Private voice engine
Draft replies in your personal style
PersonaLearn reads the messages you already sent and builds a private voice profile. Drafts come back matching your phrasing, your sentence length and how you write numbers.
100
Messages to calibrate
1
Private model per user
0
Messages sent on its own
24h
Time to hard delete
How it works
How your private model trains
The system reviews your outbound messages and maps your writing habits. When an inbound message arrives it writes a candidate reply for you to read, change or send.
01
Scan outbound messages
The engine reads your last hundred sent messages across connected networks and catalogues your sentence structure.
02
Build the private profile
A voice model for you alone compiles your punctuation habits, greeting style and numeric phrasing.
03
Generate the draft
When an inbound message arrives, PersonaLearn writes a candidate reply into the composer box.
04
Review and send
You read the text, change words if you want, and press send. The model never posts to a client on its own.
Scope
Data points the engine evaluates
PersonaLearn works from your outbound messages alone. It studies how you phrase things and ignores messages written by other people on your team.
What it reads
- Your historical outbound messages
- Your punctuation and capitalisation choices
- Your numeric habits, like 50k against 50,000
- Your typical greetings and sign-offs
What it never reads
- Outbound messages written by your colleagues
- Internal team notes and private comments
- Inbound messages from external trading partners
- Attached PDF contracts and bank statements
Workspace boundaries
Model weights stay strictly isolated
Your voice model belongs to your user profile. An admin in your company can confirm the tool is active. They cannot download the model or look inside your voice profile.
User-level separation
Each operator holds an isolated model file that cannot reach another member of staff's profile.
Admin boundary
A workspace admin can turn the feature on or off. They cannot read your style parameters.
Tenant training isolation
Your message patterns are never pooled to build a model for another company or an outside platform.
Profile purge
Deleting the profile wipes every associated style weight from our storage cluster inside 24 hours.
Your messages never train a base model for us or for any provider we call, and they are never pooled with another customer's data. On Enterprise the model runs inside your own deployment with your own keys.
Security detailInspection
Inspect your measured style
Open your profile settings and read the parameters PersonaLearn pulled out of your sent messages. Average sentence length, greeting habit, number format and emoji use are all on one page.
Everything below came from one Antwerp trader's own outbound history.
Voice profile • ben@
visible to you onlycalibrated on 100 sent messages • recompiled weekly
The off switch
One switch to erase the profile
You can disable PersonaLearn from your own settings at any moment. The composer stops producing drafts straight away and the background job deletes your voice model within 24 hours.
Drafting stops
The composer stops producing drafts the second you flip the setting off.
Weights deleted
Your style parameters are scrubbed from the database inside 24 hours.
Clean restart
Turning it back on calibrates a fresh model from scratch.
PersonaLearn drafting
off • model queued for deletion, 23h 41m remaining
Draft comparison
Same deal terms, three different voices
A buyer asks whether the shipment lands on Thursday and wants the bill. Here is the same answer, drafted for three people who write nothing alike.
Antwerp diamond trader
Blunt, no greeting, digits, short sign-off
Arrives thursday 14:00 at the vault. Invoice for 42k sent to your mail now. - B
draft • not sent
Retail account manager
Warm, full sentences, first name, friendly sign-off
Hi Sarah, yes, the shipment will reach you Thursday afternoon. I will email the invoice right away. Best, David
draft • not sent
Telecom operations lead
Formal, precise syntax, references the ticket number
Regarding ticket 8492, delivery is confirmed for Thursday afternoon. The corresponding invoice has been queued.
draft • not sent
PersonaLearn never sends a message without a person pressing send. If you want to see what it would have written across your last five hundred conversations, run shadow mode first.
Shadow modeMore of the platform
Unified inbox
Nineteen networks in one shared team timeline.
AI layer
Eight instruments, some on arrival and some on a schedule.
Total recall
Permanent retention and search across your connected channels.
Export and migration
Exports up to your plan ceiling and one-pass migration from 12 systems.
Apps
Native clients for macOS, Windows, iOS and Android.
Integrations
Sync for CRMs, billing systems and ticketing.
Draft faster without losing your voice
Turn PersonaLearn on for your seat. Calibration takes one hundred sent messages and you keep control of every outbound reply.