Forty-plus agent fields
System prompt, welcome line, language, voice, model, temperature, interruption behaviour, retries and end-of-call rules — all per agent.
Platform
Everything a calling programme needs after the call connects.
Veytrix AI is a phone-calling platform, not a chatbot with a number bolted on. Every capability below exists because a team running real campaigns asked for it.
- 40+ agent fields
- 3 call types
- English · Hindi · Telugu
- 82 API endpoints
Build
Configuring the agent
Two agents in the same workspace can behave nothing alike — the behaviour is configuration, not a fork.
Swap STT, LLM and TTS
Three independent choices, not a bundle. If one vendor degrades you change a dropdown, not a product.
Knowledge base
Give an agent your fee structure or policy document and it answers from that text instead of improvising a number.
Response latency controls
Silence windows and echo protection per agent, so a thoughtful caller is not cut off and a brisk one is not left waiting.
Multilingual, Indian-first
English, Hindi and Telugu, with the agent following whichever language the caller actually uses.
Prompt templates
Start from a working prompt for your industry rather than a blank box, then edit it down.
Run
Getting the calls out
Pacing, windows and retries are the platform's job — not something to reimplement in a spreadsheet.
Campaigns and batches
Upload a CSV, map columns to prompt variables, pick an agent and a calling window. Dialling paces itself against your concurrency limit.
Inbound on the same agent
Point a number at an agent and it answers as well as places calls — same prompt, same knowledge base, same rating template.
Retries and call-backs
“Call me at four” is captured and the call is placed. No-answers retry on the policy you set rather than hammering a number.
Bring your own telephony
Rent Indian +91 numbers from us or connect the Plivo account you already have, keeping your caller ID and number reputation.
Windows and DNC
Calling hours and opt-out handling are enforced by the platform, not left to whoever built the campaign.
Calendar booking
A confirmed slot goes into Google Calendar, with the invite sent while the caller is still on the line.
After the call
Where a calling programme actually pays back
The call is the visible part. What arrives in the sixty seconds after hang-up is what changes anyone's day.
Transcripts and summaries
Speaker-separated, on every call. Reading fifty is the most useful hour of your first month — how to read them.
Ratings against your rubric
You define the dimensions; every finished call is scored against that same rubric with a written justification. Call four hundred is judged like call one.
Structured extraction
Name the fields — budget, notice period, pincode, preferred slot — and free-form conversation comes back as columns you can sort.
Outcomes into your system
Webhooks push the result where the work happens. An outcome nobody wrote down is a call you paid for twice — the detail.
Analytics and dashboards
Per-agent stats, screening funnel, cost tracking, engagement buckets and rating distribution — live in the dashboard, not a monthly export.
Call recordings
Capture full call audio to your own S3 bucket, then play, download or share it from the Recordings page.
Govern
Once it is more than one team
Who can change an agent, who can see a transcript, and what happened to the numbers last week.
Multi-org console
A super-admin platform view across every organization: usage, agents, calls and revenue — with per-org limits and plans.
RBAC and audit log
Four enforced roles and a full audit trail of every action — access changes take effect in seconds, no re-login required.
Connect
Fits the stack you already run
REST API and webhooks
Place calls, create agents and read results programmatically. Developer surface.
Google Sheets
Export to the sheet the hiring manager already opens, instead of asking them to learn a new dashboard.
WhatsApp follow-ups
Template campaigns from the same workspace and the same wallet as the calls.