Inbound vs outbound agents
In one sentence
An inbound voice agent answers contacts that come to the business while an outbound agent initiates contact, and although they share the same technology they differ so much in regulation, deliverability, intent and economics that they are substantially different businesses.
Definition
An inbound agent answers people who contact the business. An outbound agent rings people who did not.
They run on the same technology, and they are completely different propositions the moment you look past that.
This is the single most consequential split in the voice agent category. Which side an agent is on determines its regulatory exposure, its deliverability risk, its economics and its conversation design, all at once.
Regulatory exposure
- Inbound: the person initiated contact, so contact consent is not at issue. Recording and processing consent still apply.
- Outbound: contact consent is the central question. In the US, TCPA requires prior express written consent for marketing calls to mobiles using artificial or prerecorded voices, and the FCC confirmed in February 2024 that AI-generated voices fall within that restriction. Statutory damages accrue per call.
- This asymmetry is enormous, and it is the primary reason inbound is a simpler business.
Deliverability and intent
- Inbound: no deliverability question, the call is already happening. The person wants something, so intent is high and the job is to satisfy it.
- Outbound: spam labeling and attestation decide whether calls are answered at all, and remediation is slow and opaque. The person was doing something else, so the first job is to earn the right to continue.
- That produces completely different conversation design: an inbound agent can ask clarifying questions, while an outbound agent has seconds before the call ends.
Economics
- Inbound: value per conversation is high because intent is high, and volume is bounded by demand.
- Outbound: value per conversation is low because most calls are unwanted, and volume is bounded by consent and by deliverability.
- Inbound economics are structurally better and require no demand generation.
The website position
- On-site voice is inbound in the strongest sense: the visitor arrived, stayed, and chose to speak.
- No contact consent question, no deliverability, no attestation, no number.
- This is the least regulated and highest-intent position in the entire category, and it is worth stating explicitly, because buyers researching voice agents meet the outbound literature first.
Common misconception
That inbound and outbound agents are the same product pointed in different directions. The regulatory, economic and design differences make them substantially different businesses.
Why it matters commercially
Outbound carries the category legal risk and its deliverability problems. Inbound carries neither. For a buyer, working out which one is being sold determines what questions to ask before anything else.
In voice specifically
The split is specifically a voice phenomenon. The TCPA restriction that drives it is written around artificial or prerecorded voice on phone calls, and spam labeling attaches to a phone number, so both hang on the voice channel in a way a text message does not carry over.
Where AsqVox fits
The Orb is inbound in the strongest available sense: the visitor is already on the site and initiates the conversation. None of the outbound regulatory or deliverability apparatus, no contact consent question, no attestation, no number to protect from spam labeling, applies to it.
Visual
The same technology, two different businesses
Inbound and outbound share the stack and almost nothing else. On-site voice is the strongest inbound position of all.
Statistics
Every figure carries its source and year. Vendor numbers are labelled as vendor numbers, and where no reliable figure exists this page says so rather than borrowing one.
The TCPA restricts calls using artificial or prerecorded voices, with prior express written consent generally required for marketing calls to mobile numbers.
Prior express written consent requiredindustry rangeUS Telephone Consumer Protection Act, 2026 - A statutory requirement, not a measurement. Anchor a citation on the TCPA itself and current FCC guidance, not on the review year recorded here.
In February 2024 the FCC issued a declaratory ruling confirming that AI-generated voices fall within the artificial or prerecorded voice restriction.
AI voices covered, February 2024industry rangeFCC declaratory ruling, 2024 - A dated regulatory ruling, and the reason the whole TCPA question now attaches squarely to AI voice agents. Cite it by its date.
Statutory damages are commonly cited as USD 500 per violation, rising to up to USD 1,500 for willful or knowing violations, and they accrue per call.
USD 500 per call, up to USD 1,500 willfulindustry rangeUS Telephone Consumer Protection Act statutory damages, 2026 - Per-call accrual is what makes this large at campaign scale. The figures are the statutory amounts, not an estimate of any particular settlement.
STIR/SHAKEN attestation operates at three levels, with full attestation generally requiring a direct carrier relationship and verified number ownership.
Three attestation levelsindustry rangeSTIR/SHAKEN caller authentication framework, 2026 - The framework fact behind outbound deliverability. Full attestation is a gate an outbound program has to clear and an on-site agent never encounters.
Spam labeling is applied by proprietary analytics with no published thresholds and limited appeal.
Opaque proprietary labelingindustry rangeCarrier and analytics spam-labeling practice, 2026 - The opacity is the point: an outbound program can be labeled with no stated reason and slow recourse, which is why answer rates are unpredictable.
A human-handled call costs roughly USD 7 to USD 12 against roughly USD 0.40 for an agent-handled call, and this applies to both directions; the difference between inbound and outbound is in value per conversation, not in cost.
USD 7 to 12 human vs USD 0.40 agentindustry rangeCommonly cited industry per-call cost comparison, 2026 - A widely repeated cost band rather than an audited figure. Use it for the order-of-magnitude gap, and note that it says nothing about which direction converts better.
There is no reliable published comparison of conversion or value per conversation between inbound and outbound voice agents.
-no reliable figureThe argument that inbound economics are better rests on intent and regulatory structure, not on a head-to-head study. Do not attach a conversion multiple you cannot source.
There is no independently verified published data on the answer-rate impact of spam labeling.
-no reliable figureThe effect is real and widely reported, but the specific answer-rate figures in circulation are vendor or anecdotal. Describe the direction, not a percentage.
Examples
In practice
A company builds one voice agent and points it at both inbound enquiries and outbound follow-up. The inbound deployment performs well. The outbound deployment is halted by legal review over consent standards and, separately, sees answer rates collapse from spam labeling within three weeks. The technology was identical; the businesses were not.
The everyday version
An inbound agent answers people who contacted you. An outbound agent rings people who did not. They use the same technology and they are completely different propositions. The outbound one comes with serious legal rules and the problem that nobody answers unknown numbers any more. A voice agent on your own website has neither issue, because the customer came to you.
Usage
Who says it
- Contact center practitioners and voice AI vendors, as a fundamental product distinction.
- Legal and compliance teams, who treat the two entirely separately.
Where it turns up
- At the top of an RFP as a scope definition, determining which later requirements apply.
- Outbound scope drags in consent management, do-not-call handling, attestation, pacing and calling-time controls that inbound does not.
Common misuse
- Treating the two as one product category.
- Applying outbound compliance concerns to inbound or on-site voice, or failing to note their absence when it is an advantage.
- Assuming outbound conversation design transfers to inbound, where the intent situation is reversed.
Questions people ask
What is the difference between an inbound and an outbound voice agent?
An inbound agent answers contacts that come to the business; an outbound agent initiates contact. They share the same technology, but they differ in regulation, deliverability, intent and economics enough to be substantially different businesses. Inbound has high intent and no contact consent question; outbound carries TCPA exposure and spam-labeling risk.
Why is outbound voice riskier than inbound?
Outbound carries the category legal and deliverability problems. Contact consent is central, TCPA requires prior express written consent for marketing calls to mobiles and attaches statutory damages per call, and spam labeling can cut answer rates with no stated reason and slow appeal. Inbound has none of this, because the person initiated the contact.
Does TCPA apply to AI-generated voices?
Yes for the US outbound case. The TCPA restricts calls using artificial or prerecorded voices, and in February 2024 the FCC issued a declaratory ruling confirming AI-generated voices fall within that restriction. Statutory damages are commonly cited as USD 500 per violation, rising to up to USD 1,500 for willful or knowing violations, and they accrue per call.
Is a website voice agent inbound or outbound?
Inbound, in the strongest sense available. The visitor arrived on the site, stayed, and chose to speak, so there is no contact consent question, no deliverability problem, no attestation and no number to protect from spam labeling. It is the least regulated and highest-intent position in the entire voice agent category.
Last reviewed 4 August 2026. Written and reviewed by Dhruv Dholakia, founder of AsqVox.