Autessa for Contact Centers

The whole contact center on one AI platform.Or AI added to the one you have.

Run phones, IVR, AI voice agents, texting and ticketing on Autessa, or plug Autessa into the platform you already run through APIs. Either way, card numbers are redacted before a recording is stored, callers are checked for cloned voices and spoofed numbers, and every transcript is graded and searchable.

Start building
  • Call redaction
  • Voice security
  • Automatic grading
  • Semantic search
  • Call data lake

Two ways in

Build the contact center on Autessa, or add Autessa to the one you run.

Start fresh with every piece of the floor in one platform, or keep your current phone system and call Autessa for the parts that make it safer and smarter. Both paths share the same redaction, security and analysis underneath, so you can move from one to the other at your own pace.

PATH A · BUILD

Your whole contact center on Autessa

Numbers, flows, AI agents, texting, ticketing and customer context, working together out of the box.

  • Phone numbers

    US business numbers in one click or by area code, the numbers in your own Twilio account, or texting on the number customers already know while calls stay with your carrier.

  • Call flows and IVR

    A visual call flow editor with greetings, business hours in any time zone, nestable keypad menus, ring groups with whisper and press 1, wait audio and transcribed voicemail.

  • AI voice agents

    Agents that answer inbound calls and place outbound ones with retries, calling hours and answering machine detection, then hand off with cold or warm transfers. In English and Spanish.

  • AI Receptionist

    Answers every call and text on your main line, takes a message with the reason, screens before it transfers, blocks spam and alerts the right person when a call is urgent.

  • Texting with consent and 10DLC

    A shared texting inbox, AI text replies, STOP, START and HELP handled for you, A2P 10DLC registration with pre-submit checks, a consent registry and a hosted consent site.

  • Support Desk ticketing

    Email-to-ticket, AI triage, SLAs, AI-drafted replies and a knowledge base, with a branded customer portal where customers follow their requests and an AI help assistant answers.

  • CRM context

    Contacts, companies, deals and email history on the same platform as your calls, tickets and agents, so every conversation starts with who the customer is.

PATH B · ADD ON

Autessa added to the contact center you have

APIs your current platform calls, so the recordings, callers and transcripts you already have get the same treatment.

  • SecurePass call recording redaction

    Post one recording and get back redacted audio and a redacted transcript, or point the command-line client at an S3 prefix and clean an entire archive. Results land back in your bucket.

    POST /clients/call-redaction/redact
  • AI Application Firewall API

    Score recordings and voicemails for cloned voices and the watermark Autessa agent voices carry, run caller verification challenges, and inspect every turn of your own AI agents.

    POST /clients/aaf/voice/score
  • Evaluations grading API

    Grade transcripts from any agent, wherever it runs: one call now, or batch jobs in the background with progress, pass rate and signed completion webhooks.

    /clients/evaluations
  • AutessaDB data lake

    Load call records from CSV and Parquet exports, append or replace, and query millions to billions of rows with analytical SQL, plain-language Ask and data agents.

    /clients/autessadb/analytics

Redaction · SecurePass

Card numbers, CVVs and SSNs never reach your recordings.

SecurePass transcribes each call with word timings, finds sensitive details in the words and in keypad tones, maps them to exact times in the audio and silences them. The transcript marks every span with its start and end time. Only the clean copy is kept.

  • Built for PCI-sensitive data. Card numbers, expiry dates and security codes, plus Social Security numbers, bank and account numbers, dates of birth, phone numbers, emails and addresses.
  • Real calls, not clean examples. A card read in four groups between the agent's “uh huh”s, a corrected digit, or digits typed as touch tones are all found and silenced.
  • English and Spanish. Measured on English and Spanish call recordings, including a PCI-specific benchmark.
  • On every Autessa number. Turn on Remove personal info before saving, and every recording, transcript and voicemail is cleaned before it is stored. Choose silence or a tone.
  • On the archive you already have. Point SecurePass at your S3 bucket with a role you grant, and redacted audio and transcripts are written back to it.

Security for voice · AI Agent Firewall

Know the caller is real before anything sensitive happens.

Cloned voices and spoofed numbers fool people and AI alike. The AI Agent Firewall checks every turn of every call your agents take, and makes the caller prove who they are before a sensitive tool runs. Start in Monitor to see what it would catch on your real traffic, then switch to Enforce.

Voice deepfake detection

Everything a caller says gets a 0 to 1 synthetic-voice score, labeled Real, Synthetic or Unsure, with a stricter threshold while they verify. Autessa agent voices carry an inaudible watermark, so replayed agent audio is caught.

Caller ID check (STIR/SHAKEN)

The carrier’s attestation is read on every incoming call, Verified, Partly verified, Not checked or Failed, and each result gets its own action: allow, flag or ask to verify.

Word of the day and duress word

Staff calling in say today’s word inside random words, so a recording cannot pass. A duress word silently disables sensitive tools and alerts admins and the contacts you name.

Verify before use

Put verification on any sensitive tool. On calls, authenticator codes are typed on the keypad and never heard by the AI, with lockouts scoped to the conversation so attackers cannot lock staff out.

Auto-analysis of every call

Every call graded. Every word searchable. Every trend in SQL.

Quality assurance stops being a sample. Every transcript is scored automatically, agents practice what the scores show, and every call lands in tables you can search by meaning and query at any scale.

Automatic grading · Evaluations

Grade every transcript, not a handful a week.

Each conversation is scored for groundedness, task completeness, argument faithfulness and efficiency, and an expected-outcome judge returns Met, Partially met or Not met with a cited justification.

  • AutoEval on every production conversation of your Autessa agents, voice included, or on a sample
  • Evaluation notes become your own grading rubric for the judge
  • Any transcript from any platform through the grading API, one call or a batch job
  • A human verdict always takes precedence, with human vs automated comparison

Coaching · Roleplay

Turn the grades into practice.

Coaches assign realistic scenarios and agents practice live voice calls with AI characters that interrupt, object and push back. Every session is scored against your rubric, with quotes from the transcript.

  • Rubrics with weights, a pass mark, rating anchors and knowledge checks against your reference text
  • Session reports with strengths, growth areas and the next thing to practice
  • Delivery measures: pace, filler words, talk time and hedging
  • Assignments with due dates and readiness by team, plus a starter customer support pack

Semantic search · AutessaDB

Find any call by what was meant, not what was typed.

Every call on an Autessa number is logged in the Phone calls table with its transcript and recording. AutessaDB indexes text, file and audio columns for semantic search on every write.

  • Search transcripts and voicemail transcripts by meaning across every call
  • Audio columns get transcripts automatically, so uploaded recordings are searchable too
  • AI enrichment adds summaries, key facts and likely questions to indexed values
  • Agents search the same tables to answer questions about your calls

Call data lake · AutessaDB

Call records at contact-center scale.

A columnar data lake for millions to billions of rows, on the same platform as your tables and agents. Load call records from CSV, Parquet or existing AutessaDB tables and ask anything.

  • Read-only ANSI SQL with joins, CTEs and window functions in the SQL Workbench
  • Ask in plain language and review the proposed SQL before it runs
  • Data agents that answer data questions, and dashboards in custom apps
  • Masked columns still group, count and join, with row and column security on every query

Agent assist and automation

Help during every call, and the follow-up done after it.

The people and AI agents on your floor get the context, safeguards and write-ups they need, and the work after the call happens on its own.

Screened transfers with a whisper

The AI Receptionist puts a call through only with a real reason and a matching teammate. The teammate hears who is calling and why, then presses 1 to take it.

AI Receptionist →

Urgent alerts and a morning summary

Calls that need someone today alert the right people by email, and text if you choose. Each weekday morning, one email lists the important calls first.

AI Receptionist →

Secure keypad capture

Callers type card, bank and SSN digits on the keypad out of the AI’s hearing, and recording pauses and resumes around sensitive details.

Phone & Texting →

Live-call tools for agents

Voice agents transfer, take voicemail, press keys, read caller info and ask for keypad input mid-call, and speak short filler phrases in their own voice while tools run.

Phone & Texting →

Ticket copilot

Every Support Desk ticket gets a summary, sentiment, a suggested reply with sources, similar past tickets and next steps, plus AI reply drafts reviewed before they go out.

Support Desk →

Call summaries

Upload a recorded call to Meeting Recorder for a transcript and an AI summary with action items by person, or add a trigger on the Phone calls table so an agent writes up every call as it ends.

Meeting Recorder →

Response checks on every AI reply

Replies are reviewed before they are sent, in every channel: facts must come from the conversation, sensitive data is caught, and a check that cannot run fails closed.

Evaluations →

Step-by-step guides for agents

Do a task once in the Autodocs browser and get a how-to guide with a highlighted screenshot for every step, refreshed automatically when the workflow runs again.

Autodocs →

Grow and innovate

A contact center that changes as fast as you do.

Every Autessa app is 100% customizable, code and all. Ask Autobot, the AI builder, for a change in plain language, or have your developers edit the code directly. New queues, new languages, new agents and new reports ship when you need them, not when a vendor roadmap allows.

  1. New call flows

    Branch by business hours, keypad choice or closed dates in the visual call flow editor.

  2. New AI agents, tested before launch

    Run test sets with simulated callers, from an impatient customer to a frustrated one who wants a human, against a draft before any customer hears it.

  3. New training

    Write a Roleplay scenario for the call your team keeps getting wrong, with its own character, objections and rubric.

  4. New dashboards

    Put the same read-only data lake queries behind a dashboard in a custom app for supervisors and leadership.

  5. New automations

    Add triggers to the Phone calls table that run tools and agents when a call starts, ends or is redacted.

FAQ

Contact center AI on Autessa, answered.

What is Autessa for Contact Centers?

Autessa for Contact Centers is one AI platform for running a contact center or adding AI to the one you already have. You can run phone numbers, call flows and IVR, AI voice agents, the AI Receptionist, texting and Support Desk ticketing on Autessa, or call Autessa APIs from your current platform. Either way you get call redaction, voice security from the AI Agent Firewall, automatic grading of transcripts with Evaluations, coaching with Roleplay, and semantic search and analytics over every call in AutessaDB.

Can I add Autessa to my existing contact center platform without switching?

Yes. Each capability is available as an API. SecurePass redacts recordings you send it, one call at a time or a whole S3 bucket at once. The AI Application Firewall API scores recordings for cloned voices and runs caller verification challenges. The Evaluations grading API grades transcripts from any agent, one at a time or in batch jobs with signed webhooks. And the AutessaDB data lake loads call records from CSV and Parquet exports so you can query them with SQL or ask in plain language.

How does Autessa remove card numbers from call recordings?

SecurePass transcribes each call with word timings, finds sensitive details in the words and in keypad tones, maps them to exact times in the audio, and silences them. The redacted transcript marks each span with its start and end time. It handles card numbers read in groups between an agent’s acknowledgments, corrected digits and numbers typed on the keypad, and it removes card numbers, expiry dates, security codes, Social Security numbers, bank and account numbers, dates of birth and contact details. On Autessa phone numbers, turn on Remove personal info before saving and only the cleaned recording and transcript are ever stored.

Can Autessa detect deepfake or cloned voices on contact center calls?

Yes. The AI Agent Firewall gives everything a caller says a voice score from 0 to 1 for signs of a synthetic or cloned voice, labeled Real, Synthetic or Unsure, and applies a stricter threshold while a caller is verifying. It also reads the carrier’s STIR/SHAKEN caller ID result on every incoming call, can require verification before a sensitive tool runs, takes authenticator codes on the keypad so the AI never hears them, and supports a word of the day and a duress word for staff. Recordings from other systems can be scored through the API.

Can Autessa automatically grade every call transcript?

Yes. Evaluations scores conversations for groundedness, task completeness, argument faithfulness and efficiency, and an expected-outcome judge returns Met, Partially met or Not met with a cited justification. AutoEval can run on every production conversation of an Autessa agent, voice included, or on a sample. For calls handled anywhere else, send the transcript to the Evaluations grading API, one at a time or as a batch job, and add evaluation notes as your own grading rubric.

How do I search across all of our call transcripts?

Every call on an Autessa number is logged in the Phone calls table in AutessaDB, with its transcript and recording. AutessaDB has built-in semantic search on text, file and audio columns, indexed automatically on every write, so you can find calls by meaning, such as callers upset about a double charge, rather than exact words. Audio columns get transcripts automatically, and agents can search the same tables to answer questions about your calls.

Does Autessa support Spanish-language contact center calls?

Yes. AI voice agents speak English and Spanish, with a voice, greeting and extra instructions per language and built-in filler phrases in both. SecurePass call redaction is measured on English and Spanish call recordings, including a PCI-specific benchmark.

Can we customize the contact center apps to how we work?

Yes. Every Autessa app is 100% customizable, code and all. Ask Autobot, the AI builder, for a change in plain language, or have your developers edit the code directly. Build new call flows in the visual editor, test new AI agents against simulated callers before launch, and put dashboards over your call data in custom apps.

Build your contact center on Autessa, or add Autessa to the one you have.

See redaction, voice security and automatic grading on calls like yours, then grow from there.

Security and compliance documentation lives at trust.autessa.com.

Start building