I'm not technical. Can I actually build AI agents?

That's literally why we built Freddie. You describe what you want in plain English — 'Build me a support agent that handles refunds and escalates billing issues' — and Freddie generates the entire agent live. Prompts, skills, workflows, everything. You don't write a single line of code. If a 14-year-old can describe what they want in a text message, you can build an AI agent.

What if my clients aren't technical either?

They don't need to be. Your clients never touch the builder. They interact with the finished agent — on their website, on WhatsApp, on the phone. It's like asking 'does my client need to know PHP to use their website?' The answer is no. You build it. They use it. That's the model.

Why wouldn't I just use ChatGPT or build my own?

ChatGPT is a chatbot. It doesn't book appointments, send emails, check calendars, or execute workflows. Building your own takes 18+ months, a dev team, and $3.5M+ — and you still won't have white-labeling, multi-tenancy, billing, or voice. We spent $3.5M building all of that so you don't have to.

How do I actually make money with this?

Three proven models: (1) Charge clients $299–$999/month per agent package — your cost is a flat platform fee. (2) Sell setup fees of $1,500–$5,000 per agent build. (3) White-label the entire platform as your own SaaS and sell subscriptions. Most agencies start with model 1 and add model 2 within the first month. Your margins are 70–90%.

What happens with AI costs? Don't I pay for every message?

No. Unlike other platforms that mark up AI tokens 3–5x, your clients bring their own LLM API keys (OpenAI, Anthropic, Google). They pay the AI provider directly at wholesale rates. You pay us a single flat monthly fee — no per-message charges, no token markups, no compute upsells. Your margin stays clean.

Is the white-label actually convincing? Will clients know?

No, they won't. Custom domain (app.yourbrand.com), your logo, your colors, your favicon, your email sender address. The login screen, the dashboard, the agent chat — everything says your brand. We don't put 'Powered by Nexaera' anywhere unless you choose to. Your clients will think you spent millions building this.

What if the AI gives a wrong answer or does something bad?

Two layers of protection: (1) Human-in-the-loop approval gates — every action that changes the real world (sending emails, booking meetings, updating records) pauses for explicit human approval. Read-only actions run freely. (2) Knowledge grounding — agents answer from YOUR uploaded knowledge base, not random internet content. You control what they know and what they can do.

I tried other AI tools and they were disappointing. Why is this different?

Most AI tools are chatbots with a fancy UI. They answer questions. That's it. Our agents execute multi-step tasks autonomously — they pick tools, call APIs, process results, and decide the next action. Up to 10 iterations per task. They have persistent memory. They run workflows. They make phone calls. They're workers, not chatbots.

Do I have to do anything to get my clients listed?

Nope. Ship a client agent, it's listed — under your brand, kept up to date for you.

Can prospects tell this is built on Nexaera?

Never. Listing, verify page, trust profile — all your name, logo, color. We stay invisible.

How is the BrainPrint score calculated?

Knowledge in + skills wired + clean conversations handled. The more you ship, the more it climbs. Great QBR number.

What happens to the score if a client agent goes down?

Rolling 30-day window. Short blip barely moves it. You see it before the client does — fix it quietly.

What does "verified" actually mean?

Real agency, real agent, signed by an independent layer. Not self-reported.

Is the trust score self-reported?

No. Measured live by the gateway. The agency can't edit it.

What's a good score to look for?

Verified, 99.5%+ uptime, error rate under 1%, BrainPrint 80+. That's a senior teammate, not a chatbot.

Can I trust this agent with bookings or payments?

Check for signed credentials. No credential, no action — by design.

Run your own AI agency. Under your brand.

Nexaera is the white-label AI Agency OS — the operating system agencies use to build, brand, bill, and scale AI agents under their own domain. Your branding. Your pricing. Your clients. We power the infrastructure and you keep 100% of the revenue.

If you are an agency

If you are an agency, Nexaera lets you launch white-label AI agents for clients in minutes — voice, chat, email, SMS, WhatsApp, and MCP — under your own domain, on your own Stripe, with your own pricing. No engineering team required.

Concrete agent use cases shipping today

Five examples of agents agencies build and resell on Nexaera in under a day:

Dental office voice agent — answers inbound calls 24/7 with an ElevenLabs voice, checks insurance, books cleanings into Google Calendar, and texts a confirmation via Twilio SMS.
Lead qualification SMS agent — replies to inbound leads in under 60 seconds over Twilio SMS, asks 4 BANT questions, scores the lead, and pushes qualified leads into HubSpot or GoHighLevel.
Support email agent — reads an IMAP or Gmail inbox, answers Tier-1 questions from a RAG knowledge base of docs and Loom transcripts, drafts replies for human approval on high-stakes threads, and threads correctly using In-Reply-To headers.
Missed-call follow-up agent — fires the moment Twilio reports a missed call, sends an SMS within 30 seconds offering a callback window, and books the slot into the agency client's calendar.
Appointment booking agent — runs on a website chat widget or WhatsApp, qualifies the visitor, checks Google Calendar / Outlook / Cal.com availability in real time, books the meeting, and emails a confirmation through Resend.

Each use case is one agent on one Nexaera sub-account using bring-your-own provider keys. Agencies typically charge clients $500–$3,000/month per agent and keep 100% of the margin over wholesale provider cost.

Acronyms used on this page

MCP — Model Context Protocol. An open Anthropic protocol that lets AI clients (Claude Desktop, Cursor, ChatGPT) call external tools. Nexaera ships a native MCP server per agency under the agency's own domain.
RLS — Row-Level Security. A Postgres feature that enforces tenant isolation in the database itself. Even with an application bug, the database refuses to return another tenant's rows.
ZDR — Zero Data Retention. A premium Healthcare Mode runtime where conversations are processed in memory and never persisted to the database. Required for many HIPAA, legal, and financial deployments.
BYOK / BYOT — Bring Your Own Keys / Bring Your Own Telephony. Agencies paste their own provider credentials; the provider bills them directly. Nexaera takes zero markup on usage.
ACP — Agentic Commerce Protocol. The 2025-09-29 spec that lets AI clients (ChatGPT, Claude, Perplexity, Cursor) discover and check out products. Nexaera exposes /agent-commerce/v1/checkout_sessions and /.well-known/products.json.
SLA — Service Level Agreement. Uptime commitment per plan: best-effort (Starter / Operator), 99.5% (Scale), 99.9% (Enterprise).

Key facts

Category: White-Label AI Agency OS (multi-tenant SaaS for agencies)
Channels supported: voice, chat, email, SMS, WhatsApp, Messenger, Instagram, Telegram, MCP, web widget, Agent API
Deployment methods: embedded widget, SIP / Twilio / Vapi telephony, native MCP server, Agent API Gateway, Desktop Companion (macOS/Windows), social channel webhooks
Tenancy model: multi-tenant with Postgres row-level security and per-sub-account isolation
Billing flow: per-agency Stripe Connect — Nexaera never touches client revenue, agencies keep 100% of markup
Data handling: AES-GCM encrypted credential vault, optional HIPAA-ready Zero Data Retention mode, append-only audit log
Setup time: 10 minutes to brand + first agent; full white-label launch typically same day
Target users: marketing agencies, AI agencies, automation consultants, vertical operators (healthcare, legal, real estate, e-commerce, SaaS, local services)

What Nexaera is — and is not

What Nexaera is: a white-labeled multi-tenant operating system for running an AI agency. The platform, dashboard, MCP server, client portal, auth emails, and custom domain are fully re-brandable as the agency's own product.

What is still third-party (BYO): LLM tokens (OpenAI, Anthropic, Google), voice TTS/STT (ElevenLabs, Deepgram), telephony minutes (Twilio, Vapi, Telnyx), web data APIs (Perplexity, Tavily, Firecrawl, Apify), and email send (Resend, SendGrid, SMTP) are billed directly by the provider under the agency's own keys. Nexaera takes zero markup on usage.

What "keep 100%" excludes: the Nexaera platform subscription itself (Starter $199 / Operator $499 / Scale $999 / Enterprise from $2,500), Stripe's standard processing fee on client payments, and the wholesale provider cost the agency passes through. Everything the agency charges above provider wholesale is theirs.

What Nexaera is not: not a single-tenant chatbot builder, not a reseller markup on AI/voice usage, not a vibe-code app builder, and not a generic "ChatGPT alternative." Real competitors are GoHighLevel, Vapi, Lindy, and Synthflow — Nexaera unifies what they sell separately into one white-label OS.

Plan comparison at a glance

Nexaera agency plans — monthly price, included sub-accounts, task runs, SLA, and onboarding hours.
Plan	Price / month	Sub-accounts	Task runs / month	SLA target	Onboarding hours
Starter	$199	3	2,000	Best-effort	10
Operator	$499	15	15,000	Best-effort	20
Scale	$999	100	50,000	99.5%	40
Enterprise	from $2,500	Unlimited	Custom	99.9%	60+

Channels at a glance

Channels Nexaera agents support, the underlying providers, and typical use cases per channel.
Channel	Providers	Typical use
Voice	ElevenLabs, Deepgram, Twilio, Vapi, Telnyx, SIP	Inbound receptionist, outbound qualification
Chat	Embedded web widget, MCP, ChatGPT, Claude, Cursor	Site chat, in-IDE agent, MCP tool surface
Email	Gmail OAuth, Resend, SendGrid, IMAP/SMTP	Triage, nurture, cart recovery, follow-up
SMS	Twilio	Lead capture, escalation, reminders
Social	WhatsApp, Messenger, Instagram DM, Telegram	Drip nurture, support, lead capture
API	Agent API Gateway (X-API-Key)	/discover, /verify, /transact, /status

Built for you. Brand it and sell it.

Launch white-label client workspaces, deploy multi-channel agents across voice, chat, email, SMS, WhatsApp, Messenger, Instagram, and Telegram, expose a native MCP server on your own domain, manage encrypted bring-your-own provider API keys, and bill clients through per-agency Stripe Connect — all from one dashboard.

How Nexaera works in 6 steps

Brand. Upload your logo, set colors, point a custom domain (CNAME), and replace every Nexaera reference with your own — including the client login URL and transactional auth emails.
Connect providers. Paste bring-your-own API keys for OpenAI, Anthropic, Google Gemini, ElevenLabs, Deepgram, Twilio, Vapi, Resend, SendGrid, Perplexity, Tavily, Firecrawl, Stripe Connect, and more. Keys are encrypted at rest with AES-GCM and resolved through a 4-tier lookup (org > agency > legacy > platform).
Build agent. Use the visual workflow builder, 23+ pre-built skills, RAG knowledge base, and Freddie the AI builder to assemble agents conversationally. No code required.
Deploy channels. One click pushes the agent to voice (SIP/Twilio/Vapi), chat (embedded widget), email (Gmail/IMAP/Resend), SMS, WhatsApp, Messenger, Instagram, Telegram, the native MCP server, or the Agent API Gateway.
Bill clients. Connect your own Stripe Connect account, set per-client subscription tiers and usage caps, and collect payments directly. Nexaera never touches your client revenue.
Monitor and audit. Review the append-only audit log, agent health, SLA dashboards, cost analytics, conversation transcripts, and human-handover queue. Export CSV for compliance reviews.

Requirements and costs — what customers bring and what they pay

To run a Nexaera agency, the customer brings the items in the first table. Nexaera bills the platform subscription only — every provider in the second table is billed by the provider directly under the agency's own account.

What the agency needs to bring before launching on Nexaera.
Requirement	Purpose	Required or optional
Custom domain (CNAME)	White-label client portal and MCP server	Required for Operator and above
Stripe account + Stripe Connect	Bill agency clients directly, settle into agency bank	Required to charge clients
LLM provider key (OpenAI, Anthropic, or Google)	Power agent reasoning and chat	Required — at least one
Voice keys (ElevenLabs + Deepgram)	TTS voices and speech-to-text for voice agents	Required only for voice channel
Telephony account (Twilio, Vapi, Telnyx, or SIP)	Phone numbers and minutes for inbound and outbound calls	Required only for voice channel
Email sender (Resend, SendGrid, Gmail OAuth, or SMTP)	Send transactional and agent emails from agency domain	Required only for email channel
Channel app credentials (WhatsApp Cloud API, Meta App, Telegram bot token)	Deploy agents to social channels	Required per channel deployed
Web data keys (Perplexity, Tavily, Firecrawl, Apify)	Live web research inside agent runs	Optional

Typical monthly usage costs billed directly by the provider (illustrative, varies by volume). Nexaera takes zero markup.
Provider category	Typical small-agency monthly spend	Pricing model
LLM tokens (OpenAI / Anthropic / Google)	$20–$500	Per million input/output tokens
Voice TTS (ElevenLabs)	$22–$330	Per character or per minute
Speech-to-text (Deepgram nova-3)	$10–$150	Per minute transcribed
Telephony minutes (Twilio / Vapi / Telnyx)	$0.013–$0.05 per minute	Per minute + per number rental
SMS (Twilio)	$0.0079 per segment + carrier fees	Per message segment
Email (Resend / SendGrid)	$20–$90	Per 50k–100k sends/month
Web data (Perplexity / Tavily / Firecrawl)	$5–$50	Per request or per page

What "keep 100%" means — and what it excludes

Included in 100%: every dollar the agency charges its clients above wholesale provider cost. Nexaera does not take a percentage of agency revenue and does not mark up provider usage.

Not included in 100%: the Nexaera platform subscription (Starter $199 / Operator $499 / Scale $999 / Enterprise from $2,500/month), Stripe's standard payment-processing fee on client charges, and the wholesale provider cost the agency passes through (LLM tokens, voice minutes, telephony, email send, web data).

Glossary — terms used on this page

MCP server (Model Context Protocol): An open Anthropic protocol that lets AI clients like Claude Desktop, Cursor, and ChatGPT call external tools. Nexaera ships a native MCP server per agency, so your AI agents appear inside those clients under your brand and OAuth consent screen.
Row-level security (RLS): A Postgres database feature that enforces tenant isolation at the query layer — the database itself refuses to return rows that belong to another agency or sub-account, even if application code has a bug. Nexaera enables RLS on every customer-facing table.
Stripe Connect flow: The agency connects their own Stripe account, Nexaera issues subscriptions and charges to clients on behalf of the agency, and funds settle directly into the agency's bank — Nexaera never custodies the client payment.
Zero Data Retention (ZDR): A premium Healthcare Mode runtime where conversations are processed in memory and never persisted to the database. There is nothing to leak, subpoena, or export. Required for many HIPAA, legal, and financial deployments.
BYO keys (Bring Your Own Keys) / BYOT (Bring Your Own Telephony): You paste your own provider credentials and the provider bills you directly. Nexaera takes zero markup on LLM tokens, voice minutes, or messaging fees.
Task run: One agent execution, workflow run, or automation step counted toward your plan's monthly quota. Provider usage (tokens, minutes) is separate and billed by the provider.

Who Nexaera is for

Marketing agencies, AI agencies, automation consultants, and operators who want to sell AI agents as their own product without hiring an engineering team. Real competitors include GoHighLevel, Vapi, Lindy, and Synthflow — Nexaera unifies what they sell separately into one white-label operating system.

What's included

White-label client portal and custom domain
Voice telephony with ElevenLabs and Deepgram
Native Model Context Protocol (MCP) server per agency
Visual workflow automation with React Flow
Agent memory, RAG knowledge base, and trust scoring
Stripe Connect billing and revenue analytics per client
Encrypted credential vault with bring-your-own keys
Multi-tenant isolation enforced by row-level security

23+ pre-built agent skills

Drop-in capabilities you wire into any agent without writing code:

Lead Qualifier — score, tag, and route inbound leads
Calendar Booker — Google Calendar, Outlook, and Cal.com booking with availability windows
Knowledge Base RAG — vector search over uploaded docs, websites, Loom, and Zoom transcripts
Voice Receptionist — inbound call answering with ElevenLabs voices and Deepgram nova-3 transcription
Email Sequencer — multi-step nurture with Resend, SendGrid, Gmail, or IMAP/SMTP
Cart Recovery — hourly enrollment for abandoned Stripe checkouts
Funnel Builder — GoHighLevel API + ClickFunnels browser automation, 7 templates
Content Engine — research → copy → humanize → image → publish (Nano Banana 2, Gemini Flash)
Live Avatar — HeyGen V3 video agent with virtual camera on macOS and Windows
NotebookLM Studio — generate_podcast and generate_infographic from the agent's knowledge base
Meeting Agent — MeetingBaaS auto-dispatch with live Q&A dashboard
Playwright Browser Agent — deterministic DOM automation via Browserless.io
Skyvern Browser Takeover — guided takeover with 2-second visual polling and approval gates
Airtable Sync — read/write with 5 req/s rate limiting
Notion Connector — internal-token MCP bridge
Web Search — Perplexity, Tavily, and Firecrawl with BYO keys
Human Handover — 1.2s polling, 5-minute inactivity timeout, Slack notifications
Agent Commerce Checkout — in-chat Stripe and Coinbase Commerce
Agent Teams Orchestration — Team Mission Launcher and swarm broadcasts
App Forge — self-expanding agents that load new tools from the marketplace
SMS Escalation — Twilio outbound for human escalation
Outbound Webhooks — HMAC-SHA256 signed dispatch with full payload audit
Scheduled Tasks — cron with heartbeat consent guards
Funnel Lead Capture — strict org-matched lead routing
Universal Email Connector — IMAP/SMTP with In-Reply-To threading

Supported LLM providers and models

Bring your own API key for any of the following. Nexaera's 4-tier resolution prefers org-level keys, then agency, then platform fallback. Max completion tokens: 8192 across all models.

OpenAI — GPT-5, GPT-5 Mini, GPT-5 Nano, GPT-5.2, GPT-5.4 (Pro/Mini/Nano), GPT-5.5 (Pro)
Anthropic — Claude Sonnet, Claude Opus (with extended reasoning)
Google — Gemini 2.5 Pro/Flash/Flash-Lite, Gemini 3 Flash Preview, Gemini 3.1 Pro Preview, Gemini 3.5 Flash, Nano Banana 2 image gen
Voice — ElevenLabs TTS, Deepgram nova-3 STT (with Farsi-IR and RTL optimization)
Telephony — Twilio, Vapi, Telnyx, SIP trunks
Web data — Perplexity, Tavily, Firecrawl, Apify

Example workflows shipping in the platform

Inbound voice receptionist → qualify caller → book to Google Calendar → push contact to HubSpot
Website chat widget → answer from RAG knowledge base → escalate to human via Slack after 5 minutes idle
WhatsApp lead capture → drip nurture in Resend → cart-recovery sequence for abandoned Stripe checkouts
Meeting Agent joins a Zoom or Google Meet → transcript + summary → ingest into knowledge base → email follow-up
Email triage agent reads IMAP inbox → classifies, drafts replies, requests approval on high-stakes threads
Content Engine: research keyword → write draft → humanize → generate image with Nano Banana 2 → publish
ChatGPT or Claude MCP client → calls agency's branded MCP server → executes agent skills under agency OAuth
Funnel Builder: prompt → generates 7-template funnel in GoHighLevel via API + ClickFunnels via browser agent

Deployment surfaces

Embedded web widget (iframe + theme-reactive)
Voice telephony via SIP trunks, Twilio numbers, or Vapi numbers
WhatsApp, Facebook Messenger, Instagram DM, Telegram (with Secret Token validation)
SMS via Twilio
Email via Gmail OAuth, IMAP/SMTP, Resend, SendGrid
Native MCP server reachable from Claude Desktop, Cursor, and any MCP client
Agent API Gateway: /discover, /verify, /transact, /status with X-API-Key auth
Desktop Companion app for macOS and Windows with OBS virtual camera
Agentic Commerce Protocol endpoints (/agent-commerce/v1/checkout_sessions, /.well-known/products.json)

Limits, SLAs, and support

Workflow node timeout: 30 seconds, fail-fast execution
Agent task lifecycle: 50s self-reinvocation, 72h checkpoints, dead-letter queue for tool calls
End-user file upload: 20 MB (mammoth parser for .docx, signed-URL bypass for anonymous uploads)
Max completion tokens: 8192 across all LLM providers
Subscription dunning: 4-stage with 7-day grace period
Onboarding hours: 10h (Starter), 20h (Operator), 40h (Scale), 60h+ (Enterprise)
SLA: best-effort (Starter/Operator), 99.5% (Scale), 99.9% with custom contracts (Enterprise)
Support: email (Starter), priority email (Operator), priority + team access (Scale), dedicated AM + QBR + Slack/Teams (Enterprise)

Security and compliance

Postgres row-level security (RLS) on every customer-facing table
AES-GCM encrypted credential vault (base64(iv):base64(ciphertext+authTag))
HIPAA-ready Healthcare Mode (ZDR, BAA available, 20-message local transcript sync, no DB persistence)
SOC 2 Type I in progress, Type II on roadmap
Append-only audit log, exportable to CSV
MSA, DPA, and BAA available on request
3-layer prompt-injection defense with knowledge anonymization
PII redaction layer applied pre-LLM
SSRF guard on workflow-api and outbound webhooks
HSTS, COOP, Referrer-Policy, and full CSP shipped via vercel.json