Scroll
v-reps
V-Reps · Virtual Representatives

The product is
the configuration.

A V-Rep is a Voice, Email, or Chat representative you build from tabs, sliders, and toggles. No engineering team. No glue code. Nine, ten, or six tabs to live.

Build your first V-Rep See pricing
Voice · 9 tabs●Email · 10 tabs●Chat · 6 tabs●One console
How a V-Rep comes together

Pick a channel. Fill the tabs. Ship it.

STEP 01

Pick a channel.

Voice answers the phone. Email reads and replies in Gmail. Chat embeds as a widget. Same console, same knowledge base, same wallet.

voice · email · chat · all one tenant
STEP 02

Fill the tabs.

Each V-Rep is a tabbed config: identity, LLM, voice, tools, escalation, analytics. Sliders and toggles, not YAML. Nothing to deploy.

9 voice · 10 email · 6 chat
STEP 03

Ship it.

Save and the V-Rep is live: a phone number, a monitored inbox, or an embed snippet. Edit any tab and it updates in place.

live in minutes · no engineering ticket
Three configurations, one model

One product. Three V-Reps.

Same console, same knowledge base, same wallet. The channel decides how many tabs you fill, and what each one has.

Voice V-Rep
9tabs
Answers the phone. Picks up in < 1s.
Chat V-Rep
6tabs
Embeds as a widget. Hands off to voice.
Shared core
1console
Every V-Rep draws from the same substrate.
Every tab, every channel
Voice Email Chat Shared core
Config tabs 9 10 6 1 console
Channel I/O Phone · Telnyx / Vonage / SIP Gmail inbox Web widget n/a
LLM GPT-5.1 · OpenRouter GPT-5.1 · OpenRouter GPT-5.1 · OpenRouter swappable per V-Rep
Speech / parsing Deepgram nova-3 · 5 TTS Thread + attachments n/a n/a
Escalation rules Keyword + action Keyword + action Keyword + action one rule engine
Knowledge base Shared Shared Shared one KB
Memory Per caller Per thread Per session n/a
Tools / MCP Yes Yes Yes shared registry
Analytics Calls Replies Sessions one layer
Widget / embed Click-to-Talk n/a Embed snippet n/a
Billing wallet wallet wallet one ledger
Swipe
Configuration, not code

Watch Jeeves get configured.

Voice V-Rep · Jeeves · Visium IR
Identitytab
Name it, give it a workspace, set its status
Tab
1/ 9
Name
Jeeves
Workspace
Visium IR
Channel
Voice
Status
Live
Tab 1 / 9 · V-Rep configured. Zero lines of code written so far.
Scroll through the build
The shared core

Three V-Reps. One substrate.

Workspace · Visium IR
3 V-Reps
Jeeves · Penelope · Iris · all one tenant
All live
1 tenant · 4 role tiers · invite by email
Shared resourceUsed by
KBKnowledge baseOne corpus, indexed onceAll 3Live
$Wallet + ledgerOne blended rate, one balanceAll 3Live
TEAMTeam + rolesOwner / Admin / Member / ViewerAll 3Live
ANLAnalytics layerCalls, replies, sessions in one viewAll 3Live
ESCEscalation engineKeyword + action compositionAll 3Live
PRNPronunciation libCSV / JSON · tickers, drug namesVoiceLive
MCPMCP registryServers + custom functionsAll 3Live
Frequently asked

Questions. Answered.

Do I need an engineer to build a V-Rep?+
No. Every V-Rep is built from its tabs: identity, model, voice, tools, escalation, analytics. Sliders and toggles, not YAML. There is no deploy step and no engineering ticket. The person who owns the use case owns the V-Rep.
Can one V-Rep do voice and email at the same time?+
A single V-Rep is one channel by design (Voice or Email or Chat) so its configuration stays clean. But every V-Rep in a workspace shares the same knowledge base, wallet, team, and analytics, so a customer experience can span channels, including Chat handing off to a Voice V-Rep.
What actually differs between the three configurations?+
The channel I/O and the tab count. Voice adds Transcriber, Voice, and Call tabs (9 total). Email adds Email Connection, Monitoring, Tracking, and Logs (10 total). Chat is the lean 6-tab embed. The core, LLM, Tools, Analytics, Memory, and the knowledge base, is shared across all three.
Can I change a V-Rep after it is live?+
Yes. Edit any tab and it updates in place. No redeploy, no downtime, no version branch to merge. The same console you built it in is the console you operate it from.
Which models and voices can I pick?+
LLM runs through OpenRouter with GPT-5.1 as the default and is swappable per V-Rep. Text-to-speech is your choice of five providers: Cartesia, ElevenLabs, OpenAI, Rime, or Deepgram. Transcription is Deepgram nova-3. Every one of these is a per-V-Rep dropdown, not a contract negotiation.
How do MCP servers and custom functions work?+
Register an MCP server or define a custom function once in the Tools tab and it becomes available to any V-Rep in the workspace. One shared registry, no per-agent re-plumbing.
ConnexŪS Ai owl

Build your
first V-Rep.

Pick a channel. Fill the tabs. Ship it. Voice, Email, or Chat, live in minutes.

Build your first V-Rep See pricing