DEV Community: Alfredo Romero

Smallest.ai Just Raised $13M for Voice Quality. Here's Why Agencies Should Care.

Alfredo Romero — Thu, 06 Aug 2026 14:51:19 +0000

On July 31st, TechCrunch reported that Smallest.ai raised $13M to build ultra-fast voice AI models that sound genuinely human. The bet? Architecture beats size. Smaller, specialized models tuned for speed and naturalness outperform larger, slower alternatives.

Why This Matters for AI Voice Agencies

For the last 18 months, the narrative in voice AI has been "bigger model = better results." Vapi bets on heavy inference. Retell pitches GPT-4 integrations. The implicit message: throw compute at the problem.

Smallest.ai just challenged that. Their bet is that you can build voice that actually sounds human with a 4B parameter model optimized for latency and naturalness, not sheer scale. That's a huge shift.

Here's why it matters to you as an agency owner: If smaller, faster models deliver the same (or better) perceived quality, then the platforms that own the architecture, not the model, will win. Which means integration matters more than raw model size. Margin composition shifts. And the duct-tape stack (Retell + GoHighLevel + Zapier + Twilio) starts to look even worse.

Your clients expect voice that doesn't sound robotic. If Smallest.ai delivers that at scale, and your platform can't wire it in without three weeks and a developer, you're already losing.

What We're Doing at Hermes About It

We don't wake up each morning wondering which model to bet on. We built Hermes so you never have to make that choice. Our platform integrates Smallest.ai, GPT-Live-1, Retell, Bland, and the models that come next. You pick the model. We handle the integration.

One platform. Your brand. Your margins. You don't lose 80% of profit to tool sprawl, and you don't wait three weeks for a developer to wire in the latest model breakthrough.

That's what "by builders, for builders" actually means: We solve the engineering, so you can focus on clients.

Action Steps for Agencies Right Now

Audit your current voice model quality. If clients still comment that the agent "sounds a bit robotic," this is your wake-up call. Smallest.ai's funding signals that voice quality is now a competitive differentiator, not a feature.
Check if your platform can swap models without code. If switching from Retell to Smallest.ai means a 2-week engineering sprint, you're dead in the water. Ask your platform: Can I change models via dashboard?
Don't assume "newer = better." Smallest.ai's strength is speed and naturalness at 4B params. GPT-Live-1 is full-duplex. They solve different problems. The platform that lets you pick per-client is the one that survives.
Measure and document voice quality for each client. Call listen rates, transfer-to-human rates, hang-up time. If you can't show your client that the new model improved their metrics, it's not worth the switch cost.
Lock in your margins now. The voice quality wars are starting. Platforms that own the full stack and let you white-label will keep 80%+ margin. Platforms that bolt on APIs will bleed margin to integrations and support. Pick your side.

FAQ

Should I switch my voice AI models because Smallest.ai raised money?

Not necessarily. Funding is validation that the architecture is solid, but it doesn't mean Smallest.ai is the right model for every use case. Some clients need full-duplex (OpenAI GPT-Live-1). Others need pure speed (Bland). The real question: Can your platform let you pick per-client without engineering overhead? If yes, stay flexible. If no, you're locked in.

Why does model architecture matter more than model size?

Smaller models optimized for specific tasks (latency, naturalness, throughput) often outperform larger models that are overspecialized. Smallest.ai's bet is that a 4B parameter model tuned for voice beats a 175B model that's tuned for text. That's smart engineering. The platform that can wire in these kinds of models quickly is the one that wins with clients.

How does this affect my margins as an agency owner?

If you're on Retell plus GoHighLevel plus Zapier plus Stripe, each tool takes a cut. As voice quality becomes table stakes and platforms fragment into specialists (Smallest.ai for quality, Bland for cost, etc.), you'll need a platform that wires them all without margin bleed. Hermes: one platform, 80%+ margins. Duct-tape: five platforms, margin death by a thousand cuts.

The Bottom Line

Smallest.ai's $13M funding validates what we've known for a while: the voice quality wars are about architecture, not just model scale. That means platforms that own the full stack and let you swap models without code will dominate. Platforms that bolt APIs together will fragment.

If you're still managing five tools to run one voice agent, now is the time to consolidate. Your clients expect world-class voice. Your margins expect a single platform. Build with Hermes. Ship in 72 hours.

Ready to build voice agents on a platform that actually scales?

Start with Hermes for $149/month. First agent live in 72 hours. Compare us to Synthflow, Retell, and the others. Join the beta.

Originally published at buildwithhermes.com/blog/smallest-ai-13m-voice-quality-what-agencies-should-do-2026-08-05.

OpenAI's GPT-Live-1 vs. Hermes: Which Platform Should Agencies Choose?

Alfredo Romero — Wed, 05 Aug 2026 14:33:44 +0000

OpenAI just released GPT-Live-1 with full-duplex voice capabilities (TechCrunch, July 8). The models can speak and listen at the same time. Callers can interrupt naturally. Desktop app landed July 24. This changes the conversation quality expectations for every AI voice agency starting today.

Why This Matters for AI Voice Agencies

Full-duplex voice, the ability to speak and listen simultaneously, sounds like a technical feature. It's actually a positioning reset for the entire agency market.

Until now, voice AI agents worked in turns. Listen, process, respond. Clean. Predictable. But unnatural. Real conversations overlap. People interrupt. They ask clarifying questions mid-pitch. Humans sense the lag and disengage.

GPT-Live-1 closes that gap. Agencies using it will deliver conversations that feel like talking to a person, not a bot waiting for its turn. That's a massive jump in call quality. And your clients will expect that quality immediately. If you're using older models, you just got outcompeted on the call floor.

Desktop adoption is the real signal. When consumers interact with GPT-Live-1 on their MacBooks, they develop expectations. They bring those expectations to business calls. By Q4, "natural voice" won't be a feature. It will be table stakes.

The Model vs. Platform Trade-off

Here's the decision you're facing as an agency:

Option 1: Go Direct to OpenAI's API

You get the latest model immediately. You control the voice, tone, and personality. You pay $0.13 per minute for ChatGPT inference. No middleman. No wrapper overhead.

You also get: no CRM. No campaign orchestration. No automated follow-ups. No contact management. No compliance guardrails. No white-label billing. You'll stitch together Retell or VAPI for the voice layer, GoHighLevel for the CRM, Zapier for automation, and Stripe for billing. You're building a platform piece by piece, paying a developer $8K-$15K per month to glue it all together. You're managing vendor risk across 7 systems. When Retell's API goes down, your agency goes down.

Option 2: Use a Purpose-Built Agency Platform

Platforms like Hermes abstract away model selection. We integrate the latest models from OpenAI, Anthropic, and others. Your agencies always have access to the cutting edge without managing API keys or integrations. You deploy campaigns with built-in CRM, contact management, compliance automation, white-label billing, and analytics. You own the client relationship end-to-end, under your brand.

The trade-off: slightly higher per-minute cost ($0.24/min on Hermes, but that includes the platform, not just the voice), but you get the entire operating layer. You're not managing developers. You're not managing vendor relationships. You're running an agency.

What We're Doing at Hermes

GPT-Live-1 is already available to Hermes agencies. You can select it from the model dropdown today. No migration. No rebuilding. We'll continue to layer in newer models as OpenAI, Anthropic, and others release them. Your job is unchanged: build the campaign, select the model, launch. We handle the infrastructure.

More importantly, we've built compliance and white-label tools around model selection. TCPA safeguards automatically adjust based on the model's output speed and natural-sounding capability. White-label billing stays predictable even as models improve. Agencies don't have to choose between "latest model" and "margin protection". Hermes does that for you.

That's why we price at $149-$699/month with transparent per-minute overage at $0.24. You know exactly what each client costs. No surprise white-label kits hiding extra fees. No usage-based pricing that makes your best month cost 3x your budget.

Action Steps for Your Agency

1. Audit Your Current Model Mix
Log into your platform (Hermes, Synthflow, Retell, wherever you are) and check which models your active campaigns are using. Document model age and call quality feedback from clients. You'll spot where the newest model delivers measurable improvement vs. where it doesn't matter.

2. Run a Test Campaign with GPT-Live-1
Take your highest-value use case (probably inbound sales or appointment setting) and spin up a parallel campaign using GPT-Live-1. Run it for 48 hours with a small list. Track call completion rate, client callback rate, and sentiment. Compare to your previous model baseline.

3. Document the Client Conversation Shift
Listen to 5-10 calls from the test campaign. Write down where callers interrupted, where the agent responded naturally, where lag was eliminated. This is your sales data for why the newer model matters for your clients.

4. Decide: Build or Buy
If you're comfortable managing API keys, integrations, and developer time, go direct to OpenAI. If you want a complete operating system for your agency (CRM, billing, compliance, model abstraction, white-label), move to a purpose-built platform. Either way, get GPT-Live-1 into your production campaigns by August 15.

5. Communicate the Upgrade to Your Clients
Email your top 10 clients: "We're upgrading your agent's voice model this week to the latest from OpenAI. Expect more natural conversations and faster call times. No action needed on your end." Position it as a value add, not a price bump. Set expectations before they hear the quality jump on a live call.

FAQs

Does GPT-Live-1 cost more than older OpenAI voice models?

Not from OpenAI's side. Inference pricing stays at $0.13/minute. The difference is in adoption and industry shift. Every competitor will upgrade to GPT-Live-1 this quarter. If you don't, your call quality will lag visibly by Q4. The cost of staying on older models is margin loss, not extra fees.

Can I mix models? Use GPT-Live-1 for some campaigns and older models for others?

Yes. Test on new campaigns before migrating existing high-volume ones. Use GPT-Live-1 for high-touch clients and sales work (where natural conversation matters most). Keep older models for compliance-only campaigns if they're already working. Most agencies will standardize on GPT-Live-1 within 60 days once call quality improves.

Will Hermes increase prices because models got better?

No. We price at $149-$699/month plus $0.24/minute overage. That's locked. Model improvements are a feature we deliver, not a reason to raise rates. That's why agencies choose platforms over API-direct: predictable costs, predictable margins, no surprise "white-label kits" that hide extra fees.

What if OpenAI's next model comes out next month?

We'll integrate it. You'll see it in the model dropdown. You can test it, run campaigns with it, or stick with GPT-Live-1. No forced migrations. No compatibility breaks. The whole point of using a platform is that you never have to manage model lifecycle again.

Should I worry about Synthflow or other wrappers changing prices again?

Yes. Wrapper platforms (UI layers on top of Retell/VAPI) inherit upstream cost pressure. When the base model gets better, they often bump pricing to capture margin. Synthflow moved from tiered ($29-$1,400/mo) to usage-based plus $2K/mo white-label kits this year. Voicerr went from $28 to $199-$299/mo. If you're on a wrapper, expect prices to move. Hermes locks pricing by design: $149-$699 plus overage. You know your margin math for the year.

The Bottom Line

GPT-Live-1 is live. Full-duplex voice is now table stakes for AI voice agencies. You have two paths:

Path 1: Manage model selection, integrations, compliance, billing, and white-label yourself. You'll be 60 days behind the latest models because you're building infrastructure, not scaling agencies.

Path 2: Let a platform handle model abstraction, compliance, billing, and CRM. You focus on campaigns and clients. You stay on the cutting edge without building.

Either way, migrate your campaigns to GPT-Live-1 by mid-August. Your clients expect it. Your competition is already moving.

By builders, for builders. Hermes is the operating platform for AI voice agencies. From $149/month.

Originally published at buildwithhermes.com/blog/gpt-live-1-agencies-2026-07-30.

Rime $24M + Fish Audio $21M ARR: Why Agencies Can't Win on Price Alone

Alfredo Romero — Thu, 30 Jul 2026 14:33:41 +0000

This week: Rime raised $24M Series A for enterprise voice calls. Fish Audio hit $21M ARR with 8M active users in a $52M seed round. Grok Voice Agent Builder shipped at $0.05/minute. The voice AI funding race is accelerating fast. But for AI agencies, the real battle isn't funding. It's margin.

Why This Matters for AI Voice Agencies

Every funding announcement signals the same thing: the voice AI market is validating. Enterprises are adopting. Creators are adopting. SMBs will follow. That's the TAM expansion cycle agencies should be riding.

But watch what's happening in the funding:

Rime: $24M targeting enterprise customer calls (BPO/contact center layer). Translation: going upmarket, away from solo builders and small agencies.
Fish Audio: $21M ARR from 8M users. They're monetizing synthetic voice cloning. Translation: voices are commoditizing fast. Survival is throughput, not quality.
Grok: No-code voice agent builder, $0.05/min. Undercuts every existing player on pure price. Translation: pricing wars just got serious.

The trap is clear: if agencies race vendors on per-minute pricing, they lose. A $0.05/min provider with 10M customers has leverage you don't. Your margins get squeezed. You chase volume to stay profitable. You burn out.

The flip side: agencies who own the entire stack, voice plus CRM plus campaign orchestration plus client management, win on control, not price. They build moats. They own the relationship with their client's customers. They keep 80%+ margin.

What We're Doing at Hermes About It

Hermes is built differently. We're not racing vendors on per-minute costs. Instead, we're building the operating platform agencies need to scale clients profitably without vendor lock-in.

Here's the model:

One flat-fee platform under your brand. Your clients never see the word Hermes. They see your branding, your logo, your domain. You stay the hero.
Voice + CRM + Campaigns bundled. No duct-taping Retell + GoHighLevel + Zapier + Stripe. One platform means faster deployments, fewer integration headaches, higher retention.
Transparent cost structure. $149 to $699/month depending on scale. Voice minutes included. You know your margins. You price confidently to your clients. You keep 80%+ of what they pay.
White-label everything. Agencies brand it. You own the customer relationship. When Grok drops pricing again or Rime raises it, your clients stay with you because you're the platform they trust.

While others are racing to $0.05/min and eroding their margins, we're giving agencies the infrastructure to keep theirs intact and grow.

Action Steps for Agencies Affected

1. Audit your current stack. Add up what you're spending on Retell/VAPI, GoHighLevel, Zapier, Stripe, and developer time to glue it all together. Most agencies spend $8,000 to $15,000/month before their first dollar of revenue.

2. Calculate your true margins. Take your client fees, subtract vendor costs and internal labor. If you're below 60%, you're racing to the bottom. If you're below 40%, you're working for the infrastructure.

3. Map your white-label requirements. Can your current stack stay under your brand all the way? If clients see Retell, VAPI, GHL, or Stripe logos, you're not white-label. You're a reseller. Resellers commoditize faster.

4. Evaluate platform consolidation. The agencies winning 2026 to 2027 are consolidating. They're moving from "tools" to "platform." One login, one brand, one margin story. That's the moat.

5. Start with one new client on a platform, not a stack. Deploy them on something integrated. Compare time-to-first-call, margin, and renewal rate. You'll feel the difference immediately.

The Bottom Line

Rime, Fish Audio, and Grok are validating that voice AI is real. The market is growing. Agencies who move now own the next wave of SMB adoption. But they win by controlling margin, not chasing per-minute rates. Build on a platform you brand. Keep the economics clean. Scale sustainably.

That's how you survive when a $0.05/min competitor ships tomorrow.

Originally published at buildwithhermes.com/blog/rime-fish-audio-grok-2026-07-29.

Synthflow Moves Upmarket. Hermes Fills the Gap for Agencies.

Alfredo Romero — Wed, 29 Jul 2026 14:59:26 +0000

In June 2026, Synthflow discontinued its tiered SMB plans and moved enterprise-only, with a $30,000/year ($2,500/month) minimum and per-minute billing starting at $0.11/min. Read the full pricing breakdown here.

If you're an AI voice agency that relied on Synthflow, or you're evaluating platforms right now, here's what just happened and what you should do about it.

Why This Matters for AI Voice Agencies

Synthflow's move is the latest signal in a market consolidation: platforms that win are going upmarket. Vapi evaluated 40+ vendors before Amazon Ring picked them for 100% of inbound calls. Bland AI just raised $50M Series C and is enterprise-only. NEXQA (no, not us, a different vendor) is chasing Fortune 500 integrations.

For agencies with 1-10 clients, this leaves you shopping in an increasingly narrow list.

But the real problem isn't Synthflow's pivot. It's the per-minute pricing model that comes with it.

When you bill per-minute, the economics flip: the vendor wins when calls run long, loop, or fail to transfer. You lose. As one analysis put it:

"Per-minute vendors profit when calls run long and when the bot loops, while you absorb the cost of every failed automation.", Famulor AI Voice Pricing Analysis

For an agency running collections campaigns (long calls, sometimes looped), that's a margin killer. For one running appointment-setting (shorter calls, higher success rate), it works, but you're still paying the vendor when your bot loops twice trying to find a transfer.

Flat-rate pricing inverts the incentive. You care about efficiency. The vendor cares that you hit your campaign goals. Margins stay in your pocket.

What We're Doing at Hermes About It

Hermes is the operating platform for AI voice agencies. One platform, no API keys, no duct-taped stack, no per-minute surprises.

Pricing locked flat-rate:

Starter: $149/month, 300 included minutes, $0.24/min overage
Business: $399/month, 1,000 included minutes, $0.24/min overage
Agency: $699/month, 2,000 included minutes, $0.24/min overage
Pro: Custom for 20+ clients or custom workflows

No component-based billing. No enterprise minimums. No forced upsells to "professional" plans when you add your third client.

Built in: white-label portal under your domain. Native CRM for contacts and deals. Campaign orchestration for outbound + inbound. Transparent usage reporting (no gotchas). A2P compliance built in (DNC checking, consent tracking, submission automation).

Margin math:

You charge your client $2,000/month for a voice campaign.
You run 2,500 minutes/month on Hermes Agency plan ($699).
You pocket $1,301. Margin: 65%.

Same campaign on Synthflow at $0.13/min? 2,500 minutes = $325 base + setup fees. But loops, transfers, retries push it to $450/month. You pocket $1,550, until one campaign runs long (happens 30% of the time) and suddenly overage costs are $300. Margin: 55% on a good month, 40% on a bad one.

Hermes lets you sell with confidence: your costs don't surprise you, and your margins stay yours.

Action Steps for Agencies Affected

If Synthflow was your main platform:

Export your agents. Synthflow's API still works. Pull your agent configs, prompts, and call logs before June 30 (companies always give migration windows). Save them as JSON.
Test on Hermes. Build one existing agent on Hermes in 72 hours. Use our demo agents to see the white-label portal your clients would see. No CC required, fire up a free trial.
Run a shadow campaign. Pick one of your smaller clients. Route 500 calls through Hermes while keeping your Synthflow agent live. Compare cost, call quality, and completion rate for one week. Data beats feelings.
Price the switch. If Hermes lands at $149-399/month vs. Synthflow's $30k/year enterprise minimum, the decision makes itself. Your clients notice too.
Migrate on your timeline. No rush. Agencies that switched in the last 60 days report a 6-week transition per agent, with zero downtime if you run parallel for 2 weeks.

Three FAQs

Q: Does Hermes support my use case? (collections, appointment setting, surveys, etc.)

A: Hermes powers outbound campaigns (collections, sales, surveys), inbound (voicemail-to-text, transfers), and hybrid. NEXQA (our AI backbone) handles complex branching, transfers, and fallbacks. If Synthflow could do it, Hermes can. Check our feature matrix for a side-by-side.

Q: What if I have 50+ clients? Does Hermes scale?

A: Yes. Enterprise plan starts at custom pricing for 20+ clients and custom workflows. Our largest customer manages 180 clients and 500k+ calls/month. Dedicated onboarding, SLA, and priority support included.

Q: Can I white-label Hermes for my own clients?

A: Completely. Every client sees your brand, your domain, your logo, your phone numbers. They never see "Hermes" unless you tell them. Margin stays with you. See how white-label works.

Why This Matters Beyond Pricing

Synthflow's pivot to enterprise reflects a broader market truth: 1,000 indie agencies are building AI voice businesses with DIY stacks. They're duct-taping Retell + GoHighLevel + Zapier + Stripe + Twilio + a developer. That costs $8k-15k/month just to operate. Most never professionalize, they're trapped between "MVP that works" and "enterprise platform that costs $30k+."

Hermes exists in that gap. Built by founders who ran AI voice agencies, not by ex-enterprise sales teams chasing billion-dollar unicorns.

We're not trying to be Vapi (API-first for developers). We're not trying to be Synthflow (enterprise automation). We're building the operating layer agencies actually use every day.

Per-minute billing goes away. Multi-tenant workspace ops stay. White-label stays. Margins stay yours.

Start Today

Your clients need voice AI. The question is: do you run it on a platform that works for you, or a platform that works despite you?

Book a private walk-through of Hermes (20 min, no pressure). We'll port one of your existing agents and show you the real cost math.

Or start free: Create a Hermes workspace and build your first agent in 72 hours. We'll give you a complete onboarding, no secrets, no trial-version crippling.

By builders, for builders. Let's ship.

Alfredo Romero
CEO, Hermes
AI voice agencies deserve a platform built for them, not for enterprises.

Originally published at buildwithhermes.com/blog/synthflow-upmarket-hermes-2026-07-28.

The $0.05/min Race: What Agencies Need to Know About xAI Grok Voice Pricing

Alfredo Romero — Tue, 28 Jul 2026 14:46:20 +0000

xAI launched Grok Voice Agent Builder on July 1, 2026, a no-code platform that lets you build production voice agents in about 2 minutes. The headline price: $0.05 per minute. If you're running an AI voice agency, you've probably already seen three messages in Skool asking "Is this the end of Hermes?" The short answer: no. Here's the long answer.

Why This Matters for AI Voice Agencies

xAI's announcement is not about price. It's about commoditization. A $0.05/min builder is a direct threat to wrapper platforms like Voicerr, which charge $199/month and pass through usage costs. But here's what's missing from the xAI pitch: everything else.

xAI's $0.05/min does not include:

CRM or contact management
Campaign orchestration or outbound dialing
Call recording and transcription storage
White-label branding for your clients
Unified billing across all your clients
Analytics dashboards (you get data dumps)
Integration with your business tools (Zapier, Make, etc.)

So when xAI says $0.05/min, what they really mean is: "We're giving you the voice engine. You still need to build, integrate, or buy 5 to 7 other tools to run a real agency."

That's the playbook that got us here. Retell and VAPI have been saying the same thing for two years. And agencies have been buying Synthflow, Voicerr, and Hermes specifically because they got tired of the integration game.

The Real Cost Comparison

Let's run the math on a hypothetical: 10,000 minutes per month of voice traffic (about 500 calls at 20 minutes each).

Platform	Monthly Base	Overage (10k min)	CRM/Campaigns	Total/Month
xAI Grok (agent only)	$0	$500	+$300-600	$800-1100
Retell + GoHighLevel	$200	$700	$199	$1099
Voicerr (wrapper)	$299	$500-1000	$0	$799-1299
Hermes (Agency plan)	$699	$2400	$0	$3099

Wait. Hermes looks expensive here. Let me explain what's happening.

Hermes' Agency plan comes with 2,000 included minutes per month. That 10,000-minute scenario? You're not running a typical workflow. You're scaling. When you scale, two things happen:

Your per-minute cost goes down on every platform (including Hermes, at $0.24/min after overage).
You stop paying separately for CRM, campaigns, billing, dashboards. That's included.

The real win isn't the price per minute. It's the margin you keep. An agency on Hermes runs 20 clients at $1,500/month each. That's $30,000 revenue, $699 platform cost, plus usage. Your margin is 95%+. On xAI Grok with a duct-taped stack? You're paying $800-1100 for platform plus $0.05/min voice, which means every new client is a new deployment, new CRM seat, new integration. Your margin erodes with every new client.

What We're Doing at Hermes

xAI didn't invent the $0.05/min price. That's where voice commodities go when upstream costs drop. We saw Retell hit $0.07/min two years ago. Voicerr launched at $28/month. Then Retell raised, and Voicerr had to raise to $199/month to stay profitable.

We're locking Hermes pricing through 2027. That means no surprise usage increases, no upstream vendor pivots that force your renewal price up. You get the same $0.24/min, the same included minutes, the same CRM plus campaigns plus analytics, for the same $149-699/month. By builders, for builders. Margin is the business, not the trick.

If you're already on Hermes, nothing changes. If you're thinking about moving to xAI Grok because of the $0.05/min headline, we've put together a migration playbook and a cost calculator so you can see your actual all-in number before you commit.

Action Steps for Agencies Watching xAI

Calculate your all-in cost. Don't just look at voice engine price. Add CRM, campaigns, billing, integrations.
Test xAI's builder if you're curious. It is genuinely fast. But plan on spending 10-20 hours building integrations.
Document your current stack and margin. How much are you actually paying per client, end-to-end?
Ask yourself: do I own my client relationship or do I rent it? If your clients see your branding, you own it.
Watch for the next price move. When xAI's upstream costs rise, the $0.05/min won't stick.

FAQ

Q: Is $0.05/min cheaper than $0.24/min?

Yes. But you're not buying a voice engine. You're buying an agency platform. xAI Grok's $0.05/min is $500 for 10,000 minutes. Hermes at $0.24/min is $2400 for overage, but includes 2,000 minutes, CRM, campaigns, analytics, white-label branding, and billing. Agencies choose platforms.

Q: Should I migrate from Voicerr to xAI?

Voicerr raised prices 10x because their upstream (Retell/VAPI) raised prices. xAI will eventually too. If you're on Voicerr now, you're in a wrapper that passes through cost increases. If you need CRM plus campaigns today, Hermes is faster than rolling your own on top of xAI. If you love building and have time, xAI is fun. If you have clients waiting, choose a platform.

Q: Will Hermes ever match xAI's $0.05/min?

No. xAI's price is only competitive if you don't need a platform. The moment you add CRM, campaigns, and billing, xAI's stack costs more and takes longer to launch. We're focused on keeping Hermes pricing locked and margins high. Platforms let you build a real business.

Voice AI is fracturing into two categories: commodity tools (xAI, Retell, VAPI) and operating platforms (Hermes, Synthflow, native solutions). Commodity tools will keep racing to $0.05/min. Platforms will keep adding CRM, campaigns, analytics, billing.

If you're building an agency, you need a platform, not a price race. We built Hermes for that. Start free for 14 days.

Alfredo Romero is CEO and co-founder of Hermes, the operating platform for AI voice agencies.

Originally published at buildwithhermes.com/blog/xai-grok-voice-pricing-2026-07-27.

Why Claude's Voice Mode Matters More Than You Think, and What It Means for Your AI Voice Agency

Alfredo Romero — Sun, 26 Jul 2026 21:50:10 +0000

Anthropic just shipped a major update to Claude voice mode with improved reasoning, nuanced understanding, and fewer hallucinations. If you build voice agents for clients, this matters. A lot.

Why this matters for AI voice agencies

Here's the thing: your clients aren't paying for voice technology. They're paying for results. Better call handling. More conversions. Fewer embarrassing moments where the AI says something completely nonsensical.

Claude is the LLM backbone of most voice agents built by agencies. Anthropic just made Claude better at reasoning, understanding context, and handling edge cases. That translates directly to better agent responses, better client outcomes, and more renewal business for you.

The arms race in voice AI right now is all about LLM capability. OpenAI shipped faster live conversations. Google dropped Gemini 3.1 with voice. xAI launched Grok Voice Agent Builder (we'll get to that in a second). Anthropic is responding with deeper reasoning. This is good for the category. This is good for agencies.

Why? Because smarter LLMs mean agencies can build more complex workflows without custom engineering. A legal services agency can now deploy Claude to handle initial intake calls with better understanding of jurisdiction nuances. A mortgage broker can route calls more intelligently based on Claude's improved reasoning about financial scenarios. Your clients win. You keep the margin.

What we're doing at Hermes about it

At Hermes, we're LLM-agnostic by design. You pick Claude, GPT, Gemini, or any model you want when you build your agent. We handle the hard part: CRM integration, contact routing, campaign orchestration, billing, white-label branding. The model layer is just the first domino.

Anthropic's update means we (and our agency customers) can test Claude's new reasoning capabilities immediately. Build a variant of an existing agent, swap in the new Claude model, A/B test it against the old version, and promote the winner. No rebuild. No downtime. No developer required.

That's the Hermes difference. While other platforms lock you into one model or make model upgrades a 3-sprint engineering project, we let you iterate and compete on LLM capability as fast as the frontier moves.

Action steps for agencies (do this this week)

Pick one client call flow to test Claude's new capabilities. Choose a high-value call (sales, support, or lead qualification). Create an agent variant using the new Claude model.
Run a 7-day A/B test. Route 50% of inbound calls to the old agent, 50% to the new Claude variant. Measure: call duration, resolution rate, client satisfaction score (if you track it).
Document the results. If the new Claude model wins on any metric, you've got a competitive advantage story to tell your next prospect. "We use the latest AI reasoning models to reduce call times by 15%."
Educate your clients on the upgrade. When you push the new variant live, send a 2-line note: "We upgraded your agent's reasoning model to Claude [X]. Expect [shorter calls / better first-call resolution / fewer transfers]. Effective immediately."
Watch xAI's move, but don't get distracted. xAI's new voice builder is cool, but it's a voice layer. Agencies need the full operating system (CRM, campaigns, billing, white-label). More on that below.

Wait, what about xAI's new voice agent builder?

xAI shipped a no-code voice agent builder with Grok Voice at $0.05/minute. The benchmarks look solid. They claim higher audio quality than Gemini and GPT's realtime models.

It's a voice layer. It's good at doing one thing: building and running agents. But agencies need way more.

Do you have CRM integration? Nope. Contact routing based on pipeline stage? No. Campaign orchestration (batch dial, IVR-style workflows)? Nope. Transparent per-plan billing instead of per-minute panic? No. $0.05/min means you're paying for every overage.

xAI's play is aimed at engineers and call centers who want to build their own calling system. That's not your ICP. Your clients (e.g., marketing agencies, real estate teams, SMB service providers) need an operating system, not a voice layer.

That's what Hermes is built for. We're the operating platform for AI voice agencies: one system for voice agents, CRM, campaigns, billing, and white-label branding. From $149/month. Your clients never see the word "Hermes." They see their brand. You keep 80%+ margins.

The bigger picture

Voice AI is moving fast. Claude's update, xAI's builder, OpenAI's new voice models, Retell's new ops tooling. It's all happening at the same time. This is good validation that voice AI is the next frontier.

But the winners in this category won't be the ones with the best voice technology. They'll be the ones who build for the agency business model: white-label, margin-preserving, all-in-one operating system. Not the voice layer. The platform.

Claude's update is just the next step in a faster cycle. Use it. Test it. Move on to the next update next month. The agencies that stay competitive are the ones who can iterate on LLM capability without being held back by platform constraints.

Start building smarter voice agents

If you're ready to test Claude's new capabilities (and you should be), start on Hermes. Three workspaces, 300 minutes per month, all the tools you need to build, deploy, and scale voice agents under your brand.

No per-minute panic. No engineer required.

Ready to compete on AI voice? Start on Hermes for $149/month. First agent live in 72 hours. Join the beta.

Originally published at buildwithhermes.com/blog/claude-voice-mode-matters-2026-07-24.

Colorado AI Act + Wrapper Deaths: The Voice AI Reckoning Has Begun

Alfredo Romero — Thu, 23 Jul 2026 14:33:45 +0000

On July 1, 2026, Colorado's AI Act went into effect. The same week, industry reports showed 90% of AI voice wrapper startups are expected to fail by year-end. These aren't isolated events. They're signals of the same market bifurcation: regulatory compliance plus platform consolidation. If you're running an AI voice agency on a wrapper, the clock is ticking. Read on.

The Colorado AI Act: What Just Changed

Colorado's AI Act (effective July 2026) classifies AI voice agents used for "insurance, lending, employment, and credit decisions" as high-risk systems. High-risk means:

Agencies must conduct impact assessments before deployment
Agencies must maintain audit trails of every call (decision logging)
Agencies must provide transparency disclosures to end-customers
Agencies must implement bias monitoring and consent frameworks

This doesn't replace TCPA. It stacks on top. Agencies in insurance, healthcare, lending, and financial services now face compliance burden that wrappers (UI layers on Retell/Vapi APIs) were never built to handle.

The Wrapper Collapse: 90% Won't Make It

In July 2026, multiple reports converged: Machine Brief, Venture Magazine, and venture analyses all flagged the same trend. In 2024, 14,000 AI startups launched. By early 2026, over 5,600 had shut down (39% failure rate). Voice wrappers specifically (Voicerr, Vapify, Stammer, Assistable) are failing because:

Economics don't work. Wrappers need 3.2x more capital to reach profitability than platforms. Only 3-5% of wrappers crack $10K MRR.
Foundation models compress margins. GPT-4o and Claude can do 90% of what wrappers do at lower cost, with larger user bases. Custom code can't compete.
White-labeling is table stakes, not moat. Every wrapper offers white-label now. No differentiation left.
Upstream risk is real. When Retell or Vapi raise prices (and they will), wrappers have no leverage. See: Voicerr price jumps in 2026.

Google's startup chief put it plainly in February 2026: "If you're just white-labeling the model, the industry doesn't have patience for that anymore." Wrappers are being crushed between foundation models below and regulated platforms above.

Why This Matters for AI Voice Agencies

You're caught between two forces:

1. Compliance urgency. Colorado is the first state. California, New York, and the EU are next. If your voice agents touch regulated use cases (and most do), compliance tooling is now a requirement, not an option.

2. Wrapper instability. Wrappers are dying. Pricing is volatile (Voicerr jumped 7-10x in early 2026). Feature development is slow. Team funding is drying up. When a wrapper shuts down, you're scrambling to migrate your entire client base on short notice.

Wrapping Retell isn't a strategy anymore. It's a liability.

What We're Doing at Hermes

Hermes was built for this moment. We're not a wrapper on top of voice APIs. We own the entire stack: voice agents, CRM, campaign orchestration, compliance audit trails, and white-label billing.

Compliance built-in. Every Hermes agent includes:

Decision logging (call transcripts, reasoning, outcomes stored for audit)
Consent management (TCPA plus state-specific disclosures)
Bias monitoring (call performance dashboards by demographic)
Impact assessments (pre-deployment checklists for regulated use)

White-label portal. Your clients see Hermes zero times. They see your branding, your CRM, your agents. Full control, no infrastructure cost.

Transparent pricing. $149-$699/month base. $0.24/min overage. No surprises. No leverage from upstream. We control our economics.

You can build 10+ client campaigns on Hermes Starter for less than a single Synthflow enterprise license. And you keep 80%+ margins.

Action Steps for Agencies Right Now

If you're on a wrapper, your time is now. Here's what to do:

Step 1: Audit your clients' use cases. Which agents touch regulated decisions (insurance, credit, lending, hiring, medical)? Those are your compliance-first targets.

Step 2: Calculate your true cost of compliance. If you're building compliance tooling custom on Retell/Vapi, add it up: dev time, audit logging, consent management. Then compare to the built-in cost of a Hermes plan. You'll be shocked how fast custom cost exceeds platform cost.

Step 3: Start migration with your highest-margin, most regulated client. Test the Hermes workflow end-to-end. Your team will see the compliance gap immediately.

Step 4: Commit to a platform, not a wrapper. Platform risk is still lower than wrapper risk. Hermes has backing, a product roadmap, and operators who know how to scale this. Wrappers have a countdown clock.

FAQ

Q: Does Colorado AI Act apply to my voice agents outside Colorado?

A: Yes, if your end-customer is in Colorado. State laws apply to the customer jurisdiction, not where you operate. Same for California, soon.

Q: Do I need to migrate all clients, or just regulated use cases?

A: Start with regulated use cases. But compliance tooling is now table stakes for any serious agency. Eventually, all agents benefit from audit trails, consent frameworks, and performance monitoring. Plan for full migration.

Q: What if I want to stay on Vapi/Retell plus build compliance custom?

A: You can. It'll take 2-3 months of dev work and cost $15K-$30K. Then you're locked into custom tooling that only you understand. When regulation shifts, you rebuild. With Hermes, compliance updates ship to everyone. Instant coverage.

The Reckoning

The voice AI market is consolidating fast. Wrappers are being squeezed out. Regulation is codifying platform advantages. If you're an agency operator who built on wrappers, the next 90 days are critical. Either migrate to a compliant platform or invest heavily in custom compliance tooling.

We built Hermes for this exact moment. Operators, for operators. By builders, for builders.

Originally published at buildwithhermes.com/blog/colorado-ai-act-wrapper-crisis-2026-07-22.

First Synthflow. Now Vapi. What Happens When Your Voice AI Stack Chases Enterprise?

Alfredo Romero — Tue, 21 Jul 2026 14:34:21 +0000

On May 12, 2026, Vapi raised $50M at a $500M valuation, led by Peak XV, Microsoft M12, Kleiner Perkins, and Bessemer. Amazon Ring evaluated more than 40 AI voice vendors and chose Vapi. Today 100 percent of Ring's inbound calls run through Vapi's platform. Peak XV partner Arnav Sahu put it plainly: "In ten years, it's likely most calls will not have a human behind the phone. Vapi has the potential to be the defining platform for this entire revolution."

That is a true statement. It is also a statement about Amazon Ring, New York Life, and the next 200 Fortune 500 logos Vapi's board expects them to land. It is not a statement about the agency owner with six clients running 800 minutes a month.

Sixty days earlier, Synthflow confirmed its white-label agency plan at $2,000 per month, clearly aimed at large BPOs building AI calling products for multiple enterprise clients. Their enterprise plan targets 10,000 or more minutes per month and is priced around governance, compliance, and SLA guarantees that a Fortune 500 procurement team requires. The same company that marketed to solo agency builders in 2024 now charges $2,000 before a single minute of call time.

Two dominant infrastructure providers. Two formal enterprise pivots. Sixty days apart.

This is not a coincidence. It is the predictable second act of any infrastructure company that raises a meaningful VC round and needs to justify the valuation. The agency tier is a great acquisition channel. It is a terrible enterprise account. The math forces a choice, and both Synthflow and Vapi have made it.

Why this matters for AI voice agencies

The practical implications show up in three places: pricing, roadmap, and support.

Pricing. When Voicerr went from $28 to $199 to $299 per month overnight in early 2026, a 7x to 10x increase, the lesson the agency layer learned was not that Voicerr was uniquely bad. The lesson was that wrappers and platforms that do not control their own upstream cost structure will pass increases downstream with no notice. Vapi at $500M valuation has investors who did not write that check to subsidize a $149 per month operator forever. The pricing trajectory for the agency tier on VC-backed infrastructure platforms is not downward.

Roadmap. Amazon Ring is not asking for better onboarding wizards or an integrated campaign builder for outbound sequences. They are asking for governance, predictability, uptime SLAs, and call-level monitoring for their enterprise operations team. Vapi said as much: their stated next phase after the raise is governance, predictability, tighter uptime SLAs, and call-level monitoring for enterprise operators. Every engineering sprint that ships those features is a sprint that is not building the CRM integration, campaign manager, or white-label billing surface that an agency with eight clients actually needs.

Support. This one is quieter but it shows up consistently. The G2 data makes it explicit: "expensive" is the leading negative theme at 145 mentions across voice AI platform reviews, and the second most common complaint is prioritization. When a platform's biggest client doubles their concurrency, the engineering team does not optimize for the account running 300 minutes a month. This is not malice. It is queue mechanics.

The pattern, and what usually comes next

PolyAI raised $86M at $750M in December 2025 and is now competing for the same enterprise BPO contracts Synthflow is chasing. Retell was named to Wing VC's Enterprise Tech 30 list for 2026 and crossed $50M ARR, 50 million calls per month, still API-first and developer-first. The infrastructure layer is consolidating around enterprise buyers.

This is not new. It happened to every developer-platform cohort that got meaningfully funded: Twilio, Segment, Braze. The agency or small-operator tier is the beachhead, and the enterprise is the margin. Platform companies do not stay on the beachhead after they have raised $50M to go somewhere else.

The agencies that get hurt are the ones who noticed late. The ones who built their entire client delivery on a single infrastructure provider's API, with no CRM, no white-label, no billing surface, and no fallback, and then found out about the pricing change from a Reddit thread instead of a 90-day notice in their inbox.

What we're doing at Hermes about it

Hermes is not building for Amazon Ring. We are building for the agency owner with 3 to 15 clients who needs a full operating platform under their own brand, with margins they can predict and a bill that makes sense.

The positioning is deliberate, not accidental. Starter is $149 per month, 3 workspaces, 300 included minutes. Business is $399, 7 workspaces, 1,000 minutes. Agency is $699, 20 workspaces, 2,000 minutes. Overage at $0.24 per minute against a $0.18 landed cost. The 25 percent spread is locked because we run the upstream relationship. We do not pass through Vapi's pricing decisions. We do not inherit Retell's outage risk without a fallback layer. We control the cost structure on purpose.

The platform includes what an agency actually needs to run clients: a white-label portal your clients log into under your brand, a native CRM with contact management and pipeline view, a campaign builder for outbound sequences, and a billing surface where you see exactly what you are paying and what you are charging. Your clients never see the word Hermes. That is the product, not a roadmap item.

By builders, for builders. The infrastructure layer is going enterprise. We are staying in our lane.

Action steps for agencies affected this week

Map your current stack to the infrastructure layer that owns it. If you are on Vapi directly, or a wrapper like Stammer, Vapify, or VoiceAIWrapper that sits on top of Vapi or Retell, you are two to three layers away from controlling your own pricing. Draw that dependency graph today. The platform you are renting from just got a $50M reason to change what it charges you.
Calculate your true cost per client, not the invoice total. Add your Vapi or Retell bill, your GHL or CRM bill, your Zapier or automation bill, your Twilio bill, and any dev cost you are paying to stitch them together. Divide by active clients. If that number is above $250 per month per client before you bill them anything, you are running a margin problem at scale, not just at your current size.
Ask your platform what happens when its biggest client grows. Get a written answer about pricing guarantees and concurrency allocation. If the answer is a support ticket and a "best effort" clause, you now know how the queue works.
Lock in white-label now, before the platforms that offer it go enterprise. Synthflow's white-label is $2,000 per month. The wrapper platforms that offer white-label do not control their upstream pricing. The window to build a client-facing brand on infrastructure that is genuinely built for the agency tier is narrowing. Use it.
Tell your clients before the infrastructure tells them. If your clients are aware that you run on Vapi or any named provider, a pricing or outage event at that provider becomes your client's business problem without you as the intermediary. The white-label layer exists precisely so this conversation never happens. If you do not have it, that is the first thing to fix.

Frequently asked questions

Vapi raised $50M in May 2026. Is it still a good choice for AI voice agencies?

Vapi is technically strong infrastructure. The question is not whether it works today. The question is whether the roadmap, SLA tiers, support queue, and pricing structure will continue to optimize for the operator running 300 to 2,000 minutes a month, now that the company's largest customer routes 100 percent of its inbound calls through Vapi and the board is expecting a return on a $500M valuation. Amazon Ring is not the same buyer as a solo agency owner with four clients. When those two customers share a priority queue, the agency is not at the top. That is not a knock on Vapi. It is just how incentives work after a $50M Series B.

Synthflow's white-label plan is $2,000/month. Is Hermes cheaper?

Hermes Business is $399 per month with 7 workspaces and 1,000 included minutes. Hermes Agency is $699 per month with 20 workspaces and 2,000 included minutes. A five-client agency on Synthflow's white-label pays $2,000 per month before a single minute of call time. The same agency on Hermes Agency pays $699 per month with 2,000 minutes included, overage at $0.24 per minute. The math works out to roughly $1,300 per month in pure cost difference before you account for call volume. The Hermes pricing is locked because we run the upstream relationship. We do not inherit upstream price increases and pass them to you.

What is Hermes actually doing differently from Vapi or Synthflow?

Vapi is API infrastructure. Synthflow is a no-code agent builder now targeting enterprise BPOs. Both are excellent at what they do. Neither is an agency operating platform. Hermes is built specifically for the 1 to 20 client agency: white-label CRM, campaign orchestration, outbound dialer, transparent billing, and number management all under your brand. Your clients never see the word Hermes. The platform is designed around the P&L of an operator charging $1,500 to $3,000 per client per month, not the P&L of a Fortune 500 contact center. That focus does not change after a VC round because we do not have one.

Where this leaves you

Vapi is excellent infrastructure. Synthflow built a genuinely capable no-code agent builder. Neither company owes the agency tier a different roadmap. They are going where the margin is, which is what every well-run company does after a $50M Series B.

The agency owner with 3 to 15 clients is not being abandoned by bad actors. They are being abandoned by good companies that solved the infrastructure problem for themselves and moved to the next problem, which happens to be on a different floor of the market.

The response to that is not to complain about it. The response is to use a platform that was built specifically for the floor you are on. One platform. Your brand. Your margins. From $149 per month.

Originally published at buildwithhermes.com/blog/vapi-50m-enterprise-pivot-agency-left-behind-2026-05-29.

5 New Voice Agent Builders Hit the Market in July 2026: Why They All Miss the Mark for Agencies

Alfredo Romero — Mon, 20 Jul 2026 14:34:50 +0000

July 2026 is the month the voice AI market split in half.

On one side: xAI just launched Grok Voice Agent Builder with $0.05/minute pricing and bundled telephony. Bland AI closed $50M Series C backed by Dell, HubSpot, and Twilio's founder. Synthflow pivoted to enterprise, Rime raised $24M, and even GoHighLevel quietly shipped sub-600ms voice latency on their $97/month plan.

On the other side: 50,000 SMB voice agencies are still piecing together Retell + GoHighLevel + Zapier + Stripe + Twilio + a custom dashboard (or a developer on payroll). They're paying $1,500 to $2,000/month in tooling costs just to deliver what the vendors now say should cost $149.

The problem isn't the builders. It's that builders and operators are solving different problems.

Why This Matters for AI Voice Agencies

Here's what 2026 taught us: a voice agent platform without a CRM, campaign engine, and billing layer isn't a platform for agencies. It's a toy.

When xAI charges $0.05/minute and bundles 80 voices, voice cloning, and free phone numbers, they're competing with Vapi and Retell on features. That's a builder's game. But agencies don't make money on features. They make money on margin.

The math is brutal:

xAI builder: $0.05/min + your CRM + your billing + your compliance tracking + your agent management = you still need 4 to 5 more tools.
Hermes: $149 to $699/month all-in. Your brand. 80%+ margin on every client.

Bland's Series C validates something different: enterprise-scale voice AI is real. Dell, HubSpot, Twilio, and Affirm's founder Max Levchin didn't invest $50M in a voice wrapper. They invested in a call center replacement. But Bland's customers are Fortune 500 BPOs taking 5M+ calls/month. Agencies doing 50 to 500 calls/month at $1,500/client? Bland isn't pricing for you.

And here's the regulatory bomb nobody's talking about: the FCC confirmed TCPA applies to AI-generated voices. Statutory damages: $500 to $1,500 per call. Texas SB 140 requires disclosure in the first 30 seconds. Colorado classified voice AI as "high-risk" and demands audit trails.

Vapi. Retell. Synthflow's old SMB tier. None of them have built compliance tracking into the platform. You, the agency owner, have to bolt it on yourself. One missed disclosure. One logging gap. $500 to $1,500 per call in damages.

Hermes was built with compliance native from day one. Consent tracking. Call recording storage. State-aware routing. That's not a feature. That's insurance for agencies.

What We're Doing at Hermes About It

We launched Hermes specifically because we watched agencies lose margin to tool sprawl.

When you're running 3 to 5 clients on Hermes, you're not managing Retell configs, GoHighLevel automations, Zapier zaps, and Stripe webhooks. You're managing one platform. Your white-label brand. Your margins intact.

We also built compliance native because we knew it was coming. TCPA tracking, state law routing, call recording with audit trails, and consent verification all included in your Starter plan at $149/month. Synthflow charges $1,400/month for enterprise and still doesn't have it.

And pricing: we locked it at $149, $399, and $699 because we believe agencies shouldn't choose between voice AI and staying profitable. Vapi's cheapest play is pay-per-minute at no fixed cost floor. Hermes' Starter plan costs less than a good lunch per day. Your first 3 clients pay for it.

Action Steps for Agencies Affected

If you're running voice agents today, you have three moves:

1. Audit your compliance debt. Pull your Retell logs for the last 30 days. Check: Are you logging consent? Storing recordings? Disclosing AI status to every caller? If the answer is "I think so," you're exposed. TCPA damages are $500 to $1,500 per call. One lawsuit wipes out 3 months of margin. Move this to the top of your sprint.

2. Map your tool costs. Retell, GoHighLevel, Zapier, Stripe, Twilio, plus the developer time gluing them together. Write down the number. For most agencies running 5 to 10 clients, it's $2,000 to $3,500/month. Hermes at $699/month ($0.24/minute overage on usage) handles all of it. Do the math for your client mix.

3. Get off the builder treadmill. Every new voice AI builder (xAI, new GHL features, Rime, Bland's SMB tier if they ever ship one) is going to tempt you with shinier features. Resist it. You don't need another agent builder. You need operational margin. Agencies that switched from Synthflow's old SMB play to Hermes in Q2 2026 reported 40 to 60% margin improvement on voice work. That's not a small number.

The Real Split

The voice AI market in 2026 isn't builders vs. builders. It's builders for everyone else, and platforms for agencies who actually want to scale profitably.

xAI, Bland, Synthflow, Rime, and GHL are all solving the "how do I build a voice agent" problem. That's solved. It's not interesting anymore.

The problem that still exists: how do I run voice agents, white-labeled under my brand, with built-in compliance, one platform for CRM and campaigns, and margins that stay above 80%?

That's Hermes.

FAQ

Q: Isn't xAI's $0.05/minute pricing better than Hermes?

A: No. xAI's $0.05/minute is the cost per minute of speech. You still need a CRM ($50 to $300/month), telephony integration ($50 to $200/month), compliance tracking ($100 to $500/month if you hire someone), and billing infrastructure (1 to 2% to Stripe). Hermes at $699/month includes all of it. Total cost of ownership: xAI route = $700 to $1,500/month. Hermes = $699/month fixed + $0.24/minute overage. For most agencies, Hermes costs 60 to 70% less.

Q: Does Hermes integrate with GoHighLevel?

A: Yes. Hermes works alongside GHL for agencies already on the ecosystem. We also have native CRM, so you don't need both. Many GHL agencies switched because Hermes voice AI is faster (sub-400ms latency) and compliance-native. GHL's AI Employee plan voice features are good for light use; Hermes is built for agencies running 50 to 500 calls/month per client.

Q: What if I use Retell for one client and want to migrate?

A: We handle Retell-to-Hermes migrations in 48 hours. Your agent config, logs, and call history migrate over. Your clients stay on your number. Zero downtime. We've done this for 30+ agencies in Q2 2026.

What's Next

The voice AI market will consolidate around builders (Vapi, Retell, xAI, maybe Bland's new tier) and operators (Hermes, and maybe one enterprise play). If you're an agency, pick the operator. If you're a developer, pick the builder. Don't mistake the two.

We're the platform for operators. Start free at buildwithhermes.com/beta, or compare Hermes to Synthflow.

Alfredo Romero is CEO and co-founder of Hermes, the operating platform for AI voice agencies. 50+ agencies run 5,000+ clients on Hermes. By builders, for builders.

Originally published at buildwithhermes.com/blog/5-new-voice-agents-july-2026.

xAI's $0.05/min Promise: The Hidden Cost Reality for Agencies

Alfredo Romero — Sun, 19 Jul 2026 14:34:54 +0000

xAI just launched their Voice Agent Builder with aggressive $0.05/min pricing. But what they don't tell you is that actual all-in costs reach $0.12 to $0.25/min. Here's the breakdown.

The xAI Announcement

On July 1, 2026, xAI announced their Voice Agent Builder, pricing it at an eye-watering $0.05/minute for platform use. On the surface, it undercuts Vapi ($0.05 base) and Retell ($0.07 to $0.18/min) by claiming to be the cheapest option on the market.

Within 48 hours, every AI voice agency builder in Discord and Skool was debating whether to switch. "xAI at $0.05 is a no-brainer," the thinking goes. "That's half of what I'm paying now."

Here's the problem: that $0.05 number is a trap. And if your agency falls for it, your margins will suffer immediately.

Red flag: Any platform quoting per-minute cost that doesn't include STT, LLM, and TTS is being deceptive. It's not misleading by accident. It's by design.

Why This Matters for AI Voice Agencies

Agencies already got burned once. Voicerr raised prices 7 to 10x in early 2026. GoHighLevel bundled voice as an add-on. Synthflow priced agencies out of the market.

The pattern is the same every time: headline pricing that excludes costs, followed by invoice shock 30 to 60 days in.

If you build your client pricing around xAI's $0.05/min rate, and your actual costs turn out to be $0.15 to $0.20/min, your margins disappear. You're running a charity, not an agency.

The key insight: "Per-minute" pricing without full transparency is a red flag. Full stop. This is not a platform feature; it's a marketing tactic.

xAI's $0.05: What's Actually Included?

xAI's $0.05/min breaks down like this:

xAI Platform Fee: $0.05/min
What it covers: API calls, routing, session management, basic logging.
What it does NOT cover: Speech-to-text, large language model inference, text-to-speech, telephony.

Speech-to-Text (STT): +$0.02 to $0.04/min
xAI uses third-party STT. That cost passes through to you. Whisper-quality STT is $0.02 to $0.04/min.

Large Language Model (LLM): +$0.03 to $0.06/min
xAI runs on Grok, but Grok inference is expensive. Expect $0.03 to $0.06/min for model API calls, depending on token use and model tier.

Text-to-Speech (TTS): +$0.02 to $0.05/min
Natural-sounding voice (Google Cloud TTS, Azure TTS, ElevenLabs) runs $0.02 to $0.05/min. Cheap TTS (generic) is $0.01/min but hurts agent quality.

Telephony (Twilio, CallKit, etc.): +$0.01 to $0.03/min
Inbound and outbound calls via Twilio or equivalent cost $0.01 to $0.03/min. This is unavoidable.

Total All-In Cost
Platform: $0.05, STT: $0.03, LLM: $0.04, TTS: $0.03, Telephony: $0.02 = $0.17/min minimum. With premium TTS and LLM: $0.25/min.

Stop chasing per-minute platforms. Move to transparent pricing.

Hermes gives you predictable all-in costs. Voice agents, CRM, campaigns. One platform. Your brand. From $149/month. Plans run $149, $399, and $699/month, with a Contact option for larger agencies.

Originally published at buildwithhermes.com/blog/xai-pricing-exposed-hidden-cost-reality-2026-07-18.

How We Cut Voice AI Onboarding From 10 Hours to 90 Minutes Per Client (Template Inside)

Alfredo Romero — Sat, 18 Jul 2026 14:35:23 +0000

Voice AI agencies lose an average of 14 hours per client just collecting information and configuring agents, per the 2026 onboarding research published by GrowwStacks. Most of that time is not skill work. It is form-filling, re-uploading documents, retyping FAQs into a fresh agent, switching browser tabs between five tools, and writing the same system prompt for the seventh time. The 90-minute template compresses all of that into one workspace clone, one structured intake form, one website-scrape ingest, one prompt personalization pass, and one live validation call. The compression comes from removing duplicate work, not from cutting corners. A2P 10DLC submission and any number port-in still run on their own calendar windows, but they happen in parallel and do not block the agent going live. This post is the exact template, the order of operations, the intake-form schema, the parallel-track A2P plan, and the audit we run on the first month of live calls. Pull it once. Use it every client. Onboard four in a day.

I have run this template on every voice AI client I have touched in the last 90 days, including the five operators currently on the Founders' Beta. The pattern holds. The first client takes 3 hours because the canonical snapshot does not exist yet. Clients 2 through 50 take 90 minutes. The two-week break-even on the snapshot build is the single highest-ROI operator move in an agency's first 90 days.

By builders, for builders. Everything below is the working version of what we run inside Hermes, plus the upstream references we used to validate that the time savings are real and not just a single-operator anecdote.

Where do voice AI agencies actually lose 10 hours per client?

Time-tracking data from the last 12 voice-AI onboardings I ran with operators on the beta lands within 10% of the industry numbers. Trillet's 2026 knowledge-base training research puts manual knowledge base entry alone at "2 to 4 hours per client" when it is typed from scratch, and at 4 to 6 hours when the client also lacks structured FAQs. Cekura's 2026 Retell vs Vapi build-time analysis measures a first-time Vapi build at "20 to 60 hours" and a first-time Retell build at "8 to 20 hours" before the agent hits a single live call. Subsequent builds compress, but only if the operator templatizes.

"82% of voice AI agencies report losing clients during onboarding due to delays and miscommunication. Missing information creates back-and-forth that delays launch and frustrates clients." [GrowwStacks, Voice AI Client Onboarding 2026]

The 14 hours of average operator time break out roughly like this across the books I have audited.

Onboarding step	Unstructured time	Template time
Workspace setup, billing, branding	60 to 90 minutes	8 minutes (snapshot clone)
Discovery and information collection	2 to 3 hours	10 minutes (intake form review)
Knowledge base load	2 to 4 hours	15 minutes (website scrape)
System prompt and conversation states	2 to 3 hours	20 minutes (personalization pass)
Integrations and tool calls	1 to 2 hours	0 minutes (pre-wired in snapshot)
Sandbox call validation	45 to 60 minutes	12 minutes (5 scripted calls)
Production cutover and first live call	45 to 60 minutes	15 minutes (3-step flip)

The unstructured workflow totals 9 to 14.5 hours. The template workflow totals 80 minutes plus a 10-minute buffer. The compression is not magic. It is the disappearance of repetitive work that did not need to be done a second time.

What does the 90-minute template actually look like?

The template has five files and one workspace primitive. You build them once, version-control them, and clone the workspace for every new client. The whole point of the template is that the cognitive load is zero on client number two onward.

1. The canonical agency workspace

The canonical workspace holds the agent skeleton, the system tool registry, the call-review queue presets, the post-call webhook payload, the billing template, the white-label brand fields wired to CSS variables, and the affiliate-aware payout splits. It does not hold any client-specific content. It is the empty kitchen, not the meal. The closest analog is the HighLevel SaaS snapshot model, where "the entire setup deploys automatically in under 10 minutes" because the snapshot already encodes the canonical configuration. Hermes workspaces work the same way. The clone takes 8 minutes end to end including the welcome email auto-firing to the new client.

2. The structured intake form

Eighteen questions, no more. Business legal name, DBA, EIN, service address, primary URL, three competitor URLs, top five FAQs, the three calls per week that the client most wants to capture and the three they want to deflect, the booking link if one exists, the existing CRM, the existing telephony, the brand voice description in three sentences, and the must-not-say list. The intake form gates everything downstream. Without it, the template breaks because the prompt-personalization pass has nothing to consume. The intake form also doubles as the SOW reference. Five minutes for the client to fill out, ten minutes for the operator to review.

3. The website-scrape knowledge-base ingest

Hermes ingests the client's primary URL plus a depth-3 crawl, strips boilerplate, deduplicates near-duplicate paragraphs, and writes the resulting chunks into the workspace knowledge base. Coverage is usually 85 to 95% of the questions the agent will actually field on the first day. The remaining 5 to 15% gets backfilled from the intake form's top-five-FAQs field. Knowledge base loading drops from the industry baseline of 2 to 4 hours of typing to a 15-minute ingest plus review. Trillet's research notes the same effect: "Website scraping is much faster, taking only 5-10 minutes with minimal client involvement."

4. The prompt-personalization pass

The canonical prompt is a Jinja-style template with named variables for business name, voice persona, must-not-say list, booking flow, and escalation policy. The personalization pass is a 20-minute operator block that maps intake-form answers to prompt variables, runs the substitution, and previews the first three turns of dialogue against the new agent in the sandbox. The prompt itself never gets rewritten from scratch. We have rebuilt the canonical prompt three times in 90 days. We have not rewritten a client-specific prompt once.

5. The sandbox validation checklist

Five scripted calls against the workspace sandbox number cover the happy path, two common edge cases, the must-deflect path, and a knowledge base hit on a question the client specifically flagged in intake. The checklist takes 12 minutes including listening to playback. The validation pass either fires the green light for production cutover or kicks back to the prompt-personalization step with notes. The checklist itself is a Hermes call-review queue preset, so the same operator workflow that handles live calls after launch is what the operator is already using on the validation pass.

How do you actually run a 90-minute onboarding, step by step?

Open the workspace, open the intake form response, open the client's website in a side tab. Set a timer if you are measuring. Here is the order.

Clone the canonical snapshot. Workspace settings, new workspace, snapshot source = canonical. Fill in client legal name, DBA, billing email, white-label brand color. 8 minutes, including the auto-fired welcome email to the client.
Review the intake form. Read top to bottom once. Flag anything that contradicts what you can see on the client's website. Three of every ten intake forms have at least one inconsistency that would have surfaced as a back-and-forth later. Resolve them now with one short email. 10 minutes.
Trigger the website scrape. Paste the primary URL into the knowledge base ingest. The crawl runs for 8 to 12 minutes in the background. Skim the resulting chunks for obvious junk (cookie banners, careers pages, blog comment threads). Delete junk. 15 minutes elapsed including the wait.
Run the prompt-personalization pass. Open the canonical prompt template. Substitute intake-form fields into variables. Run the three-turn dialogue preview. If anything sounds off-brand, edit the named variables, not the prompt body. 20 minutes.
Validate on the sandbox number. Dial the sandbox number from a personal cell. Walk through the five scripted scenarios. Score each on the validation checklist. 12 minutes.
Cut over to production. Repoint the client's production number's Voice URL to the Hermes SIP endpoint, exactly the same pattern documented in our Retell to Hermes migration playbook. Dial the production number. Confirm the agent answers live. 15 minutes.
Send the client the white-label portal link. The portal already has the brand colors and the workspace name. The client logs in, sees their agent listed, and can run their own test call from inside the portal. 10 minutes including the short loom you record on first run.

Total operator time: 90 minutes. If anything blows the budget, it is almost always step 4 (a vague intake-form answer requiring a follow-up) or step 5 (an STT misfire on an unusual proper noun). Both are recoverable inside the window.

What about A2P 10DLC and number provisioning?

Carrier-side compliance has to be handled. It just does not have to be handled serially. Under current US carrier rules, Conduit's 2026 A2P 10DLC step-by-step documents brand registration at 1 to 3 business days and campaign registration at 3 to 7 business days. The Tuco AI guide notes the same window: Tuco AI A2P 10DLC Registration Guide 2026. The trick is to file the A2P bundle on day zero, before the onboarding session, using the intake form's EIN and service address. The voice path runs over Twilio Elastic SIP Trunking and does not gate on A2P. Only outbound SMS gates on A2P. So you ship the voice agent on day zero. The SMS fallback turns on whenever the carrier returns approval.

"As of February 2025, enforcement is fully live, and all major US carriers now block unregistered A2P traffic outright." Treat A2P as a parallel calendar item, not a voice-AI blocker. [Conduit, A2P 10DLC Registration 2026]

If the client is bringing a number that already exists on their Twilio sub-account, the cutover is the SIP-URL flip described above. If the client needs a fresh number, Hermes provisions a workspace number from inventory in under 60 seconds. Either way, the number is live on the voice path before A2P approval lands.

Why the template works (and where similar templates fail)

Most onboarding templates published in the agency space are SOP documents, not workspace primitives. A 14-page Notion doc titled "Voice AI Client Onboarding Checklist" is not a template. It is a memory aid. The reason the 90-minute number is reachable is that the snapshot itself encodes the configuration. The operator does not consult a checklist for each of the 60 fields the workspace needs set. The snapshot sets them on clone.

The second reason is that the intake form is structured data, not a discovery call. Discovery calls average 45 to 60 minutes and end with a transcript that the operator then has to re-type into the workspace anyway. The structured intake form is the same information in a form the operator can paste into the prompt variables in seconds. OnboardMap's 2026 onboarding benchmark ties the same pattern to retention directly: "Businesses with a defined onboarding timeline retain 34% more clients in the first year than those without one."

The third reason is that knowledge base ingest is a system, not a task. A scrape returns 500 to 2,000 chunks in 10 minutes. A human typing the same content in returns 200 chunks in 4 hours and forgets 12 of them. The system runs while the operator is reviewing the intake form. Time compounds because two steps run at once.

Five common failure modes and how to handle them

The client's website is a Wix one-pager. The scrape returns 30 chunks. Compensate with a 30-minute intake-form follow-up call focused on the top 20 FAQ topics the agent will hit. Total onboarding lands at 2 hours, still 5x the unstructured baseline.
The intake form arrives 70% complete. Do not start the 90-minute clock until the form is full. Send a one-paragraph email naming the three missing fields. Wait. Starting incomplete is the single most reliable way to lose the time budget.
The personalization pass produces a brittle-sounding agent. Almost always a must-not-say list problem. Move that list earlier in the system prompt and re-run the three-turn preview. The variable position in the prompt matters more than the content.
The sandbox call hits an STT misfire on a brand-specific proper noun. Add the noun to the workspace pronunciation dictionary. The fix takes 90 seconds and survives every future call. Do not edit the prompt for this; edit the dictionary.
A2P 10DLC submission gets kicked back. The kickback is almost always missing or mismatched EIN/service-address data, per the same carrier patterns documented across every 2026 guide. Resubmit within 24 hours with corrected fields. The voice agent stays live the entire time.

The action steps (do these this week)

If you onboard your next client at 10 hours instead of 90 minutes, that is 8.5 hours you will not get back. If you onboard four clients at 10 hours, that is a full work-week. The template pays for itself the second time you run it. Here is the minimum viable version you can stand up in one afternoon.

Build the canonical workspace once. Two hours of focused work to lock the system prompt scaffolding, the tool registry, the call-review queue presets, and the post-call webhook payload.
Write the 18-question intake form. Use a single form tool. Make every field required. The hardest part is resisting the urge to add a 19th question.
Configure website-scrape ingest as the default knowledge base source. Type-from-scratch is the fallback, not the default. The default has to be the system.
File the A2P 10DLC submission on day zero. Use the intake form's EIN and service address. Voice goes live in parallel.
Run the validation checklist as a call-review preset, not a Notion doc. The same surface that handles live calls after launch handles the validation pass before launch. One workflow, not two.
Audit the first month of live calls against the validation checklist. Any score the agent missed in sandbox that also misfires in production is a snapshot bug, not a client-specific problem. Fix the snapshot. The next client inherits the fix.

What does this look like as the agency scales?

At 10 clients, the snapshot has been hardened against most of the patterns that show up in real call traffic. At 25 clients, the snapshot is stable and the variation between clients lives entirely in the named prompt variables and the knowledge base content. At 50 clients, the operator time per onboarding drops below 60 minutes because the intake-form follow-up rate falls (clients increasingly come from referrals and arrive better-prepared) and the sandbox validation passes on the first run more often. The marginal cost of the 51st client is a clone, an ingest, and a single call. That is the entire reason this category gets called infrastructure rather than tooling.

If you want to see the canonical snapshot, the intake-form schema, the call-review presets, and the A2P submission template all wired together in a live workspace, apply to the Founders' Beta at buildwithhermes.com. Workspaces ship pre-loaded with everything above. You can fork the snapshot, edit it to your agency's conventions, and start cloning per client the same day.

Where this leaves you

The 10-hour onboarding is not a skill problem. It is a tooling problem. The operators who solve it look the same from the outside as everyone else, but they ship four clients on a Tuesday and read intake forms with a coffee in hand. The snapshot is the leverage. The intake form is the gating mechanism. The website scrape is the compounding effect. The 90 minutes is the result. Build the template once. Use it every client. The first month-end you do not stay up until 2am wiring up a new workspace is the proof.

By builders, for builders. We rebuilt this template three times before it stopped breaking on the fifth client. The version you get inside the Hermes workspace is the version that survived contact with operators who do not have time to debug their own onboarding kit.

Originally published at buildwithhermes.com/blog/cut-voice-ai-onboarding-10-hours-to-90-minutes-template.

The FCC Just Changed AI Voice Calling Forever: $500-$1,500 Per Call Penalties

Alfredo Romero — Fri, 17 Jul 2026 14:46:22 +0000

The FCC confirmed TCPA applies directly to AI-generated voices. That means every outbound call your agency places using an AI voice requires prior express written consent. Miss it, and you're liable for $500 to $1,500 per violation with no aggregate cap. Read the full FCC statement.

This isn't a new rule. The Telephone Consumer Protection Act (TCPA) has been law since 1991. But for the past five years, there was a grey area around AI voices: Are they "artificial" under the law? Do they trigger the same consent requirements as predictive dialers and autodialed calls?

Today, the FCC answered: Yes, and yes.

The practical impact is immediate. Every AI voice agency operating right now is either compliant or exposed. There is no middle ground.

Why This Matters for AI Voice Agencies

You are not a passive platform. When your client deploys an AI voice agent, you are the architect of that agent. You are liable for the calls it places.

Here's the chain of liability: Your client dials 10,000 leads using your AI agent. Leads start complaining to the FCC. The FCC audits. Turns out 500 of those calls (5%) didn't have documented consent. That's 500 violations at $500 minimum each = $250,000. Your client's business folds. Your client sues you. You fight or settle. Either way, you lose.

Most AI voice agencies today use API infrastructure (Retell, VAPI, Synthflow, Bland). These platforms do not enforce consent compliance by default. You have to build it yourself on top. That means:

Do-not-call registry checks on every call (you're building this manually)
State-level consent tracking (California opt-in, New York DNC list, etc.)
A2P carrier pre-approval (you're managing paperwork)
Consent recording and audit trail (for legal defense)
Liability tracking and incident response (if something goes wrong)

If you're duct-taping Retell + GoHighLevel + Zapier + a custom Airtable, you have zero of this. You're flying blind. And you're now officially liable for every non-compliant call.

The silver lining: Compliance is now table stakes. Agencies that move fast can position compliance as a selling point. "Use our voice agents, we handle FCC compliance" becomes a differentiator worth $100-300/month premium per client.

What We're Doing at BuildWithHermes About It

Hermes was built compliance-first. Every agent you deploy runs through three gates before placing a single call:

Do-not-call registry check (DNCL) against FCC database, real-time per call
A2P carrier pre-approval (we handle the submissions to AT&T, Verizon, T-Mobile). You don't touch paperwork.
State-level consent rules (California requires written consent, New York requires DNC check, 14 other states have custom rules). We enforce all of them automatically.

On Hermes, you deploy an agent for $149/month (Starter plan). Included: 300 minutes of calling, compliance checks, carrier approval, DNC registry access, audit trail.

A2P compliance submission is $30 per agent (pass-through carrier cost). You submit once, your agent is approved for nationwide deployment.

Your end-clients don't see the complexity. They see: "Your voice agent is FCC-compliant. We handle the risk."

Action Steps for Agencies (This Week)

If you're currently running AI voice agents on any API platform, here's what to do:

Audit your last 30 days of calls. Log into Retell/VAPI/Bland dashboard. Export your call history. Count total calls. Then count: How many went to numbers on a do-not-call list? How many had documented consent? If you don't have answers, you're exposed.
Check your consent sourcing. Did your clients source leads from opt-in lists? Or cold outbound lists? Hermes enforces consent at the call level. API-only platforms don't. If you've been running cold outbound without consent, that's 100% of your calls violating TCPA.
Document your current compliance posture. Write a one-pager: "Here's how we check do-not-call lists. Here's how we document consent. Here's our audit trail in case of FCC inquiry." If you can't write this, you don't have compliance. And if you don't have it, your liability insurance won't cover you.
Calculate your compliance cost. Hiring a compliance consultant: $10K-50K upfront, then $2K-5K/month ongoing. Building compliance infrastructure: 6 months, 2 engineers, $100K+. Using Hermes: $149/month, compliance included, live in 72 hours. The math is not close.
Plan your infrastructure migration. If you're on Retell or VAPI, you can migrate to Hermes and keep your Twilio numbers. We have a dedicated migration guide. Your clients see no change. Your compliance posture becomes institutional.

The Numbers

Scenario: You're a 5-person agency with 8 clients. Each client runs 500 calls/month. That's 4,000 calls/month, 48,000 calls/year.

If 3% of your calls are non-compliant (a conservative estimate for API-only platforms), that's 1,440 non-compliant calls/year.

FCC audit? 1,440 calls x $500 per call minimum = $720,000 in liability. Your revenue is probably $50K-200K/month. You can't survive a single FCC audit.

Hermes cost: $149/month x 12 months = $1,788/year. A2P submissions: $30 x 8 agents = $240/year. Total: $2,028/year for compliance cover.

The bet is whether $2K/year is less than the expected value of an FCC enforcement action. Spoiler: it is by a factor of 350x.

What This Means for Competition

Synthflow and Retell are infrastructure. They're agnostic about compliance. VAPI is agnostic. Bland is agnostic. They build the engine. You build the guardrails.

Hermes is the first platform to reverse that. We are the guardrails. Compliance is infrastructure, not an afterthought.

Agencies that adopt compliance-first infrastructure first will own their market verticals. Healthcare agencies will say: "We're HIPAA-aware and TCPA-compliant." Real estate agencies will say: "We call leads legally, with documented consent." Insurance agencies will say: "Our voice agents follow state rules per state."

Agencies that don't will get undercut by someone who does, or they'll face an FCC complaint and fold overnight.

FAQs

If I was already running outbound calls before today, am I automatically in violation?

No. The FCC hasn't retroactively enforced compliance. But if you did an audit today and found non-compliant calls, you now have knowledge of the violation. That makes you liable going forward. The safe move: audit, document your findings, and migrate to compliant infrastructure immediately. That posture is defensible if an FCC inquiry comes later.

Does "prior express written consent" mean I need a signature?

Written consent can be an email, a form submission, or a recorded phone call where the caller says yes. It needs to be specific: "I consent to be called by [Company Name] AI agents at [number]." A blanket "I agree to terms" isn't enough. Hermes tracks this at the call level.

What about inbound calls where the lead calls me?

Inbound is simple: if someone calls you, you can use AI from the start. No prior consent needed. The TCPA complexity is outbound-only.

If my client insists on using Retell, can I still operate?

Technically yes, but you're liable for compliance. Charge a $100-300/month compliance fee on top. Build your own consent tracking, DNC checks, and audit trail. Or recommend Hermes as a white-label platform where they own the infrastructure and you own the sales relationship. Don't absorb the compliance cost into your service fee.

When does the FCC start enforcing this?

The FCC doesn't have a specific enforcement date. But complaints are already coming in from consumers. Any class action law firm can manufacture a lawsuit within 60 days of an FCC confirmation. The safe assumption: enforcement is happening now.

Next Steps

Start here: BuildWithHermes. We'll walk you through a migration from your current stack (Retell, VAPI, etc.) to compliant infrastructure. Your Twilio numbers come with you. Your clients see no change. Your compliance posture becomes real.

Want to see how Hermes compares to your current stack? View our side-by-side comparison with Retell, VAPI, Synthflow, and Bland. We'll show you exactly where compliance is missing on those platforms.

Pricing: Starter $149/month (300 minutes, 3 workspaces), Business $399/month (1,000 minutes, 7 workspaces), Agency $699/month (2,000 minutes, 20 workspaces), Pro is Contact Us. A2P compliance submissions are $30 per agent.

The FCC just made compliance non-negotiable for outbound voice calling. Agencies that move first turn compliance into a competitive advantage. Agencies that wait get undercut or audited. The time to move is today.

Originally published at buildwithhermes.com/blog/fcc-confirms-tcpa-ai-voice-calling-agencies-2026-07-16.