customers
All customer stories
Top Voice AI companies are building with Assembly.
resources
Latest Release
Universal 3.5 Pro
Our new flagship speech-to-text model handles real-world audio the way it actually happens. Available for real-time and pre-recorded audio.
resources
Build AI voice agents that handle account inquiries, payment processing, and fraud verification — with 43% better alphanumeric accuracy for account numbers, PII redaction by default, and the compliance infrastructure financial institutions require. Ship end-to-end with our Voice Agent API, or drop Universal-3.5 Pro Realtime into your existing stack.
Account services overview
Caller verified · Account ****4532 · Priority tier
Identity
Verified
PII Redacted
Active
Stated reason for call
Suggested resolution
Financial services callers navigate 4-5 IVR menus before reaching a human — if they don't hang up first. Account numbers get misheard, transactions fail identity verification, and every misrouted call is a compliance exposure. Modern voice agents — built on accurate streaming STT with 43% better alphanumeric recognition, a managed LLM, and PII redaction — resolve account inquiries faster, capture sensitive data correctly, and maintain an auditable transcript of every interaction.
Better alphanumeric accuracy — account numbers, routing numbers, confirmation codes.
P50 median streaming latency for Universal-3.5 Pro Realtime.
PCI DSS v4.0, SOC 2 Type 2, and ISO 27001:2022 certified.
Audio processed daily in production.
Two ways to build
Ship a working voice agent in an afternoon, or drop industry-leading STT into the platform you already run — both with PII redaction and compliance infrastructure built in.
Our proprietary voice stack via one WebSocket. Build account inquiry agents, payment processors, and fraud verification flows with zero infra to manage.
Best for
Account inquiries, balance checks, payment processing43% better alphanumeric accuracy for account numbersTeams shipping fast — working agent in an afternoonPCI DSS v4.0 and SOC 2 Type 2 certifiedFree tier available · No credit card required
The STT layer for your financial services voice platform. Works natively with your preferred orchestrator, compliance tooling, and core banking integrations.
Best for
Teams using LiveKit, Pipecat, or Twilio as their orchestratorCascading architectures (STT → LLM → TTS)PII redaction and audio de-identification built inComplex core banking and payment gateway integrationsHigh-scale deployments where margin and full control matterNo concurrency caps · Autoscaling included
Voice Agent API — recommended
# Voice Agent API: financial services account inquiry agent
import asyncio, json, websockets
API_KEY = "YOUR_API_KEY"
async def run_agent():
async with websockets.connect(
"wss://agents.assemblyai.com/v1/ws",
additional_headers={"Authorization": f"Bearer {API_KEY}"},
) as ws:
await ws.send(json.dumps({
"type": "session.update",
"session": {
"system_prompt": (
"You are an account services agent for First National Bank. "
"Verify caller identity before discussing account details. "
"Never read full account or card numbers aloud."
),
"greeting": "Hi, this is First National — how can I help today?",
"input": {"keyterms": ["First National", "rewards checking",
"premium savings", "Reg E", "provisional credit"]},
"output": {"voice": "ivy"},
},
}))
# Stream audio in, get audio + transcript back
async for msg in ws:
handle(json.loads(msg)) # transcript.user, reply.audio, tool.call, ...
Universal-3.5 Pro Realtime + LiveKit — BYO stack
# LiveKit + AssemblyAI STT in a financial services agent pipeline
from livekit.agents import Agent, AgentSession, TurnHandlingOptions
from livekit.plugins import assemblyai, cartesia, openai, silero
class FinancialAgent(Agent):
def __init__(self):
super().__init__(
instructions=(
"You are an account services agent for First National Bank. "
"Verify identity before disclosing account information. Be concise."
),
)
async def entrypoint(ctx):
session = AgentSession(
stt=assemblyai.STT(
model="u3-rt-pro",
min_turn_silence=100,
max_turn_silence=3000, # callers pause to look up accounts / read numbers
vad_threshold=0.3,
keyterms_prompt=["First National", "rewards checking",
"premium savings", "Reg E", "provisional credit"],
),
llm=openai.LLM(model="gpt-4o"),
tts=cartesia.TTS(),
vad=silero.VAD.load(activation_threshold=0.3),
turn_handling=TurnHandlingOptions(
turn_detection="stt",
endpointing={"min_delay": 0}, # avoid additive latency in STT-driven turns
),
)
await session.start(room=ctx.room, agent=FinancialAgent())
43% better accuracy on account numbers, routing numbers, confirmation codes, and email addresses spoken over the phone — the data that matters most in financial services.
PCI DSS v4.0, SOC 2 Type 2, and ISO 27001:2022 certified. PII redaction masks card numbers, SSNs, account numbers, and DOBs before transcripts reach downstream systems.
Sentiment analysis detects frustrated callers for priority routing. Entity detection captures transaction amounts and dates. Topic detection flags escalation triggers in real time.
The transcription accuracy, reliability, and speed of AssemblyAI's API have greatly enhanced our operations.
Raj Shankar, SVP Product — Calabrio
Read more We require a leading edge speech-to-text provider that can meet our specialized needs: fast, accurate, targeted, and multilingual.
Super
.png)
Read more