AI Sales Call FAQ 2026: 25 Questions on Pricing, Compliance, and ROI

Gong charges a platform fee of approximately $5,000-$7,000 per year plus a per-user seat fee of $1,400-$1,600 per user per year. A 10-seat team typically costs $19,000-$23,000 in year one. Gong does not publish prices on its website; all pricing is via custom quote. Vendr and Vertice data suggests median contract values of $40,000-$60,000 for mid-market teams.

ZoomInfo acquired Chorus.ai in 2021 and has since bundled Chorus functionality into the ZoomInfo Sales platform. As of 2026, Chorus is not sold as a standalone product - you must purchase a ZoomInfo licence. ZoomInfo pricing depends heavily on seat count, data volume, and add-ons. A combined Chorus + ZoomInfo package for a 10-seat team typically runs $25,000-$50,000 per year. If you already have ZoomInfo licences, Chorus adds minimal marginal cost.

Clari acquired Wingman (the AI sales coaching tool) in 2022 and rebranded it as Clari Copilot. Clari Copilot provides real-time conversation coaching, post-call summaries, and call library features similar to Gong and Chorus. It can be purchased as a standalone product ($500-$700/user/year) or as part of the full Clari Revenue Platform. The standalone Clari Copilot is typically 30-50% cheaper than Gong on a per-seat basis, at the cost of less mature revenue intelligence features.

All three record and transcribe sales calls, provide AI-generated summaries, and offer a call library. Where they differ: Gong has the most mature deal intelligence and revenue forecasting integration. Chorus (bundled with ZoomInfo) gives you data enrichment in the same platform. Clari Copilot focuses on real-time coaching during the call and integrates tightly with Clari's forecasting module. Gong is typically the most expensive; Clari Copilot is often the cheapest standalone option.

Revenue intelligence tools like Gong and Chorus are primarily post-meeting tools - they analyse calls that have already happened. They do not generate pipeline. For pipeline generation, you need AI SDR (email outbound) or voice AI (phone outreach) tools. Gong's engagement features and Chorus's ZoomInfo integration provide some prospecting context, but the core product is call analysis, not prospecting.

An AI SDR (Sales Development Representative) is software that automates the prospecting and outreach functions traditionally performed by a junior sales hire. It identifies target accounts and contacts, writes personalised outreach emails using AI, manages follow-up sequences, handles objections in email threads, and books meetings directly to an AE's calendar. Current AI SDR platforms (11x Alice, Artisan Ava) operate autonomously - they do not require a human to approve each email before sending.

11x does not publish pricing on its website. Vendr's public marketplace listing, read 23 September 2026, shows a median contract value of $55,050 a year, a low of $38,250, a high of $89,850, a named Advanced tier at $45,000 a year carrying 3,500 new prospects a month, and a redline threshold estimate of $30,000. Buyer reports put retail Alice around $5,000/month. Always negotiate: Vendr's redline suggests there is room down to about $30,000 but little below it.

Artisan publishes nothing. As of 23 September 2026 artisan.co/pricing lists three plans sized by leads contacted per month - Team at about 2,500, Scale at about 6,000, and Enterprise - each marked 'Pricing scoped on your plan'. The self-serve credit tiers Artisan published earlier in 2026 have been withdrawn. The only public anchors are Vendr's: a median Artisan AI contract of $30,000 a year across a $18,300 to $80,371 range, and a listed Ava: Accelerate tier at $12,000 a year for 12,000 leads contacted annually. Note that leads contacted is not the same as leads sourced, and the AI dialer is a per-seat add-on on top of the plan, so ask for both lines separately in writing.

Both are agent-class AI SDR platforms for email outbound (each runs within human-configured sequences rather than setting its own strategy). Alice (11x) is typically positioned at higher-volume enterprise deployments and is priced higher. Ava (Artisan) has a more transparent pricing structure, a lower entry price, and has released more public product updates. Alice also offers Julian (an AI voice agent for inbound qualification) as a companion product. On G2, Artisan has a lower average rating (4.0 vs 4.4 for 11x, as of August 2026) - likely reflecting the difficulty of AI SDR adoption rather than product quality alone.

Regie.ai is better described as an AI SEP (Sales Engagement Platform) rather than a pure AI SDR. Unlike 11x and Artisan, Regie does not operate as a fully autonomous agent. It requires human reps to approve and send emails, rather than sending autonomously. Regie's AI primarily handles content generation, sequence management, and data package curation. It is closer to an AI-augmented tool for your human SDRs than a replacement for them. Pricing: $3,000-$7,000/month platform fee plus $59-$89/user/month.

Industry benchmarks and vendor case studies (which are self-selected and optimistic) suggest 10-30 meetings per month. Realistic steady-state targets after 90-day tuning: 10-20 meetings/month for most deployments. Deployments in niche verticals or with complex products typically see 5-15 meetings/month. First 90 days are typically below steady state as sequences are tuned. Do not evaluate ROI on first-90-days data.

All four are voice AI infrastructure platforms - they provide the APIs and orchestration layer to build AI phone agents. Retell: a published $0.07-$0.31/min band assembled from four separately metered lines, about $0.149/min for a realistic sales stack, 250-400ms latency, known for ease of setup and a published add-on menu. Vapi: $0.05/min hosting with models passed through at provider cost, $0.082-$0.129/min all-in on Vapi's own calculator, the cheapest of the four and the broadest on model flexibility. Bland: $0.14/min on the free Start tier and $0.12/min on Build with a $299/month platform fee, known for its built-in dialer and high-volume outbound campaigns. Synthflow: no-code interface, but it publishes no per-minute rate at all now, only that enterprise contracts start at $30,000 a year.

Rates read from each vendor's pricing page on 23 September 2026: Vapi $0.082-$0.129/min all-in on its own calculator, built on $0.05/min hosting plus at-cost pass-throughs and free Vapi Telephony. Retell a published $0.07-$0.31/min band, about $0.149/min for a realistic sales stack. Bland $0.14/min on Start with no platform fee, $0.12/min on Build with $299/month; its $499/month Scale tier has been withdrawn. Synthflow publishes no rate at all, only that enterprise contracts start at $30,000 a year. Self-hosted (assemble your own STT + LLM + TTS): $0.04-$0.08/min but requires engineering resources. LLM inference is typically the largest cost component in a self-hosted stack, and on Retell's published list it is the line with the widest range, from $0.0016 to $0.32 a minute.

Neither platform bundles the model into a flat rate. Retell charges $0.055/min for its own voice infrastructure and meters the model separately, from $0.0016/min on the cheapest published option to $0.32 on the most expensive, with recommended-tier models at $0.064-$0.16; text-to-speech is $0.015 or $0.040 and telephony $0.015, which puts a realistic stack at $0.149/min. Vapi charges $0.05/min hosting and passes providers through at cost: Deepgram $0.0095-$0.0099, an OpenAI model $0.0077-$0.0452, an ElevenLabs voice $0.0146-$0.0238, with its own telephony free. That is $0.082-$0.129/min all-in on Vapi's calculator. For a 2-minute call: Vapi $0.16-$0.26, Retell about $0.30 on the mid stack.

Technically yes - platforms like Bland and Retell support outbound calling campaigns at scale. Legally, this requires prior express written consent from every number you dial under FCC Ruling 24-17 (February 2024). Without consent, each call is a potential TCPA violation at $500-$1,500 per call. Outbound appointment confirmation to contacts who submitted a demo request (implied consent) is a materially safer use case than cold outbound.

Latency (the delay between a human speaking and the AI responding) is the primary quality metric for voice AI. Sub-500ms latency feels natural in conversation. 500-800ms latency is noticeable but acceptable for transactional calls. Above 800ms feels robotic and frustrating. Retell achieves 250-400ms with its optimised pipeline. Vapi achieves 300-500ms. Self-hosted stacks with GPT-4o can run 600-1000ms depending on infrastructure. For sales calls where natural conversation rhythm matters, sub-500ms is the target.

Under FCC Ruling 24-17 and general FTC guidance, AI-generated or AI-assisted calls should identify as AI at the start of the interaction. Best practice regardless of jurisdiction: 'Hi, I am an AI assistant calling on behalf of [Company].' This covers TCPA disclosure obligations, satisfies best-practice standards in two-party consent states, and builds trust with prospects who increasingly expect AI disclosure.

No legal requirement exists in the US to disclose that an email was written by AI (as of April 2026). CAN-SPAM requires commercial emails to identify the sending entity, include a physical address, and provide a working unsubscribe mechanism. GDPR requires lawful basis for processing EU prospect data. There is no FTC or SEC rule requiring 'written by AI' disclosure in sales emails at this time.

High if consent disclosure is not enabled. California CIPA requires all-party consent for recording. Gong and Chorus both have automated consent disclosure features (typically an audio beep at the start of recording). These must be explicitly enabled in account settings - they are not always on by default. The Ambriz v Google ruling (N.D. Cal., motion to dismiss denied Feb 2025) confirmed that AI analysis of call content can trigger CIPA obligations independent of traditional recording. Verify your consent disclosure is active before recording California contacts.

TCPA violations are enforced through private rights of action (individual plaintiffs) and FTC/state AG enforcement. Statutory damages: $500/call for negligent violation, $1,500/call for wilful violation. TCPA class actions can aggregate thousands of calls into a single action. Cases typically settle for $0.10-$0.50 per class member call, but plaintiffs' attorneys take these cases on contingency, so any credible TCPA claim gets filed. A 5,000-call campaign without consent: realistic settlement range $500,000-$2,000,000 after legal fees.