AI call QA that scores every call against your goals
Set goals and competencies for each agent. After every call, NuTalk marks each goal met or not met, says why and scores the call out of 100.
Scored after the call · Ready within hours or overnight, depending on your plan
What is AI call QA?
AI call QA, or automated call quality assurance, reviews each call against a scorecard instead of a supervisor spot-checking a few by hand. It's the part of conversation intelligence that answers one question: did the agent do its job?
In NuTalk, the scorecard is the goals and competencies you set for each agent, and every verdict comes with a written justification you can check against the transcript.
Goals: met or not met
Yes-or-no outcomes such as "Booked an appointment", set per channel, each with the reason for the verdict.
Competencies: 0 to 5
Skills such as empathy or handling objections, rated on the 3 or 4 performance levels you write.
Overall: out of 100
One number per call that weighs the goals and competencies together.
How it works
How NuTalk scores a call
Set goals and competencies
Write goals for each channel, and competencies for each agent: what good looks like and 3 or 4 performance levels. NuTalk can draft a first set from your agent's instructions.
It scores the call after it ends
Once the transcript is ready, the evaluation runs in a queue. Scores are ready within hours or overnight, depending on your plan.
Review the verdicts
Open any evaluation on its call, next to the recording and transcript, or browse the log by goal or channel.
Per agent
See how each agent is doing
Every evaluation adds up on the agent's page, so you fix patterns instead of chasing single calls.
Goal success rates
Each goal's success rate by day, week, month or quarter, filtered by channel or agent version.
Suggestions for each agent
When a competency averages below 3 out of 5, the agent's page flags it and points to what to tighten in the instructions.
WhatsApp chats too
Chats are scored with the same competencies and the goals you set for WhatsApp, once the conversation goes quiet.
Goals and competencies via the API
Create and update them from your own systems with the REST API.
Plans
Call QA on every plan
- Goals, competencies and an overall score out of 100
- WhatsApp chats scored too
- Goal success rates and suggestions per agent
Pay as you go
Start free$0 / month · USD
Free credits to start, then minutes as you need them. The option if you only want a NuTalk number.
Included
Each evaluation is charged in credits.
NuTalk SMB
Self-serve$299 / month · USD
Unlimited AI calls on your own line through the AiBox, one at a time; fair-use review above 5,000 min/mo per line.
Included
Evaluations are confirmed for your package.
NuTalk Call Center
Sales-assistedFrom $599 / month · USD
Unlimited AI calls for campaigns, outbound and inbound; fair-use review above 5,000 min/mo per line.
Included
Evaluations are confirmed for your package.
- AiBox installation for NuTalk SMB: $149 USD one time, quoted and billed separately.
- Connection to your PBX or phone system (SIP) or Zoom Phone is scoped and set up with our team.
- Simultaneous-call capacity, numbers, carrier usage and hardware are confirmed for your package. Fair use applies: we review usage above 5,000 minutes a month per line.
Limits
What it scores, and what it doesn't
- It scores the calls and chats your NuTalk agents handle. It doesn't import or score calls your team takes on another phone system.
- Scores come after the call, never during it: within hours or overnight, depending on your plan.
- A score is an AI judgment. Read the justification and the transcript before acting on a single result.
- Suggestions are per agent, based on each competency's average. A single call gets a justification, not coaching notes.
FAQ
Call QA questions
What is call center QA?
Call center quality assurance means reviewing calls against a scorecard to check that agents follow the process and treat customers well. Traditional QA samples a few calls by hand. By default, NuTalk scores every call your AI agents handle against the goals and competencies you define, and writes down why each goal was or wasn't met.
How is each call scored?
Each goal is marked met or not met with a written justification. Each competency gets one of the performance levels you write, with a written justification; every competency counts the same. The call also gets an overall score out of 100.
How soon after the call is it scored?
Evaluations run in a queue after the call, once the transcript is ready. Scores are ready within hours or overnight, depending on your plan. Nothing is scored during the call.
Does it score WhatsApp chats?
Yes. Once a chat has been quiet for the time you set on the channel, it's scored with the agent's competencies and the goals you set for WhatsApp. Each evaluation links back to its conversation.
Can I choose what gets scored?
Yes. You write the goals and competencies and can change them at any time. Scoring is on for every channel by default, for 100% of calls and chats; you can lower the sample rate or turn it off per channel.
Is call QA included in my plan?
Yes, on every plan. With pay as you go, each evaluation is charged in credits. On NuTalk SMB and NuTalk Call Center, evaluations are confirmed for your package.
Where does the scoring run?
Call scoring uses AI models from vetted cloud partners that don't keep the data or train on it, or NuTalk's own servers for an extra fee. NuTalk keeps your call history and quality logs as described in our privacy policy. We don't use your conversations to train general-purpose AI models.
Goals · Competencies · Overall score
See how your calls score
Request a demo, then build your first agent for free. Move to a flat plan when the calls add up.