Roark Live Bench · Suite v1.3 · Sweep of 23 October 2026
Voice AI models, ranked on live phone calls.
Every speech-to-text, language model and text-to-speech model, run in every combination on the same real call to the same agent. A model is ranked on every stack it runs in; the stacks are the evidence.
Our picks
Task completionMedian responseCost per minute- Best overallAssemblyAI Universal-3.6 Pro + OpenAI GPT-6.1 Sol + Cartesia Sonic-3.697%3.9 s$0.055
- Best valueAssemblyAI Universal-3.5 Pro + Google Gemini 3.8 Flash + Cartesia Sonic-3.597%4.3 s$0.048
- FastestOpenAI Whisper-1 + Google Gemini 3.8 Flash + OpenAI TTS-1 HD88%0.5 s$0.023
- Best speech-to-speechgrok-voice-think-fast-2.082%2.1 s$0.083
- Best in customer supportAssemblyAI Universal-3.6 Pro + OpenAI GPT-6.1 Sol + ElevenLabs v4 Turbo100%3.9 s$0.096
- Best in healthcareAssemblyAI Universal-3.5 Pro + Google Gemini 3.8 Flash + Cartesia Sonic-3.696%4.4 s$0.041
- Best in home servicesAssemblyAI Universal-3.6 Pro + OpenAI GPT-6.1 Sol + Cartesia Sonic-3.596%4.0 s$0.058
- Best in insuranceAssemblyAI Universal-3.5 Pro + OpenAI GPT-4o + Cartesia Sonic-3.698%3.2 s$0.042
The read
- 430 of 642 cascades complete the task in at least 90% of calls, and the top 10 sit within 1 point of each other. What separates the top stacks is speed and cost.
- Median response time runs from 0.5 s to 6.0 s on the same agent and the same caller, a spread of 5.5 s that comes entirely from the models behind it.
- No stack wins everything. The fastest, OpenAI Whisper-1 + Google Gemini 3.8 Flash + OpenAI TTS-1 HD at 0.5 s, ranks 530 of 642 on task completion; the most complete is Speechmatics Linden-1 (own VAD) + OpenAI GPT-6.1 Sol + OpenAI TTS-1 HD at 99%.
- The language model matters most: swapping it moves task completion by 7 points, against 3 for the speech engine and 2 for the transcriber.
Cascade vs Speech-to-Speech
The first choice to make. A cascade chains three models: one transcribes, one decides what to say, one speaks. A speech-to-speech model does all three at once. Both run the same agent on the same calls here.
- Cascades lead on task completion, tool use and time to first word; speech-to-speech leads on response time and cost per task. Neither wins outright, so the right choice depends on what your calls need.
- The biggest difference is task completion: 95% for cascades against 77% for speech-to-speech.
| Measure | Cascade | Speech-to-speech |
|---|---|---|
| Task completion | 95% (leads) | 77% |
| Tool use | 72% (leads) | 55% |
| Time to first word | 0.7 s (leads) | 1.3 s |
| Response time | 3.6 s | 2.6 s (leads) |
| Cost per task | $0.15 | $0.15 (leads) |
The three strongest of each, averaged. Cascades: Speechmatics Linden-1 (own VAD) + OpenAI GPT-6.1 Sol + Cartesia Sonic-3.5; AssemblyAI Universal-3.6 Pro + OpenAI GPT-6.1 Sol + Rime Coda; Speechmatics Linden-1 (own VAD) + OpenAI GPT-6.1 Sol + Rime Coda. Speech-to-speech: grok-voice-think-fast-2.0; gemini-3.8-live; amazon.nova-2-sonic-v1:0. Each measure is scored 0 to 100 on its own scale, further out is better; the table gives the same figures in their own units.
Which to use
- Choose speech-to-speech when the pause matters more than the outcome: short transactional calls, callers who interrupt, queues where every half second costs you callers. Expect to pay more, and choose the model carefully (gpt-live-1 (backend gpt-6.1-sol) leads here).
- Choose a cascade when the call has to complete, when cost matters, or when you need to see which stage to fix. It is 13 points ahead on task completion here.
- Costs are what each stack’s providers charged for these calls, at the rates in effect when they ran. Telephony and hosting are not included.
Speech-to-Text
The transcriber turns the caller's speech into text. It decides what the agent thinks it heard, and nothing after it can start until it finishes.
- AssemblyAI Universal-3.6 Pro, AssemblyAI Universal-3.5 Pro, Deepgram Nova-3 and Speechmatics Enhanced (realtime) are level: none leads by more than the margin of error. Of them, AssemblyAI Universal-3.6 Pro costs least, at $0.030 a call. Avoid Speechmatics Linden-1 (own VAD): last on semantic word error rate (2.2%).
- Speechmatics Linden-1 (own VAD) delivers its transcript in 0.49 s, OpenAI Whisper-1 in 1.93 s: 1.44 s added to every turn of the call.
- AssemblyAI Universal-3.6 Pro is the most accurate: it changes the meaning of 0.9% of the caller's words, against 2.2% for Speechmatics Linden-1 (own VAD).
- In 341 of 648 cascades the transcriber is the slowest step of the turn, whichever model runs it.
Open a model to see where it ranks and the stacks it works best in.
Language model
The language model decides what the agent says and does. It has the most influence on whether the task gets done, and usually takes the largest share of each turn.
- Pick OpenAI GPT-6.1 Sol: it leads on task completion (95%), tool use (71%) and instruction following (4.2 / 5). If speed matters more, Anthropic Claude Haiku 4.5 is the quickest, at 0.36 s, and completes 91%. Avoid OpenAI GPT-5.1: last on tool use (33%) and instruction following (4.1 / 5).
- Anthropic Claude Haiku 4.5 starts answering in 0.36 s, Google Gemini 3.8 Flash in 1.47 s: 1.11 s added to every turn of the call.
- OpenAI GPT-6.1 Sol uses its tools most reliably: it handles its tools without a mistake on 71% of calls, against 33% for OpenAI GPT-5.1.
- In 161 of 648 cascades the language model is the slowest step of the turn, whichever model runs it.
Open a model to see where it ranks and the stacks it works best in.
Text-to-Speech
The speech engine is the voice the caller hears, and the last step before they hear anything.
- Cartesia Sonic-3.6, Cartesia Sonic-3.5 and ElevenLabs v4 Turbo are level: none leads by more than the margin of error. Of them, Cartesia Sonic-3.5 costs least, at $0.165 a call. Avoid OpenAI TTS-1 HD: last on readback accuracy (80%), word error rate (8.7%) and time to first audio byte (2.09 s).
- ElevenLabs v4 Turbo starts speaking in 0.11 s, OpenAI TTS-1 HD in 2.09 s: 1.97 s added to every turn of the call.
- Cartesia Sonic-3.6 is the most accurate: it reads codes, amounts and dates back correctly 87% of the time, against 80% for OpenAI TTS-1 HD.
- Cartesia Sonic-3.6 sounds the most human: 3.8 / 5, against 3.4 / 5 for ElevenLabs Flash v2.5.
- In 99 of 648 cascades the speech engine is the slowest step of the turn, most often with OpenAI TTS-1 HD (54 of them).
Open a model to see where it ranks and the stacks it works best in.
Speech-to-Speech models
One model listens, reasons and speaks, so there is nothing to swap and no stages to split. Each is judged on the whole call, with the same calls and the same scoring as every cascade.
- The model matters as much as the architecture: gpt-live-1 (backend gpt-6.1-sol) completes 86% of tasks, ElevenLabs Realtime v1 Mini 42%.
- grok-voice-think-fast-2.0 answers in 2.1 s against 4.1 s for gemini-2.5-flash-native-audio-preview-12-2025.
- Price varies as much as quality: $0.06 per completed task for amazon.nova-2-sonic-v1:0, $0.69 for gpt-realtime-2.
Open a model for its full report, every figure it was scored on and the calls behind them.
Every stack
The evidence behind the model boards: each combination on its own calls, scored on the same recipe. Open a stack for its full report, every figure it was scored on and the calls behind them.
Each score carries its 95% margin of error. A rank is the best place a stack could hold once every margin is allowed for, so stacks the calls cannot tell apart share one and are listed by the low end of their margin; hover a rank for the worst place it could hold. 92 stacks are level with the leader: their calls cannot tell them apart. Among them, 31 Google Gemini 3.8 Flash, 29 OpenAI GPT-6.1 Sol, 19 Anthropic Claude Sonnet 5.5, 9 Anthropic Claude Haiku 4.5, 2 OpenAI GPT-4o mini, 1 OpenAI GPT-4.1 and 1 speech-to-speech; 48 AssemblyAI, 23 Deepgram and 20 Speechmatics; 48 Cartesia, 33 ElevenLabs, 9 Rime and 1 OpenAI. A cascade's figures are 25% its own calls and 75% what every other stack shows about its transcriber, language model and voice.
Not ranked here for missing data: Speechmatics Linden-1 + OpenAI GPT-6.1 Sol + OpenAI TTS-1 HD (no response time or tool use or human-likeness or readback accuracy or voice consistency).
By industry
Each industry ranked on its own calls: the same five callers, its own tasks and tools. A stack has a handful of calls in each, so a cascade is ranked on what its transcriber, language model and voice show across every stack in that industry, and a speech-to-speech model on its own calls there. Pairings show on the main board. Every panel is on one scale, so a bar reads against the other industries as well as its own.
Healthcare
Reschedule an appointment. A patient moves a dermatology appointment around a work schedule, corrects their date of birth, narrows the times twice, and asks about the change fee, parking and insurance.
Home services
Urgent repair: triage and booking. A caller with a burst pipe needs safety advice first, then an emergency visit. The unit on file is out of date, the callback number gets corrected, and the caller pushes for an earlier visit.
Insurance
Report a new auto claim. A driver who was just rear-ended reports a claim, corrects the policy number, spells the other car’s plate, asks about the deductible, a rental and their premium, and needs a tow.
Customer support
Internet outage troubleshooting. A technical caller who has already rebooted the gateway pushes back on scripted steps, wants the line fixed, and asks for a bigger credit for the downtime.
How it works
Every number on this page comes from real phone calls, placed and scored the same way for every stack.
Real calls, four industries, five callers
Roark places real phone calls over the public phone network to a reference agent for each of four industries: rescheduling a clinic appointment, booking an urgent home repair, opening an auto insurance claim and fixing an internet outage. Each agent has its own prompt and tools, identical for every stack. The tasks carry the details real calls trip on: a corrected date of birth, a policy number spelled out, a constraint added halfway through.
Every industry is called by the same five simulated callers: a quiet US English line, a Spanish-accented caller on a city street, an Indian English caller in a busy café, a flustered Southern caller with a baby crying, and a fast caller who interrupts over construction noise. They are equally clear about what they want, so a call fails because of the agent, not the caller. Only the models change, so any difference on the board comes from them.
Each caller rings each industry once for a cascade and four times for a speech-to-speech model. A cascade's figures also draw on its parts, heard in dozens of other stacks; a speech-to-speech model has only its own calls, so it takes more of them.
That makes industries comparable over the same callers, and callers over the same tasks, and the board can be re-ranked by either. One caller within one industry is too few calls to judge, so the board never shows it.
655
stacks
20
calls per stack
13,876
calls scored
Every model, in every combination
9 transcribers, 8 language models and 9 voices are run in every combination with each other, beside 13 speech-to-speech models. A model's own figures pool every stack it runs in, adjusted for its partners in each one, so every model is compared as if it ran with the same partners. A transcriber's figure rests on hundreds of calls with every language model and every voice, not on one lucky pairing.
That measures a model where it actually works: inside a running agent, on live phone audio, through interruptions, corrections and real turn-taking. A test that feeds a model clean audio files or one prompt at a time never sees any of that.
One score, weighted by what matters on a call
Each section ranks on a Roark score. Every figure scores 0 to 100 between two published points: full marks where doing better makes no difference to the caller, zero where the call breaks down. Each then counts at its weight, and the weights follow one order everywhere: whether the caller got what they rang for, then whether the agent got it right, then speed.
A stack, cascade or speech-to-speech, is scored on the outcome and the reply first, then on what each of its components got right on its own calls: the words its transcriber got wrong in a way that changes the meaning, whether its voice read codes and amounts back right, how human it sounded, and whether its tools were handled. A component that leads its section lifts every stack that carries it, by the same recipe for every stack. A figure a stack cannot carry, such as a semantic word error rate for a speech-to-speech model with no transcript, is credited at the field median, so the stack is neither flattered nor punished for it.
On a full sweep each cascade's figure is 25% what its own calls measured and 75% what every other stack shows about its transcriber, language model and voice, fitted across the whole sweep. The split is the same for every stack, so a lucky run does not lead and a stack's score does not move with how noisy its calls were. In an industry a stack has a handful of calls, so a cascade's figure there is what its transcriber, language model and voice show across every stack in that industry's calls, and a speech-to-speech model's is its own calls there. How a particular pairing beats or misses its parts shows on the main board, where each stack has all of its calls. The best overall pick, on the main board and in each industry, is the stack whose score is highest at the low end of its margin of error, so a few lucky calls cannot carry a stack past one whose score rests on hundreds. Stacks that share a rank are listed the same way, so the first row is the pick.
That order comes from real-world voice agent calls, where getting the conversation right predicts how a call ends more strongly than speed does. The latency points come from the same calls: replying faster than about 2 seconds made no measurable difference, and calls went measurably worse once the agent took more than about 3 seconds to reply.
Stacks and speech-to-speech
| Figure | Weight | Full marks | Zero |
|---|---|---|---|
Task completionscenario_adherence_check | 21% | 100% | 80% |
| An agent that completes fewer than four calls in five is not fit to deploy. On a full sweep, a cascade's figure is 25% what its own calls (20 or more) measured and 75% what every other stack shows about its transcriber, language model and voice, the same split for every stack, so a lucky run does not lead. Where the sweep is too small for that, and for a speech-to-speech model, the figure is the stack's own. In an industry a stack has a handful of calls, so a cascade's figure there is what its transcriber, language model and voice show across every stack in that industry's calls, and a speech-to-speech model's is its own calls there. How a particular pairing beats or misses its parts shows on the main board, where each stack has all of its calls. | |||
Response timeresponse_time | 17% | 2.0 s | 6.0 s |
| In real calls, replying faster than 2 s made no measurable difference, and calls went worse past 3 s. | |||
Instruction followinginstruction_follow | 13% | 5.0 / 5 | 3.0 / 5 |
| Every model scores between 3 and 5 on the judge's 1 to 5 scale, so that is the range. | |||
Semantic word error ratestt_semantic_word_error_rate | 13% | 0.5% | 3% |
| At 0.5% the best transcribers change the meaning of one word in two hundred on phone audio; at 3% one word in thirty sends the agent the wrong way. | |||
Tool usetool_use | 13% | 100% | 50% |
| An agent that mishandles its tools on half its calls cannot be trusted with the systems it runs on. | |||
Human-likenessvoice_human_likeness | 8% | 4.5 / 5 | 2.5 / 5 |
| 4.5 out of 5 is close to a person on the phone; 2.5 sounds plainly synthetic. | |||
Time to first wordtime_to_first_word | 4% | 1.0 s | 4.0 s |
| A greeting within 1 s sounds natural; 4 s of silence sounds like a dead line. | |||
Readback accuracytts_entity_readback_accuracy | 4% | 100% | 60% |
| Reading a code, amount or date wrong breaks the task it was for; the weakest voices get two in five wrong. | |||
Word error ratetts_word_error_rate | 4% | 2% | 12% |
| At 2% the voice says what it was given; at 12% one word in eight comes out wrong. | |||
Voice consistencyvoice_consistency_check | 4% | 100% | 80% |
| A voice that changes speaker on one call in five is not one a business can put its name to. | |||
Shown, not scored: Turn budget, Barge-in handlingbarge_in_yield_appropriateness
Counted only with price factored in: Cost per task
Language model
| Figure | Weight | Full marks | Zero |
|---|---|---|---|
Task completionscenario_adherence_check | 36% | 100% | 80% |
| An agent that completes fewer than four calls in five is not fit to deploy. | |||
Tool usetool_use | 27% | 100% | 50% |
| An agent that mishandles its tools on half its calls cannot be trusted with the systems it runs on. | |||
Instruction followinginstruction_follow | 18% | 5.0 / 5 | 3.0 / 5 |
| Every model scores between 3 and 5 on the judge's 1 to 5 scale, so that is the range. | |||
Time to first tokenllm_time_to_first_token | 18% | 0.50 s | 2.00 s |
| Under 0.5 s is not noticeable; at 2 s the first token alone takes two thirds of the 3 s past which real calls go worse. | |||
Shown, not scored: Prompt cache hitsllm_prompt_cache_hit_rate
Counted only with price factored in: Cost per call
Speech-to-Text
| Figure | Weight | Full marks | Zero |
|---|---|---|---|
Semantic word error ratestt_semantic_word_error_rate | 71% | 0.5% | 3% |
| At 0.5% the best transcribers change the meaning of one word in two hundred on phone audio; at 3% one word in thirty sends the agent the wrong way. | |||
Time to final transcriptstt_time_to_final_transcript | 29% | 1.00 s | 2.50 s |
| The fastest transcribers finish in about 1 s; past 2.5 s the transcript alone uses most of a natural reply time. | |||
Shown, not scored: Word error ratestt_word_error_rate
Counted only with price factored in: Cost per call
Text-to-Speech
| Figure | Weight | Full marks | Zero |
|---|---|---|---|
Readback accuracytts_entity_readback_accuracy | 32% | 100% | 60% |
| Reading a code, amount or date wrong breaks the task it was for; the weakest voices get two in five wrong. | |||
Word error ratetts_word_error_rate | 21% | 2% | 12% |
| At 2% the voice says what it was given; at 12% one word in eight comes out wrong. | |||
Time to first audio bytetts_time_to_first_byte | 21% | 0.20 s | 1.50 s |
| Under 0.2 s is not noticeable; 1.5 s is half the point where calls go worse. | |||
Human-likenessvoice_human_likeness | 16% | 4.5 / 5 | 2.5 / 5 |
| 4.5 out of 5 is close to a person on the phone; 2.5 sounds plainly synthetic. | |||
Voice consistencyvoice_consistency_check | 11% | 100% | 80% |
| A voice that changes speaker on one call in five is not one a business can put its name to. | |||
Counted only with price factored in: Cost per call
Factor in price and cost takes 25% of the score; the other weights keep their proportions. Price is scored on a log scale, so each doubling costs the same points.
A figure listed as shown, not scored is on the section’s boards and outside its score. Where a section has not collected a scored figure yet, the others share its weight.
Every call recorded and judged
Every call is recorded. Judge models score each transcript for the outcome and the conversation, and the timings come from the recording. Every figure links to the calls behind it.
Roark is on both sides of every call: it plays the caller, so it keeps the exact text its caller said, and it runs the agent under test, so it sees what the agent’s transcriber heard and what its model wrote for its voice. The transcript and voice figures come from lining those up, with no hand-made transcript in between. Each carries a margin from the synthetic voice or Roark’s transcriber in the loop. Every model faces the same ones under the same conditions, so the figures compare models on the same calls through the same voice and the same ear, though those slips need not fall evenly on each model. The rates will not match a vendor’s figure on a curated set; read them against each other.
An example: one call from this sweep
Speechmatics Linden-1 (own VAD) + OpenAI GPT-6.1 Sol + OpenAI TTS-1 HD, 23 October 2026. Highlighted turns are the ones that affected a score. Every stack's report links to its calls.
See how your own agent compares
Roark runs the same kind of calls against your agent, on your scenarios, with the same scoring.
Rank on Roark score
#226of 654 stacks
51 / 100
Leader AssemblyAI Universal-3.6 Pro + OpenAI GPT-6.1 Sol + Cartesia Sonic-3.6, 68 / 100
Where it ranks
Its models
Task completion
Task completion
How much of the caller’s task the agent completed: the share of the scenario’s checks a judge model found met in the transcript, averaged over calls. Before v1.3, the share of calls judged 4 or better out of 5.
#214 of 661 on the board · best 99%
Flow adherence
How closely the agent followed the intended conversation flow, judged per call. 5 = fully followed.
#100 of 661 on the board · best 5.0 / 5
Instruction following
How well the agent stuck to the instructions in its own prompt. 5 = followed throughout.
#12 of 661 on the board · best 5.0 / 5
Turn-taking
Barge-in handling
When the caller interrupts, how often the agent made the right call: yielding to a real correction, and holding the floor through a filler.
Not scored by this sweep.
Incorrect interruptions
How often the agent interrupted the caller when it should have kept listening.
#1 of 661 on the board · best 0%
Overtalk
Share of the conversation where both sides were speaking over each other.
#448 of 661 on the board · best 0.0%
Agent self-cutoffs
Times the agent cut its own sentence off mid-word.
#1 of 661 on the board · best 0.0 per call
Understanding & effort
Caller effort
How hard the caller had to work: repeating themselves, correcting, re-explaining. 1 = effortless.
Not scored by this sweep.
Repeated-info loops
Times per call the agent re-asked for something the caller had already given.
#180 of 493 on the board · best 0.0 per call
Speech-to-Text
Semantic word error rate
The caller's words the agent's transcriber got wrong in a way that changes the meaning, as a judge reads each difference.
#446 of 648 on the board · best 0.0%
Word error rate
The caller's words the agent's transcriber missed, changed or invented, against what the caller actually said.
#514 of 648 on the board · best 1.2%
Text-to-Speech
Readback accuracy
Share of the codes, amounts, dates and phone numbers the agent wrote that its voice read out correctly. One wrong digit is a miss.
#573 of 603 on the board · best 100%
Word error rate
The share of the words the agent wrote that its voice dropped or changed, read back from Roark’s transcription of the recording; the same transcriber hears every voice.
#613 of 648 on the board · best 1.9%
Voice
Human-likeness
How human the agent's voice sounds, scored by an audio model on intonation, rhythm and expressiveness. 5 = hard to tell from a person.
#49 of 659 on the board · best 4.3 / 5
Voice consistency
Share of calls where every line the agent spoke was the same speaker, with no change of speaker, gender or language voice partway. It checks who is speaking, not how they sound.
Not scored by this sweep.
#1 of 659 on the board · best 100%
Latency
Response time
The silence between the caller finishing and the agent starting to speak, measured from the recording on every turn of the call.
#81 of 660 on the board · best 0.50 s
Time to first word
How long the caller waits before the agent first speaks after pickup.
#310 of 661 on the board · best 0.45 s
Latency spikes
Turns where the response took far longer than the rest of the conversation.
#55 of 661 on the board · best 0.0 per call
By industry
| Industry | Task completion | Latency P50 | Barge-in handling |
|---|---|---|---|
| Healthcare | 100% | 2.86 s | — |
| Home services | 85% | 2.41 s | — |
| Insurance | 98% | 2.63 s | — |
| Customer support | 91% | 3.68 s | — |
By caller
| Caller | Task completion | Latency P50 | Barge-in handling |
|---|---|---|---|
| US English, quiet line | 94% | 2.63 s | — |
| Spanish-accented English, city street | 89% | 3.22 s | — |
| Indian English, busy café | 96% | 3.21 s | — |
| Southern US, flustered, crying baby | 98% | 2.93 s | — |
| Fast and interrupting, construction outside | 91% | 2.35 s | 100% |
The calls behind the numbers
Customer support
AGENTBrightwave Internet support, this is Alex. What’s going on with your service?
CUSTOMERHi, Alex.
CUSTOMERMy internet has been down for about two hours now. I work from home.
CUSTOMERSo it's been a problem. There's a road crew digging up the street just outside my building. I already rebooted the gateway twice and checked all the cables.
CUSTOMERThe internet light is blinking orange.
CUSTOMERCan you check if the road work is causing this outage? My account number is BW883017.
CUSTOMERSorry, last digit is one.
AGENTThanks for the correction—reading this back: I’ve got your account as BW883011. Here’s what I’m seeing and what I did: - No area outage is showing for your address, so nothing official tied to the road work on our side. - The line test showed good signal, but your gateway wasn’t registered. - I pushed a remote reset to your Brightwave BX7 just now, and it’s now showing as registered and online. Devices may take about 2 minutes to fully reconnect. On your side, over the next couple of minutes: 1. Watch the internet light—it should go to solid white (or solid green, depending on your model) once it’s fully online. 2. Try loading a site on your work computer and one other device (like your phone on Wi‑Fi). If, after 2–3 minutes, the light is still blinking orange or you still can’t get online, tell me exactly what the light is doing and whether any device can connect, and I’ll move to the next step (likely a tech visit, given the road work).
CUSTOMERRight?
CUSTOMERGot
AGENTYou should be coming back online now. I’m seeing your gateway as registered and online on my end, so within a minute or two: - The internet light should be solid (not blinking orange). - Your devices should reconnect, or you can manually reconnect Wi‑Fi if needed. When you get a second, try a couple of things: - Open a website on your work computer. - Run a quick speed test if you use one for work (optional). If anything’s still down or that light stays blinking orange, tell me what the lights look like now and whether any device can get online, and I’ll take it from there.
CUSTOMERit. The internet lights just went.
CUSTOMERAha! The light just went solid white.
CUSTOMERHow long until my devices reconnect?
CUSTOMERAlso, since I lost two hours of work. Can you issue a credit for the downtime?
AGENTSolid white is exactly what we want—that means the connection is back up. On your devices: - Most should reconnect within 1–3 minutes on their own. - If something doesn’t, just toggle Wi‑Fi off/on or disconnect/reconnect to your Wi‑Fi network and it should pick it up right away. About the credit: I’ve applied a $10.00 service credit on account BW883011 for the downtime you experienced. You’ll see that $10.00 credit on your next bill. If any specific device is still not online after a couple of minutes, tell me which device and what it shows (e.g., “connected, no internet” or Wi‑Fi not connecting), and I’ll walk you through just that piece.
CUSTOMERI lost two hours of work.
CUSTOMERCan you make it more than $10?
AGENTI understand why you’re asking, especially if you lost actual work time. I’m sorry this hit you in the middle of your day. Right now I’ve applied a $10.00 credit on account BW883011, which will show on your next bill. That’s the standard credit I’m allowed to issue directly for a short outage like this. I don’t have the ability to increase that amount myself. If you want to request more based on the impact to your work, the next step is to contact our billing team so they can review it: - From your Brightwave account portal: use Support → Billing Chat and mention today’s 2‑hour outage and the existing $10 credit. - Or by phone: ask for Billing and reference account BW883011 and today’s downtime. They’re the ones who can review and, if appropriate, approve anything beyond the standard credit.
CUSTOMEROkay.
CUSTOMERUnderstood.
CUSTOMERThank you for your help, Alex.
CUSTOMERGoodbye.
AGENTYou’re welcome. I’m glad we got you back online.
AGENTQuick recap before you go: - Your gateway is online with a solid white light. - A $10.00 credit has been applied to account BW883011 and will appear on your next bill. If anything acts up again, just reach out. Goodbye, and take care.
Healthcare
AGENTRiverbend Health Clinic, this is Alex. How can I help you today?
CUSTOMERHi, Alex.
CUSTOMERI am having one appointment with dermatology next Wednesday morning.
CUSTOMERI need to move it because I have a work shift now. At that time, can you help me reschedule?
CUSTOMERMy name is Arjun Mehta.
AGENTI can help with that. I’ll need a couple more details to look you up and find your appointment. 1. What is your date of birth? 2. When you say “next Wednesday
CUSTOMERLa. Sorry. I got distracted there. My date of birth is March 17th, 1948.
CUSTOMERSorry. No. 1984. March 17th, 1984.
CUSTOMERAbout the.
CUSTOMERAppointment. It's next Wednesday morning.
CUSTOMERI need to move it to after 3 p.m..
AGENTThanks for the correction, Mr. Mehta. I have you as: - Name: Arjun Mehta. - Date of birth: March 17, 1984. - Current dermatology visit: next Wednesday at 9:15 a.m. with Dr. Okafor, Main Campus, Building B. You asked to move it to after 3 p.m.. With Dr. Okafor, the next available late-day options I can offer are: - Tuesday at 3:45 p.m. - Thursday at 4:10 p.m. There’s also a Friday slot, but it’s 11:30 a.m., which is before 3 p.m., so that doesn’t meet your request. Would you like to move your appointment to: - Tuesday at 3:45 p.m., or. - Thursday at 4:10 p.m.? Once you choose, I’ll reschedule it and read back the new details and confirmation code.
CUSTOMERTuesday. Does not work after all. I pick up my daughter on Tuesdays.
CUSTOMERWhat about Friday? Any slots available after 3 p.m.? If not, then I will take Thursday at 4:10 p.m..
AGENTThanks for the update. I checked Friday for you—Dr. Okafor only has 11:30 a.m. available that day, nothing after 3 p.m. Since you said you’ll take Thursday if Friday doesn’t work, I’ve gone ahead and moved your appointment. Here are your new details: - Name: Arjun Mehta - Date of birth: March 17, 1984 - Department: Dermatology - Provider: Dr. Okafor - New day and time: Thursday at 4:10 p.m. - Location: Main Campus, Building B - Confirmation code: RVB-3M8P2 Please arrive 15 minutes early with a photo ID and your insurance card. Do you want me to repeat any of that, or help with anything else today?
CUSTOMERNo.
CUSTOMERThat's clear. I have a few quick questions.
CUSTOMERIs there a fee for changing the appointment? Where should I park?
CUSTOMERAnd does Doctor Okafor accept Aetna insurance?
AGENTHere are the answers to those: 1. Fee for changing the appointment. - The clinic’s policy is: changes made more than 24 hours before the appointment are free. - Changes within 24 hours of the appointment time carry a $40 fee. Since we moved your appointment well in advance, there should be no change fee for this reschedule. 2. Where to park. - There is free patient parking in Lot C, with the entrance on River Road. - Give yourself a few extra minutes in case the lot is busy. 3. Does Dr. Okafor accept Aetna insurance? - I don’t have provider-by-insurance-panel access in this system, so I can’t safely confirm Aetna coverage for Dr. Okafor from here. - To be sure, please either: - Call the clinic’s main line and ask if Dr. Okafor is in-network for Aetna, and/or. - Check with Aetna (website or phone) using the clinic and provider name. If you’d like, I can restate your appointment details once more so you have everything in one place.
CUSTOMERYes.
CUSTOMERPlease.
CUSTOMERKindly read. Back the new appointment details and confirmation code.
AGENTHere are your updated appointment details: - Name: Arjun Mehta. - Date of birth: March 17, 1984. - Department: Dermatology. - Provider: Dr. Okafor. - Day and time: Thursday at 4:10 p.m. - Location: Main Campus, Building B. - Confirmation code: RVB-3M8P2. Arrival instructions: Please arrive 15 minutes early with a photo ID and your insurance card.
CUSTOMERYes, that's exactly
AGENTI’m glad that all looks right. Before
CUSTOMERright. Thank you for your help. Alex.
AGENTwe wrap up, do you want me to: - Note anything about your Aetna insurance for the front desk, or. - Go over parking or the change-fee policy one more time?
Home services
AGENTCopperline Home Services, this is Alex. What can I help you with?
CUSTOMERHi. I need emergency plumbing help.
CUSTOMERThere's a pipe under my kitchen sink that's burst. Water is spreading across the floor, heading toward a power strip. I haven't shut the water off yet.
CUSTOMERCan you send someone out right away?
AGENTThanks for telling me what’s going on—that sounds urgent. First, let’s keep you safe and stop the water as much as possible: 1. Stay away from that power strip and any outlets near the water. - Don’t touch the strip, any cords,
CUSTOMERYes.
CUSTOMERI'm already on it.
CUSTOMERJust need someone out here fast.
CUSTOMERHow soon can you get a plumber over?
AGENTWe can get an emergency plumber there today between 2:00 PM and 4:00 PM. Before I book that, a couple quick things so I quote
CUSTOMEROkay, just give me the earliest emergency window you have.
CUSTOMERWhat's the cost for the visit?
CUSTOMERCan anyone come before two?
CUSTOMERI really need the earliest window.
CUSTOMERWhat's the cost for the visit?
AGENTRight now the earliest emergency arrival window I can offer is today, 2:00 PM–4:00 PM—there’s
CUSTOMERRight.
CUSTOMERNothing.
AGENTThanks for hanging in there.
CUSTOMERGot
AGENTGot it, thanks. Earliest emergency window: - Today between 2:00 PM and 4:00
CUSTOMERit. Right.
AGENTI’ve got you—let’s move quickly. 1) Earliest emergency window + cost. - The earliest emergency arrival I can book is today, 2:00 PM–4:00 PM. There’s nothing available earlier in the system. - Your account shows you as a Copperline Care member, so your emergency callout fee is $99 for the visit. Is that okay to proceed with? 2) Confirm your info. I show this on file—tell me what’s right or what needs updating: - Service address: 1480 Larkspur Avenue, Unit 4B. - Best callback number: the one you’re calling from now. Do I have your permission to: 1) Open an emergency plumbing ticket for the burst pipe under the kitchen sink, and. 2) Book the 2:00 PM–4:00
CUSTOMERYeah. That's fine.
CUSTOMERBut I moved last month. It's unit 4D now, not four B, and my callback number is (503) 555-0139.
CUSTOMERNo, 0193. Go ahead and book the visit.
CUSTOMERYep. That's fine. Just to confirm, I moved to unit 4D last month, not four B.
CUSTOMERMy address is 1480 Larkspur Avenue, unit for De.
CUSTOMERMy callback number is (503) 555-0139.
CUSTOMERNo, 0193.
CUSTOMERDate code is pound 2719.
CUSTOMERWho's coming out and how will I know they're on the way?
CUSTOMERCan you give me the ticket number?
AGENTYou’re all set—I’ve booked this as an emergency. Here’s what we’ve got: - Service window (emergency): Today, 2:00 PM–4:00 PM. - Visit fee (Copperline Care member): $99 emergency callout. - Ticket number: CPL-58210. Contact & address I have on the ticket: - Service address: 1480 Larkspur Avenue, Unit 4D. - Gate code: #2719. - Callback number: (503) 555-0193. Please double-check: is all of that correct? If anything’s off,
CUSTOMERYeah.
AGENTPerfect,
CUSTOMERNope. That's
AGENTGot it, I’ll stick with what you told me. Here’s everything confirmed: - Ticket number: CPL-58210. -
CUSTOMERall. Correct. Thanks.
CUSTOMERRight.
AGENTGreat,
CUSTOMERThanks for your help.
AGENTYou’re welcome—glad we could get you on the schedule quickly. Quick recap of what’s set: - Today’s emergency
AGENTwindow: 2:00 PM–4:00 PM. - Address: 1480 Larkspur Avenue, Unit 4D. - Gate code: #2719. - Callback: (503) 555-0193. - Ticket number: CPL-58210. - Emergency visit fee (with Copperline Care): $99. You’ll get a text at that number about 30 minutes before the plumber arrives so you know they’re on the way. If the water starts getting close to any outlets or that power strip again before we arrive, stay clear of it and, if you can safely reach it, shut off the main water valve to the home. Call us back if anything changes or gets worse before the tech arrives.
Insurance
AGENTNorthgate Mutual claims, this is Alex. Are you somewhere safe right now?
CUSTOMERYes.
CUSTOMERI am safe.
CUSTOMERI am standing next to my car at the cafe.
AGENTI’m glad you’re safe. Is anyone injured or in any danger—either in your car or the other vehicle? If anyone is hurt, please hang up and call 911 first, then we can continue once everyone has help on the way. Assuming no one is injured, I’ll start your claim. Please tell me your auto policy number and the policyholder’s last name so I can look it up.
CUSTOMERSure. My policy number is n g.
CUSTOMER4471903. No, sorry.
CUSTOMERLast one is two.
CUSTOMERThe last name is Mehta.
AGENTThank you, Ms. Mehta. I have your policy number as NG4471902 and the last name Mehta. First, I’m glad no one is injured. Let’s get the accident details. 1. When did the accident happen? 2. Where did it happen? 3. What happened? Please briefly describe how the accident occurred. 4. Other driver’s information: - Their name (as best you know it). - Their insurance company (if you were able to get it). - Their license plate number. Also, did the police come to the scene, and if so, is there a police report number? And looking at your car right now: does it look safe to drive, or is it badly damaged (for example, leaking fluids, wheels pushed in, lights smashed)? If it’s not safe to drive, I can arrange a tow for you.
CUSTOMERK.
CUSTOMERSo the accident. Happened about 20 minutes ago. I was stopped at a red light at SW fifth Avenue and Alder Street. Another car hit the back of my car.
CUSTOMERThe other driver is Kevin Lund.
CUSTOMERHe is insured with Pacific Shore.
CUSTOMERThe license plate number is seven.
CUSTOMERK.
CUSTOMERX like X-ray. R.
CUSTOMER214.
CUSTOMERPolice did not come.
CUSTOMERThere is no report number.
CUSTOMERThe back bumper is crushed and the rear wheel rubs.
CUSTOMERI cannot drive it.
CUSTOMERSo please arrange a tow truck.
AGENTI’ve started your claim and arranged the tow. Let me read back the key details to be sure I have them right: - Policy number: NG4471902. - Other driver’s license plate: 7KXR214. Please correct me if either of those is wrong. Here’s what we’ve done and what happens next: 1. Claim opened. - Your claim number is NGM-904417. - Your assigned adjuster is Priya Shah; she will contact you within one business day to go over coverage, repairs, and next steps. 2. Accident details on file. - Loss time: about 20 minutes ago. - Location: SW 5th Avenue and Alder Street, near the café. - Description: You were stopped at a red light when you were rear‑ended. - Other driver: Kevin Lund, insured with Pacific Shore, plate 7KXR214. - Police: Did not attend; no report number. - Injuries: None reported. 3. Tow truck arranged. - Tow company: Eastside Towing. - ETA: about 45 minutes from now. - Destination: Northgate partner repair shop at 9th and Morrison. While you wait: - Remove any valuables from the car before it’s towed. - When the tow arrives, you can tell the driver you’re with Northgate Mutual and give them your claim number NGM-904417. If you’d like, I can also check your coverage details (collision, rental, towing) and explain what’s likely covered, or we can keep it simple and you can wait to speak with Priya.
CUSTOMERPlease check my coverage details now.
CUSTOMERI want to know how much I have to pay.
CUSTOMERWill I get a rental car? And is my premium going to go
AGENTHere’s what your policy shows for this claim: 1) What you may have to pay - Your collision deductible is $500. - If your damage is repairable and the repair cost is more than $500, you would generally pay the first $500 to the shop, and we’d handle the rest (up to the value of the car), subject to final claim review. - If the car is a total loss, the deductible is taken from the settlement amount. Because you were rear‑ended, we will also look at collecting from the other driver’s insurance (Pacific Shore). If they accept fault and pay, that can affect whether your deductible ultimately applies or is reimbursed—but that depends on how their investigation goes. 2) Rental car coverage - You have rental reimbursement coverage. - Limit: up to $40 per day, for up to 30 days. - It starts from today (the day the claim is opened), as long as the car is not safely drivable or is in the shop for covered repairs. Your adjuster, Priya, can help set up the rental through one of our partner agencies, or you can rent on your own and submit receipts up to the $40/day limit. 3) Towing coverage - Towing is covered up to 25 miles. - The tow we just arranged to the partner shop at 9th and Morrison will fall under that towing coverage, so you should not have out‑of‑pocket tow costs within that limit. 4) Will your premium go up? - I can’t see or predict premium changes or renewal pricing from here. - A rear‑end where you’re stopped is often treated as “not at fault,” but the final decision on fault and any impact on premium is made by our underwriting team after the claim is fully investigated. - Your adjuster can tell you how the loss is being coded (at‑fault vs. not‑at‑fault), but even then they can’t guarantee what will happen to the premium at renewal. To recap key numbers: - Policy number: NG4471902 - Claim number: NGM-904417 - Collision deductible: $500 - Rental: up to $40/day, 30 days - Towing: covered up to 25 miles Do you want me to note a preferred rental company on your claim, or would you like Priya to call you first and walk you through the rental setup?
CUSTOMERup? But he hit
CUSTOMERme. So I should not have to pay that.
AGENTYou’re right to be thinking about that,
CUSTOMERDeductible.
AGENTIn a rear‑end
CUSTOMERThat.
AGENTYou’re absolutely right that he hit you from behind, and that’s a strong sign he’s at fault. The part that’s frustrating is how the deductible works in practice. Here’s how it usually goes in a rear‑end like yours: 1) Your deductible right now. - On your policy, the collision deductible is still $500. - If we repair your car under your own policy, we have to apply that $500 up front. That’s a contract rule we can’t override at intake. 2) How you might not end up paying it in the long run. There are two common paths: - Path A: The other insurer (Pacific Shore) accepts fault quickly. - You could choose to go through Pacific Shore directly for repairs. - In that case, you usually would not pay your Northgate deductible at all, because you’re using their policy, not yours. - The trade‑off is timing and convenience—sometimes it takes them a bit to accept liability. - Path B: You go through Northgate first, then we collect from them later. - We handle your repairs now, apply your $500 deductible, and get you moving. - Then we pursue Pacific Shore. - If they accept fault and reimburse us, we can reimburse you for the deductible (in whole or in part, depending on what they pay). So: - I can’t remove the $500 deductible from your policy. - But because you were rear‑ended, there is a real possibility you won’t be out that $500 in the end, if Pacific Shore accepts liability and pays. Your adjuster, Priya, will: - Talk through whether you want to go through Northgate or directly through Pacific Shore. - Explain exactly how deductible reimbursement would work in your case.
AGENTWould you prefer I note on your claim that you’re very concerned about not paying the deductible because you were clearly rear‑ended, so Priya addresses that first when she calls?