A
AdyaVoice Intelligence
AdyaVoice intelligence platform

Adya your AI calling infrastructure - RAG+Scripted

Adya is the internal operator console for outbound voice AI — script authoring, test dials, webinar and uGNET campaigns, and call analytics in one stack.

Explore console

Call modes

2

RAG · scripted

Campaigns

2

webinar · uGNET

Languages

3

en · hi · hinglish

STT latency

<300ms

Deepgram

Call modes

RAG and scripted — two engines, one platform

RAG

Knowledge-grounded calls

Upload PDFs to Qdrant, then let the LLM answer from your documents during live voice sessions. Text and voice playground available before go-live.

Scripted

Compiled conversation flows

Author marker scripts, compile to JSON graphs, test on one number, then scale via webinar or uGNET campaigns — deterministic, multi-language, no RAG retrieval.

Operator console

Everything in the sidebar

Script library

Create, edit, and compile scripts from marker text or PDF. en-IN, hi-IN, and Hinglish agents.

Open

Test call

Dial one number to validate script wording, language, and flow before any campaign.

Open

Campaigns

Webinar reminders and uGNET slot confirmations — CSV upload, concurrency, retries, pause/resume.

Open

Call logs

Filter by script, campaign, lead tag, and date. Open View transcript for the latest call on each contact.

Open

Dashboard

Call volume, campaign progress, lead tags, and AI cost across scripted operations.

Open

Platform guide

How to create scripts, campaigns, and review calls

Adya follows a simple operator workflow: author a script, validate with a test dial, launch a batch campaign from CSV, then review transcripts and lead tags. Log in to the dashboard to run each step.

Author a script

Open

Build deterministic conversation flows before any dial goes out.

  1. 1Open Script library and create a new script or select an existing one.
  2. 2Write marker text (Q:/OPT blocks) in the raw editor, or paste compiled graph JSON.
  3. 3Pick agent language — English (en-IN), Hindi (hi-IN), or Hinglish.
  4. 4Compile from markers, review the graph, save, and set status to active when ready.

Place a test call

Open

Validate wording, branching, and language on one number before scaling.

  1. 1Choose the script you want to test from the dropdown.
  2. 2Enter the destination number and optional contact name.
  3. 3Click Place test call — Adya dials via Plivo and runs the scripted graph live.
  4. 4Listen through the flow; fix the script and re-test until it sounds right.

Create a campaign

Open

Batch outbound dials from a CSV contact list with retries and concurrency.

  1. 1Go to Webinar campaigns or uGNET campaigns in the sidebar (same flow, different list type).
  2. 2Enter a campaign name, select the compiled script, and upload a CSV with to_number and name columns.
  3. 3Set parallel dials (1–20), max retries per contact (1–50), and success criteria.
  4. 4Click Create campaign (draft) — contacts are imported; nothing dials until you press Start.

Run and manage campaigns

Open

Start, pause, edit, and export while the worker dials your contact queue.

  1. 1Start launches outbound dials; Pause stops new attempts while keeping progress.
  2. 2Expand a campaign to see per-contact status — pending, connected, completed, or retrying.
  3. 3Edit replaces the contact list or updates concurrency, retries, and success criteria.
  4. 4Export CSV downloads contact outcomes; Call logs opens filtered transcripts for that campaign.

Review logs and transcripts

Open

Analyze outcomes, lead tags, and full call transcripts after dials finish.

  1. 1Open Call logs and filter by script, campaign, lead classification, connection, or date range.
  2. 2Each row groups calls by contact — click View transcript for the most recent conversation.
  3. 3The transcript page shows turns, intent, script exit, lead tag, and re-analyze options.
  4. 4Filter call logs by campaign to review outcomes for a specific batch.

RAG knowledge calls

Open

Optional path — ground live voice answers in uploaded documents instead of a fixed script.

  1. 1Upload PDFs in Knowledge base and ingest chunks into Qdrant.
  2. 2Preview stored chunks and collections before going live.
  3. 3Use the text or voice playground to test retrieval-backed replies.
  4. 4RAG mode runs alongside scripted flows — pick the engine that fits each use case.

Campaign types

Webinar vs uGNET

Webinar

Webinar reminder campaigns

Upload registrant lists, run reminder calls before the session, and track who connected or finished the script.

uGNET

uGNET slot confirmations

Upload uGNET contact lists for outbound follow-ups and slot confirmation calls with the same retry engine.

CSV format

Contact list requirements

campaigns.csv

# Required header row

to_number, name

919876543210, Adarsh

919123456789, Learner

E.164 or local numbers · name used in scripted greeting

Success criteria

When a contact stops retrying

Reminder (connected = success)

Success when the call is answered and the agent speaks. Retries only on no-answer, busy, rejected, or dial failure.

Script complete

Success when the conversation reaches a terminal script end. Retries until the script completes or max retries is hit.

Campaign lifecycle: create as draft → Start to dial → Pause to halt new attempts → Edit to swap contacts or tuning → open Call logs or View transcript to review outcomes.

Scripted workflow

From script to campaign

01

Author script

Script library · compile from markers or PDF

02

Test call

Single dial to verify flow and language

03

Launch campaign

Webinar or uGNET batch with retries

04

Review logs

Intent, lead tags, transcripts, re-analyze

Voice stack

Integrated pipeline

STT·DeepgramRAG·QdrantLLM·CerebrasTTS·SarvamTel·Plivo
scripted call trace

[STT] Deepgram → caller response captured

[Script] Graph node W0_P1 → branch YES_NO

[LLM] Cerebras → scripted reply streamed

[TTS] Sarvam → Plivo outbound channel

lead tag · script exit · campaign contact status

Dashboard tracks call volume, campaign contact progress, lead tags, and AI cost across all scripted operations.

Deploy

Up and running in four steps

Configure Plivo and API keys, author your first script, then validate with a test call before launching a campaign.

01

Set environment

FRONTEND_URL · API keys
02

Connect Plivo

from_phone · credentials
03

Author a script

Script library
04

Place a test call

Test call console
A
AdyaVoice Intelligence

Operator console access

Adya is invite-only. Sign in to author scripts, place test calls, run webinar and uGNET campaigns, and review call logs with intent and lead tagging.

Log in with credentials from your administrator. No public signup.

Script library