Make routine work actually happen
Intake, matter opening, engagement steps, file setup, follow-ups, and reminders should run consistently instead of sinking to the bottom of the list.
7/12 forms · 6/6 usable call summaries · 9/9 debriefsFive-minute private beta video
Show the operation,
not a feature tour.
The audience is mostly solo and very small firms. They already know AI can draft. What they need to see is that Referent can keep intake, email, matters, documents, tasks, and deadlines moving while the lawyer stays in control.
One story
Scattered work becomes one matter. AI prepares. The lawyer approves.
01 / Audience
Who is on the other side of the camera
The strongest shared trait is not a practice area. It is an owner-operator carrying legal work and firm operations at the same time.
438People recordsAll CRM leads live in People
89%Solo or up to 10 peopleAmong 281 leads with usable firm size
73%United StatesAmong 181 with country
476Linked notesCovering 425 current People
Shape of the known lead pool
Three useful mental models
The overloaded solo
Knows every matter personally, but intake, follow-up, billing prep, and filing keep slipping.
The lean owner
Wants more capacity without hiring an admin layer. Usually decides and pays directly.
The AI-literate skeptic
Has tried generic AI. Wants reliable execution, source context, boundaries, and a human to call.
02 / Demand
What they care about
The richer interviews repeat the same jobs. The wording changes; the operational shape barely does.
Put the practice in one context
Email, documents, notes, tasks, calendar, and current matter status should stop living across disconnected tabs and personal memory.
5/6 call summaries · 6/9 debriefsSurface what was forgotten
Lawyers ask for overdue work, deadlines, missing follow-ups, and a quick answer to “what happened last?”
6/6 call summaries · 8/9 debriefsKeep authority with the lawyer
Matter-specific context, editable drafts, privacy controls, source-linked summaries, audit history, and approval before client-facing action are trust requirements, not polish.
11 forms with trust answers · editability led at 7/11Prove fit before asking for rollout
They ask for a real demo, supported integrations, clear beta scope, price-to-value, and a believable first result—not another subscription that adds manual sync.
Demo/scope: 3/6 call summaries · 3/9 debriefsClose the remaining capture gaps
Phone calls, in-person meetings, billing systems, complex client relationships, and custom workflows are expansion asks. They are confirmed, but not a single universal request.
Medium confidence · several are not complete today
03 / Product
Connect the ask to what is already real
The video should stay on the green path. Mention the gaps only if asked; do not sell roadmap as product.
Lead asks
“Set up the client and matter without retyping everything.”
“Tell me what needs attention today.”
“Turn incoming work into the right matter or task.”
“Give me the whole matter without hunting.”
“Let AI help, but show me why and ask before acting.”
Safe to demonstrate now
Connected-source onboarding and client / matter import
Dashboard: overdue, due today, inbox, triage, schedule, activity
Triage suggestions with Create / Dismiss and destination review
Matter workspace plus State of Play, timeline, notes, email, docs, tasks
Voice/chat with matter context, source accounting, approval, activity trail
Verified against the current checked-out application code, including UI plus API/service or registered agent-tool paths. Rehearse every step in the target demo workspace before recording; code presence is not proof that each integration is enabled there. Phone-to-matter capture, in-person meeting capture, universal PMS migration, Outlook Calendar event sync, and every billing integration remain outside this demo promise.
04 / Run of show
One clean five-minute arc
Speak for roughly 4:05. Use the remaining 55 seconds for clicks, loading, pauses, and product response time. Do not narrate every control.
0:00
The operational problem
Another chatbot is not the answer. Small firms lose time to scattered context and work that only exists in someone’s head.
On screenFace to camera, then Referent home.
0:35
Start from the real practice
Connect email, calendar, and documents. Show the import building client files, matters, tasks/docs, and current context.
On screenImport progress or a short cutaway.
1:10
What needs attention today
Dashboard: overdue, due today, unlinked email, triage queue, schedule, time, recent activity.
On screenOne slow scan; no detours.
1:45
AI prepares, lawyer decides
Inspect one triage suggestion. Create the matter or task, or dismiss it. Make the approval model visible.
On screenSuggestion → source → approve.
2:35
One matter, one context
Open a seeded matter with email, documents, tasks, calendar, notes, billing context, timeline, and State of Play.
On screenOverview, then timeline/records.
3:25
Ask, then act
Speak: “What happened last in this matter, and what needs my attention next?” Keep the answer grounded in the visible matter.
On screenVoice input with matter pill.
4:10
Trust is inspectable
Show source accounting or activity. State the operating rule once: AI prepares; the lawyer approves.
On screenSources / activity / review state.
4:40
Private beta
Hands-on onboarding for small firms that want operational leverage without giving up professional judgment.
On screenreferent.law / Apply for Private Beta.
05 / Script
English teleprompter copy
527 spoken words: about 4:03 at 130 words per minute, leaving almost a minute for clicks, pauses, and product response time.
Most small law firms do not have a shortage of software. They have a shortage of time. Client emails arrive in one place, documents live somewhere else, deadlines sit in a calendar, and the real status of a matter is often in the lawyer’s head. A chatbot can help draft a paragraph, but it does not keep the practice moving. Referent is built for that operational layer.
Referent connects to the tools the firm already uses for email, calendar, and documents. During onboarding, it reads the existing practice and builds a working client file: clients, matters, tasks, documents, and a current summary. You are not starting with a blank database, and you are not spending a weekend rebuilding your firm by hand. The goal is to reach useful context quickly, with a human helping you through setup.
This is the dashboard. It gives me the operational picture for today: overdue work, tasks due today and this week, unlinked email, items waiting in triage, today’s schedule, tracked time, and recent activity. For a solo lawyer or a small team, this is the difference between searching through tabs and knowing where attention is needed right now.
Here is the triage queue. Referent can identify an email or document that needs filing, suggest a new matter, or turn an obligation into a task. But it does not quietly make a high-risk decision for the firm. I can inspect the source, choose the right destination, create the matter or task, or dismiss the suggestion. The AI does the legwork; the lawyer keeps authority. That approval model matters because client communication and legal judgment should never disappear into an autonomous black box.
Now I will open a matter. The email history, documents, tasks, calendar events, notes, billing context, and recent activity sit together. Referent also maintains a current State of Play and timeline, so I can understand what changed without reconstructing the matter from memory. This is the part many lawyers ask for: not another generic CRM, but a focused client and matter record that reflects how legal work actually moves.
I can type or speak a request in the context of this matter. For example: “What happened last in this matter, and what needs my attention next?” Referent answers from the firm’s live context rather than a blank prompt. I can also use it to create a client or matter, summarize recent work, and move routine follow-up forward. The point is not conversation for its own sake. The useful output is a cleaner matter, a prepared action, or a task that no longer lives only in someone’s head.
When Referent produces a summary, I can inspect the source accounting and activity history behind it. I can see what was used, what was excluded, what changed, and what still needs review. Client-facing or high-risk actions stay behind lawyer approval. The rule is simple: AI prepares; the lawyer approves.
Referent is for small firms that want AI to reduce operational work without handing over professional judgment. We are opening a private beta with hands-on onboarding. If you want your email, documents, tasks, and matters to work as one system, apply at referent.law.
06 / Shot list
Five visual beats worth rehearsing
Use the real application for the recording. These existing brand images are composition references, not substitutes for a working demo.
07 / Guardrails
What makes the recording believable
Credibility comes from a narrow working path and clear boundaries, not from saying “AI” more often.
Do
- Record only in a seeded demo workspace.
- Preload one matter with email, document, task, event, timeline, and one reviewable suggestion.
- Test the exact voice prompt before recording.
- Use one approval and one matter thread from start to finish.
- If import timing varies, use a short cutaway of the completed import.
- End on “Apply for Private Beta.”
Do not promise
- Phone/VXT-style automatic call capture.
- Granola-style in-person meeting capture.
- Turnkey migration from every legacy PMS.
- Outlook Calendar event sync or every accounting/billing integration.
- Autonomous legal judgment or unreviewed outbound action.
- Production readiness of any roadmap-only workflow.