safety: approval and accounting checks first

08

KPIs & dashboards

  • • firm KPIs
  • • channel ROI
  • • manager dashboards

Safety: definitions and source freshness are explicit

09

SEO & marketing ops

  • • recommendation loops
  • • GMB assets
  • • sitemap verification

Safety: public claims must be source-backed

10

Governance & approvals

  • • approval cards
  • • dry-run gates
  • • incident triage

Safety: high-risk actions require named approval

Full workflow catalog

These ten families sit on top of 250+ mapped workflow candidates and operating loops, plus hundreds of operating surfaces across jobs, tools, integrations, and agent playbooks. See the full public-safe catalog, tagged by autonomy level, and pick where to start.

Explore all use cases

Secret-safe by design

What we intentionally do not disclose.

The case study should prove scale without publishing the private map of Conduit’s operations. Public buyers need the pattern and operating judgment, not credentials, endpoints, matter facts, or client-level examples.

  • No client names, case facts, medical details, emails, phone numbers, or document links on the public page.
  • No API keys, webhook URLs, tokens, internal queue names, database table details, or credential locations.
  • No legal-advice claims, autonomous-lawyer claims, or promises that a workflow replaces attorney judgment.
  • Only public-safe system categories, rounded scale markers, and reusable operating patterns are shown.

Reusable operating loop

Every useful automation follows the same five-part path.

1

Signal

A matter, lead, document, payment, KPI, or marketing event needs attention.

2

Context

FirmOps reads approved sources and reconstructs the state without exposing raw client data.

3

Draft

The assistant prepares a summary, checklist, message draft, task plan, or exception report.

4

Gate

A human approves, edits, rejects, or escalates before external or record-changing action.

5

Measure

The workflow logs what happened so the next build gets safer and more useful.

What this proves

The Managed Firm Brain is backed by real operating pressure.

Conduit forced the same practical questions a new client brings: what data can safely be used, which draft is useful enough to review, what needs a checklist or tracker, and where a human must approve before the system changes the outside world. FirmOps starts there, then connects deeper only when a workflow proves value.

Read approved context

The system starts by reading approved firm context across matter data, intake, documents, tasks, communications, and reporting before it recommends or drafts anything.

Prepare supervised work

Agents draft summaries, checklists, follow-ups, task plans, and exception reports with enough source context for a human to inspect the work quickly.

Hold risky actions

Client-facing sends, record changes, payment actions, public posts, and legal-ops decisions stay behind human approval gates and audit trails.

Why the gates exist

The approval gates track what bar guidance asks of agentic AI.

The "Hold risky actions" layer was not a design preference. In its 2026 Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law, the State Bar of California's Committee on Professional Responsibility and Conduct says "the greater the degree of autonomy afforded to an AI system (i.e., agentic systems), the more important it is for a lawyer to implement supervisory controls and verification mechanisms," and that "the degree of lawyer diligence and supervision must correspond to the level of system access and autonomy."

  • • External sends wait for a person. The guidance says a lawyer "must not deploy an agentic AI system in a manner that permits autonomous external transmission of client information, including automated communications, filings, or data transfers, without appropriate safeguards and human review." At Conduit, client messages, records requests, and payment actions are drafted, then approved.
  • • Access is scoped first. The guidance asks lawyers to "carefully evaluate and limit the scope of such access." That is why the Managed Firm Brain starts read-first and connects deeper only when a workflow earns it.
  • • Judgment stays with the lawyer. The agent prepares the work. Attorneys at Conduit decide legal questions, and nothing on this page claims a workflow replaces that judgment.

Source: State Bar of California, Practical Guidance for the Use of Generative AI in the Practice of Law (2026). FirmOps is not a law firm and this is not legal advice. For the policy side, see the law firm AI policy guide, the ethics overview, and how security and access controls work.

The Conduit workflow pattern

A measured case-launch workflow at Conduit took one case launch from about 2.5 hours to roughly 10 minutes, and 47 manual steps became one confirmation. More important than the time savings: the workflow made matter creation, task setup, folder structure, notifications, and reporting visible as one repeatable pattern.

  • • CRM contacts and matters are created with firm-specific field mapping.
  • • Task templates and document structure follow the firm’s operating rules.
  • • Staff see the same source of truth instead of chasing updates across tools.
  • • The agent can explain what happened and what needs human attention next.

Why this matters to buyers

You see a real system.

A demo can show an owner how one bottleneck becomes source-linked evidence and approval-ready work. That is easier to judge than an abstract AI roadmap.

The product gets sharper after each useful workflow.

Every useful build teaches which CRM, API, export, approval, and reporting patterns deserve connection next. Read-only first; deeper FirmOps automation only when it earns its place.

Want to bring one workflow to the live demo?

Bring one recurring workflow to the live demo. We will show how the Managed Firm Brain starts read-only, keeps supervised work behind approval gates, and whether a first job is a fit.

Put a number on it

Conduit's stack is one example. Price your own with the software cost calculator. It starts from a real firm's stack, tags each line Stays or Consolidates, and nets the $500 a month FirmOps maintenance; what that covers is on the pricing page.