ElevenLabs held its New York Summit on 11 November and announced 13 new products and upgrades — here is the split that matters if you build voice agents. The headline is that it stopped being a voice tool. It is now a platform for voice agents, and the announcements split into two groups: things you can use today, and things that are for large enterprises with a sales call in between.

I tested the tools that claim to do the same work I do, so here is the split that matters if you build agentic workflows.

The 3 that matter for builders

1. ElevenAgents Campaign Manager — live now. Announced 9 November, and it is the one I would try first. It runs outbound campaigns: you give it contacts and a goal — a qualified lead, a confirmed appointment — and it works each contact toward that goal, retrying when nobody picks up, calling back when someone asks for Thursday at 4pm, and stopping when a person opts out. It runs on the agents and phone numbers you already have in ElevenAgents, and contacts go in by CSV or API. Phone calls and WhatsApp voice and messaging work today; email and SMS are coming soon. It ran live beta campaigns across insurance, banking and ecommerce before launch.

This is the loop I build by hand in n8n today: call, check the outcome, retry on a schedule, keep the context between touches. Campaign Manager puts that loop inside the voice platform. The honest read is that the voice agent is the front door and something still has to do the work behind it — the CRM lookup, the booking, the follow-up. That is where n8n sits in my stack, and it is still where it sits after this launch.

2. Scribe v3 and Scribe v3 Live — live now. Announced 10 November. Scribe v3 is the recorded-audio transcription model, covering 102 languages. Scribe v3 Live is the low-latency real-time variant, with speaker labelling as it goes, and it is live in ElevenAgents today. Both are in ElevenAPI. If you run voice agents, transcription quality is the part that decides whether the agent understands the caller at all, and this is the piece of the Summit that is already in the product rather than on a roadmap.

3. ElevenAI — free early access, open now. A workplace app that listens to spoken conversations, including in-person meetings, and turns them into notes, summaries and action items. It can also give live assistance during a call. ElevenLabs built notice and consent mechanisms into it, which matters — a meeting recorder that does not ask first is a liability, not a feature. Early access is free and open with the announcement. This is the one to try with no budget and no sales call.

What is enterprise-only for now

The rest of the Summit is real, but it is not for a solo builder today, and I will not pretend otherwise.

The Marketplace is a curated catalog of partner-built industry apps — healthcare, financial services, insurance, education, retail, government — from more than 20 partners at launch, including Salient for borrower servicing calls, Gail by Lula for insurance inquiries, and Zendesk for customer service. You can browse the apps and contact the developers today. You cannot buy through it yet: purchasing through existing ElevenLabs commitments is planned to start in 2027.

Model customization lets selected enterprises work with ElevenLabs researchers to fine-tune open-source language models and ElevenLabs voice models on data from their own live deployments, tuned to metrics like customer satisfaction. It begins in the coming months, for selected enterprises first. Scribe v3 Medical — a transcription model built for healthcare terminology — already exists, and a financial-services model is in development. If you are not a large enterprise, this is a roadmap item, not a product you can sign up for.

The Interaction Engine is a preview: a system that manages turn-taking, interruptions and model selection inside a conversation. Memory (an agent keeps context across interactions, so callers stop repeating themselves) and Routines (multi-step work before, during and after a conversation) are coming to ElevenAgents. None of the three is shipped yet.

The Partner Program itself is now 550+ partners across 44 countries, including Google Cloud, IBM, Deloitte, DXC and Genesys. ElevenLabs also expanded its NVIDIA work — Nemotron 3.5 Super VL is in ElevenAgents, with 15 MW of Vera Rubin compute planned on top of existing Blackwell Ultra systems. A Unity SDK for ElevenAgents landed 5 November, so game players can talk to NPCs out loud — desktop and WebGL supported, iOS, Android and XR in beta. None of that changes what I build this week.

What worked

Campaign Manager is a real product, not a demo. It shipped after live beta campaigns, it runs on the agents and numbers you already have, and the retry and callback logic is the part that usually takes a builder a week to get right.

Scribe v3 Live is in the product today. Real-time transcription with speaker labelling, inside ElevenAgents, no waitlist mentioned.

ElevenAI early access is free and open. No sales call to try the one consumer-shaped product in the whole announcement.

The API is still the product. ElevenAPI carries the new models, and the whole stack is built to be called from outside — which is why it fits an n8n workflow instead of fighting it.

What broke — or rather, what is hard

Nothing about ElevenLabs broke this week. The hard part is the gap between the announcement and the product.

Three honest limits:

1. Most of the Summit is not shippable yet. Marketplace purchasing waits until 2027. Customization starts in the coming months for selected enterprises. Memory, Routines and the Interaction Engine are coming, not live. If you buy on the keynote alone, you are buying a roadmap.
2. Campaign Manager is phone and WhatsApp only for now. Email and SMS are coming soon, and new customers have to go through sales rather than signing up in the app. Existing ElevenAgents users find it under Campaigns in the Agents sidebar.
3. The enterprise pricing is undisclosed. No prices were given for the customization work or the Marketplace. The self-serve plans are public — Free at $0, Starter at $6, Creator at $22, Pro at $99, Scale at $299, Business at $990 a month — but the Summit's biggest items have no number attached.

None of that is a reason not to use it. It is a reason to use the three things that are live, and to wait on the rest until it ships.

Who it is for

ElevenLabs fits if you run voice agents in real workflows — outbound calls with a goal, meeting notes that need to be right, transcription in more than one language — and you want the voice layer to be an API you call from n8n rather than a dashboard you babysit.

It fits agencies that run outbound for clients: Campaign Manager is the first piece of this launch that does a job an agency bills for, today.

It does not fit if you only need text-to-speech for a video or a voiceover. The free tier covers that, and the Summit news is about agents, not voices. And if you are a solo builder waiting on the Marketplace or model customization, the honest answer is to try ElevenAI and Scribe v3 now and check back when the enterprise pieces ship.

The deal, plainly

ElevenLabs' programme pays 22% of payments on Starter, Creator, Pro and Scale plans for the first 12 months of a referred subscription, and 11% on the Business tier, through PartnerStack. Enterprise purchases earn nothing. Commissions clear about 90 days after the payment they come from. If you try it through my link, that is the arrangement: https://whoisalfaz.me/go/elevenlabs

And if the voice agent you build sits behind n8n schedules, check the 3.0 guide before December. The Cron node your workflows use today is gone in 3.0: https://whoisalfaz.me/blog/how-to-upgrade-to-n8n-3-without-breaking-workflows/

ElevenLabs is now a platform for voice agents, not a voice tool. Three of the 13 announcements are live and worth your time this week. The rest is a roadmap — a good one, but a roadmap.

Alfaz Mahmud Rizve — GTM & RevOps Architect. I build autonomous revenue systems for scaling agencies and SaaS: SEO & organic growth, high-speed Next.js apps, and self-healing n8n workflows. whoisalfaz.me

A crash alert can't catch a workflow that never starts. This is the heartbeat check I built for that gap.

I run a Watchtower: one global error-handling workflow, built on n8n's Error Trigger, that catches crashes across my whole instance and pings me with a deep link to the failed run. It is good at its job. The full build is here: https://whoisalfaz.me/blog/n8n-global-error-handling/

It has two blind spots. The 3.0 upgrade makes the first one urgent, and the second one has always been there.

Blind spot 1: the dropout

In n8n 3.0, workflows built on removed nodes — Cron, Interval, Function — register zero active listeners when the container boots. The dashboard still says Published. No execution ever starts, so the Error Trigger never fires. Nothing crashes, so nothing alerts.

This is the warning box on the Watchtower blog itself: a trigger dropout produces no execution, and no execution means no error event. Your error handler is working perfectly. It simply has nothing to catch.

The first sign of trouble is a customer asking where their invoice went.

Blind spot 2: the dead container

The second blind spot is older. If a workflow eats all the RAM — say, a Code node parsing 50,000 rows at once — the container can die at the operating system level. When the orchestrator itself is dead, the Error Trigger might not fire either, because the thing that would have fired it is gone.

The Watchtower blog is direct about this: hosting is part of the error-handling design, not an afterthought. That is why my production instances run with the V8 memory limit raised. But even good hosting does not close the gap — it only makes it rarer.

The check that covers both

A crash alert answers "did something fail?" A dropout needs a different question: "did something that should have run, run?"

That is a heartbeat check. It is a second, separate workflow that does one job: for each production workflow, look at when it last ran, compare that with how often it should run, and alert when the gap passes the baseline.

Build it like this:

1. One Schedule Trigger, on its own cadence. Every 15 minutes is a sane default. The watchdog runs on a clock, not on the workflows it watches.
2. Query the executions API. For each production workflow, pull its recent executions — each run's start time and status. The n8n API exposes this; you are reading the record, not guessing from the dashboard.
3. Compare against a baseline per workflow. An hourly workflow and a weekly workflow have different "overdue" lines. Record the median gap between runs for each workflow before you rely on the check — the same baseline step the 3.0 SOP recommends before upgrading.
4. Alert with the same shape as the Watchtower. Workflow name, expected interval, last run, and a deep link to its executions page. One message per overdue workflow, not a dump of the whole fleet.

That is the automated version of the manual check from the SOP: note the gap, watch it, and treat a gap that suddenly goes infinite as a silent dropout.

Two design rules

The watchdog lives on a separate instance from the one it watches. If the watched container dies, a monitor running inside it dies too. This is the whole point of blind spot 2. The simple second home is n8n Cloud — it runs the check, your self-hosted instance runs the work, and neither can take the other down: https://whoisalfaz.me/go/n8n

One threshold per workflow, never one global number. A global "alert if nothing ran in an hour" rule will spam you about the weekly report workflow and stay quiet about the hourly one that died 55 minutes ago. Baselines are per workflow or the check is noise.

The outer ring

The heartbeat check covers dropouts and dead workflows. It does not cover a dead instance, because the check itself needs an instance to run on. That is what the outer ring is for: a plain uptime ping on your instance URL, from outside, that tells you the box is down. Host-level, five-minute interval, no n8n involved.

Three layers, each catching what the one below cannot see:

• Watchtower catches executions that start and fail.
• Heartbeat check catches workflows that never start.
• Uptime ping catches an instance that is down entirely.

Build the Watchtower first — it is the fastest win and the SOP is linked above. Then build the heartbeat check before December, while the 3.0 beta is still out. The workflows that look fine after the upgrade are exactly the ones to check first: https://whoisalfaz.me/blog/how-to-upgrade-to-n8n-3-without-breaking-workflows/

Alfaz Mahmud Rizve — GTM & RevOps Architect. I build autonomous revenue systems for scaling agencies and SaaS: SEO & organic growth, high-speed Next.js apps, and self-healing n8n workflows. whoisalfaz.me

I send every n8n workflow's mail through Brevo. This is what that looks like after months of daily use.

I run Brevo every day. It is the email layer under my n8n work: transactional mail, CRM contacts, and lists, all driven by API from the workflows themselves.

That is the whole review in one sentence, but the reason it matters is the part people skip. Email is not the hard part of outbound. Deliverability is. And Brevo is a tool that makes you do the deliverability work properly, instead of pretending it does not exist.

What it does in my stack

Apollo finds the prospect. n8n moves the record. Brevo holds the contact and sends the mail.

The connection is the API, not the dashboard. My n8n workflows push contacts into Brevo with an upsert pattern — same contact in twice does not create a duplicate — and enrol them in lists by ID. The credential is a header auth key, and the calls are plain HTTP requests from n8n. Nothing in the chain depends on a browser session or a manual export.

That shape is why Brevo fits an automation stack. A tool that only works when a human clicks through its UI is a dead end for n8n work. Brevo's API is the product here, and the dashboard is where you check the results.

I also use it for transactional mail — the receipts, alerts, and confirmations that workflows send when something happens. Same account, same API, different job from the outreach lists. Keeping both in one place means one sender reputation to watch instead of two.

What worked

API-first, for real. Contact upserts, list enrolment, and sending all work over the API with clear success codes. My pipeline draft targets a 99.8% success rate on upserts, and the failures that do happen are visible in the response, not silent.

One place for mail and CRM. The contact, the list, and the send live in the same system. When a workflow updates a contact's job title or headcount, that change is in the record the next send uses. No sync job between two tools to break.

The warming discipline is written down. Brevo does not warm your IP for you, and it should not — warming is a schedule, not a switch. My SOP for it covers the setup that actually decides whether mail lands: SPF, DKIM, and DMARC records, a gradual volume climb, and n8n throttling so a workflow cannot blast a cold IP on day one. The full guide is here: https://whoisalfaz.me/blog/brevo-cold-email-ip-warming-guide/

What broke — or rather, what is hard

Nothing about Brevo broke. The hard part is the part no vendor can fix for you: a cold IP has no reputation, and inbox providers treat unknown senders as guilty until proven otherwise.

Three honest limits:

1. Warming takes weeks. Volume climbs on a schedule. If you need to send 10,000 emails on day one, Brevo will not save you — no tool will. The schedule in the SOP is the price of a working sender reputation.
2. A shared IP borrows reputation. Start on shared sending and your deliverability rides with other senders. A dedicated IP is the fix, and it is the thing that needs warming.
3. List hygiene is your job. Bad addresses and stale contacts hurt the sender score, and Brevo will not stop you from mailing them. The dedup and domain checks in my pipeline exist because dirty data costs reputation.

None of that is a reason not to use it. It is a reason to use it the way it is built to be used: API-driven, warmed slowly, lists kept clean.

Who it is for

Brevo fits if you run outbound or transactional mail from n8n and you are willing to do the warming work. It fits agencies that want mail and CRM in one API-driven place.

It does not fit if you want zero-setup bulk sending with no DNS records and no schedule. That buyer will be unhappy with any honest tool, Brevo included.

The deal, plainly

Brevo's programme pays €100 per paid account and €5 per free signup, after the first paid referral. If you try it through my link, that is the arrangement: https://whoisalfaz.me/go/brevo

The pipeline that connects Apollo to Brevo through n8n — webhooks, dedup, CRM sync, the copy-pasteable workflow — is written up here: https://whoisalfaz.me/blog/apollo-brevo-n8n-outbound-pipeline/

And if your Brevo mail rides on n8n schedules, check the 3.0 guide before December. The Cron node your workflows use today is gone in 3.0: https://whoisalfaz.me/blog/how-to-upgrade-to-n8n-3-without-breaking-workflows/

Brevo is the email stack I actually run. Not because it is magic, but because it is honest about the work — and the work is the whole game.

Alfaz Mahmud Rizve — GTM & RevOps Architect. I build autonomous revenue systems for scaling agencies and SaaS: SEO & organic growth, high-speed Next.js apps, and self-healing n8n workflows. whoisalfaz.me

How to Use Temporary Email for Verification Codes in 4 Simple Steps

You fill out a sign-up form, hit submit, then switch tabs and wait for a 6-digit code. Today, almost every online service asks for an email verification code — downloading a PDF guide, opening a free trial, joining a forum. Using a temporary email for verification codes gives you a clean way through: grab the one-time password, finish registration, and move on.

Step 1: Get your address

Open a disposable inbox generator. A fresh temporary address is created automatically — no account, no password, no personal details.

Step 2: Copy and paste

Copy the address and paste it into the registration form or app that is asking for your email.

Step 3: Keep the tab open

Stay on the inbox tab. It listens for incoming messages automatically, so the code appears as soon as it arrives. Browsers put background tabs to sleep to save battery, so if you switch away for too long, delivery can stall — keep the inbox tab in the foreground.

Step 4: Copy your code

Open the message, copy the 6-digit code (or click the activation link), and paste it into the sign-up form. Once you are registered, close the tab. The mailbox deletes itself.

Why Is My Verification Code Not Arriving?

Most verification emails arrive within 3 to 10 seconds. If yours is delayed, one of three things is usually responsible:

1. Sender greylisting. Some large mail servers deliberately delay the first message from an unknown sender by 2 to 5 minutes, as an anti-spam measure. A 10- or 15-minute inbox gives the email enough time to land.
2. Sleeping browser tabs. Chrome and Safari pause background tabs to save battery. If your inbox tab goes to sleep, new messages may not load until you click back to it.
3. Domain blocking. Some platforms check sign-ups against lists of known disposable domains and reject them outright. If that happens, generate a fresh address — a different domain often goes through.

Troubleshooting quick fixes

Which Platforms Accept Temporary Email?

Acceptance varies a lot by platform. Community sites such as Discord, GitHub, and Epic Games generally accept disposable addresses, with codes arriving in about 2 seconds. Gaming stores such as Steam often greylist mail, so expect a 2 to 4 minute delay. Social platforms like X may rate-limit disposable sign-ups — clicking "Resend code" once at the 60-second mark usually helps. High-security AI platforms run strict MX filtering and may reject disposable domains entirely.

When Should You Use a Burner Email for OTP?

Safe uses

• Public Wi-Fi portals — cafes, hotels, and airports that demand an email before giving you internet access.
• Free whitepapers and PDFs — content gated behind an email form.
• Software trials and sandboxes — test a tool without being enrolled in marketing emails.
• Gaming communities and forums — throwaway accounts for discussion boards.
• Discount codes — stores offering 10 to 15% off in exchange for a newsletter sign-up.

Never use it here

• Banks and payment accounts — PayPal, banks, crypto exchanges, and credit card portals.
• Primary social media — accounts holding your followers, photos, and personal messages.
• Work and school accounts — anything tied to your employer, school portal, or official records.
• Paid subscriptions — services where you will need to manage billing or reset a password later.

The rule of thumb: if you might need to click "Forgot password" six months from now, use your real email. A temporary inbox is for one-off access — if you want the full background, learn what a disposable email is and how it works.

Temporary Email vs. Temporary Phone Numbers (SMS)

People often confuse the two. Email verification sends a code or a link to an email address; SMS verification sends a text to a phone number. A temporary email service only receives email — it cannot receive text messages. Before generating an inbox, check whether the sign-up box asks for your email or your phone number.

Is Temporary Email Safe?

Used for one-off sign-ups, yes. The inbox is anonymous and self-destructs, so nothing is tied to your identity and no marketing follows you home. The risks come from misuse: never use it for accounts you need to recover, and never send sensitive information through one. For a deeper look at the security and legal side, see whether temporary email is safe.

Frequently Asked Questions

Can I use a temporary email for a verification code?

Yes. You can receive one-time passwords, activation links, and confirmation emails in a disposable inbox.

Why is my verification code not arriving?

The most common reason is sender greylisting — the sending server waits 2 to 5 minutes before delivering the first message.

Can a temporary email receive SMS codes?

No. Temporary email only receives messages sent over email protocols. It cannot receive text messages.

How long does a temporary email keep my verification code?

As long as the inbox timer runs — typically 5 to 30 minutes, depending on the service. Once the timer ends, the mailbox and its messages are deleted.

Can websites detect disposable email?

Some check sign-ups against databases of known disposable domains. If a site rejects your address, generate a new one with a different domain.

Can I recover an account created with a temporary email?

No. Once the session ends and the mailbox is destroyed, that inbox can never be reopened — and neither can the account's password resets.

Is it legal to use temporary email for verification codes?

Yes. Using one to protect your privacy and avoid spam is legal in most places, as long as you are not using it to deceive a service in a way that breaks its terms.