AI voice agent QA

Every AI call checked.
Every release proven.

Loops grades every call against your own SOPs and shows what each release broke.

Rule libraries for US collections, insurance and healthcare

Loops
Outcome · Collections · rubric v15 $14,300 recoverable, back in the queue 7 calls ended with no payment plan offered
Because3.2Offer a payment plan7 calls
Sample
Sample dataYour rubricEvidence retained 7y

Works with

  • Bland
  • Pipecat
  • Twilio logo
  • Ultravox
  • Amazon Connect
Every verdictquotes the transcript line it relied on
100% of callsgraded, where manual QA typically reviews under 5%
Every ruleapproved by your compliance lead before any call is graded

The gap

Your agent is live. Can you prove it's working?

Compliance

Is it following our rules?

Manual QA typically reviews under 5% of calls. McKinsey, 2024

Operations

Is the new version better or worse?

Tests pass. Live calls drift.

Finance

What is it earning us?

Platform reports count minutes and containment, not payments collected.

Up to $1,500

per call in TCPA damages when a court finds an AI-voice call without consent willful or knowing, and $500 a call when it doesn't. The FDCPA adds actual damages, up to $1,000 in statutory damages per lawsuit, and attorney's fees. Loops shows, call by call, whether your agent followed the rules a call can settle. Consent isn't one: it lives in your records, which Loops doesn't read. Sources: 47 U.S.C. § 227(b)(3) and 15 U.S.C. § 1692k Since February 2024, the FCC treats AI-generated voices as artificial voices under the TCPA, so AI calls need the same consent as prerecorded ones. FCC ruling 24-17 Gartner expects over 40% of agentic AI projects to be canceled by the end of 2027, over cost, unclear value or weak risk controls. Gartner, June 2025

How it works

One loop, from your rules to your results

Four steps, then the loop closes: fix a rule once, and the next release is proven against the last.

1Start from your rules

Your SOP is the rubric

Upload the SOP you already have. Loops splits it into numbered rules, and your compliance lead approves them before anything is graded.

  1. 1.2Right person§2.4
  2. 2.3Disclosure first§3.1
  3. 3.2Payment plan§4.2
  4. 4.1Log the promise§5.1

Approved by Priya K, compliance

2Check every call

Every call graded, with evidence

Loops grades each call, AI or human, rule by rule. Every verdict quotes the line it relied on, and required tool calls are checked.

Call 8f3a-22913 of 4 met

  1. 1.2Right personMet
  2. 2.3Disclosure firstMissed
  3. 3.2Payment planMet
  4. 4.1Log the promiseMet

00:18 · Agent · evidence for 2.3Your balance with Alder Card is $1,284.16, 94 days past due.

3Prove every release

See exactly what a release broke

Compare each rule's miss rate, this release against the last, and open the calls behind any number that moved.

Rule 2.3 · Disclosure firstv13 → v14

1.9%missed in v14+1.6 pts vs v13

2.3 got worse in v14190 calls

4See what it earned

The number finance asks for

Payments live in your own systems, and Loops doesn't read them today. Joined to graded calls, they show what each release earned or cost, as in this sample.

Revenue impactv14 vs v13

+$18,400more in payments posted, per 10,000 accounts

  1. Promise-to-pay rate12.8% → 13.7%
  2. Cost per resolved call$4.10 → $3.62
95% range +$11,200+$25,600

Fix the rule once.

Your team writes the fix and ships it, and Loops checks the next release against the last, rule by rule. Loops never writes to your agent.

Start a free 30-day audit

Release v15Ready for review

v15: 2.3 fixed, v14's gains kept

  1. 2.3Disclosure firstMissedMet
  2. 1.2Right personKept
  3. 3.2Payment planKept
  4. 4.1Log the promiseKept
SOP update · Collections2.3Line 18Give the mini-Miranda before any balance is stated.

Approved by Priya K, compliance

Watch

Loops in 30 seconds

Two short explainers, captioned. Both use sample data, not customer results.

01

How Loops works

0:33CC

From your operating procedures to a proven release, in one loop.

Chapters5

02

The release that passed every test

0:34CC

An illustrative story: what the tests missed, and what the release earned.

Chapters4

Independent by design

The builder builds. Loops keeps score.

Your platform's QA scores calls on that platform, on criteria you write in its format. Loops grades every call against your approved SOP rules, the same way on any platform and on your human floor.

You

Plan

Write and approve the SOP.

Your platform or team

Build

Retell, Vapi, Bland, ElevenLabs or your own code runs the agent.

Loops

Test, check, prove

Grades every call and every release against your rules.

You

Improve

Your team writes the fix and decides what ships.

Compare

Loops vs platform QA and voice agent testing tools

Platform QA and test tools are good at their jobs. Loops does a narrower one and runs alongside both.

Capability LoopsIndependent audit Platform QABuilt into Retell, Vapi, Bland, ElevenLabs Agent test toolsCoval, Cekura, Hamming Your own LLM judgeBuilt in-house QA team todayManual sampling
Graded against your own SOP, rule by rule Yes: Approved, versioned rules Partly: Custom criteria, their format Partly: Custom metrics, their format Partly: Whatever your team wrote Yes: On the calls they sample
Every production call, not a sample Yes: Every call Partly: Every call or a sample you size, by platform Partly: Varies by tool Yes: Every call No: Under 5%, by hand
Human and AI calls on one rubric Yes: Same rules, same screen No: AI only No: AI only Partly: If you build it Partly: Humans only
This release against the last, per rule, in production Yes: With the calls attached Partly: Scores per criterion; no side‑by‑side release report Partly: Mostly before release Partly: If you build it No: No
Independent of whoever runs the agent Yes: Yes Partly: Built in, by the same vendor Yes: Yes No: No, the team that ships grades it Yes: Yes
Real-time latency and interruption monitoring No: No, Loops grades after the call Yes: Yes Yes: Yes No: No No: No

Based on each vendor's public documentation, September 2026. If a cell is out of date, tell us through the form and we'll fix it. Product names are trademarks of their owners; Loops is not affiliated with or endorsed by any of them.

Rule library

Compliance rules for collections, insurance and healthcare calls

Start from a library of rules for your industry, each tagged with where it comes from, then add your own. It's quality assurance and call monitoring on every call, not a sample.

Rule library ready

Collections and loan servicing

FDCPA · Reg F · TCPA · state rules

  • Give the disclosure before discussing the debtReg F
  • Confirm the right party before account detailFDCPA
  • Seven calls in seven days, per person and per debtReg F
  • Voicemails kept to the limited-content messageReg F
  • Consent on file before each AI-voice callTCPA

Sample verdictMissed

Balance quoted before the debt collection disclosure

How Loops audits collection calls

Rule library ready

Insurance

State unfair claims laws · privacy and recording laws

  • Verify the policyholder before any claim detailPrivacy law
  • Say the caller is talking to an automated agentNAIC AI bulletin
  • Never promise an outcome, amount or pay dateClaims law
  • Acknowledge the claim and say what's nextClaims law
  • Tell a caller who disputes how to get a reviewClaims law

Sample verdictMissed

"Your claim will be approved" promised an outcome

How Loops audits claims calls

Rule library ready

Healthcare patient access

HIPAA · TCPA · No Surprises Act · state medical debt laws

  • Verify identity and authority before any PHIHIPAA
  • Name the caller and give a callback numberTCPA
  • Offer self-pay patients a good faith estimateNo Surprises Act
  • No diagnosis, procedure or balance on voicemailHHS guidance
  • Quote only the balance your billing system gaveYour SOP

Sample verdictMet

Date of birth and ZIP confirmed before the balance

Patient-access rules in detail

Security and data

What Loops can touch, and what it can't

Read-only access to your calls

  • A read-only key where the platform offers one
  • Or send finished calls by webhook, with no key

Never writes to your agent

  • Never changes a prompt, flow, setting or number
  • Your team makes every change

Places no calls

  • Reads only calls your agent already made
  • Releases compare on live calls

US storage, encrypted

  • Stored on US systems, encrypted
  • Kept only as long as you set, then deleted

No training on your calls

  • Loops trains no model on what you send
  • Model providers get call text only under a DPA

Agreements on request

  • NDA and DPA on request; no health data before a BAA
  • SOC 2 planned; see the trust center

Request the security packetVisit the trust center

The free audit

Your first 30 days

No build work on Retell, Vapi, Bland, ElevenLabs, Ultravox, Twilio, Telnyx or Amazon Connect. LiveKit, Pipecat and in-house agents add an OpenTelemetry exporter.

  1. Day 0
    Connect
    • Ask for an NDA and DPA; for healthcare, a BAA comes before any health data
    • Connect your platform, choose which calls count
    • Send the SOP you already have
  2. Days 1 to 2
    Approve the rules
    • Loops turns your SOP into numbered rules
    • Your compliance lead reviews every rule
    • Each rule sits beside the SOP line it came from
  3. After approval
    First audit
    • Last month's calls, graded rule by rule
    • Every verdict quotes the line it relied on
    • No call history? We grade new calls from day one
  4. Day 30
    Your report
    • How often Loops agreed with your QA team
    • What changed between releases, rule by rule
    • Nothing renews. Delete your data any time

FAQ

Questions about auditing AI voice agent calls

Does Loops work with my platform?

Yes, if the platform can send a finished call. Retell, Vapi, Bland, ElevenLabs, LiveKit, Pipecat, Twilio, Telnyx, Ultravox and Amazon Connect are supported today, and anything else that posts a transcript by API or webhook works the same way.

setup · 1 webhook, no SDK, no change to your agent
Does Loops change my agent or write my prompts?

No. Loops reads finished calls and grades them. It never writes to your agent, your prompts or your dialer, and it places no calls.

read only · places no calls
Why not grade calls with our own LLM judge?

Because a judge built by the team that ships the agent grades its own work. Loops is independent, uses the same approved rubric across every release, and quotes the transcript line behind each verdict so the grade can be checked.

your compliance lead approves every rule before anything is graded
What can I hand an examiner or a client auditor?

A rule-by-rule report for any date range, with the quoted transcript line under every verdict and a record of which SOP revision was in force at the time.

FDCPA and Reg F for collections · HIPAA patient access for healthcare
Can Loops put a dollar figure on a release?

Not today. Payments live in your own systems, and Loops doesn't read them. What Loops measures is each rule's miss rate, release against release, on live calls, with the calls behind every change listed.

v14 vs v13 · rule by rule
When does the first audit start?

Once your compliance lead approves the rules. Send the SOP you already have, approve the numbered rules Loops drafts from it, and Loops grades last month's calls against them.

30-day audit · no change to your stack

Free 30-day audit

Find out what your agent said last month.

Tell us what you run, and a person on our team will reply with next steps.

  • Every call from last month, graded against your SOP
  • NDA and DPA available on request
  • The report is yours to keep, whether or not you continue
What runs your agents? Pick any
What would you like first?

We use your email only to reply about this audit. See our privacy policy.

Prefer to talk? Book a 30-minute call.