Which languages does the SDK support?

Go is the official one. But the REST API works with any language (Python, Node, Ruby, etc.). Standard HTTP client.

How does multi-tenant work?

Each API key isolates the tenant. Rate limit, billing and data stay separated. Ideal for platforms offering agents to multiple customers.

Does it support audio and image?

Yes. Native multimodal: the agent receives audio, transcribes, analyzes images and replies with audio if you want. No extra orchestration.

Is there a free tier?

Yes. Free API key for testing. When you go beyond free volume, pick a paid plan.

Connects via MCP · Cursor, Claude Code, Gemini

Build an AI Assistanton WhatsApp in minutes.

Configure via Cursor, Claude Code or Gemini. No dashboard, no API key, no server.

No credit card7 days freeSign up in 1 minute

claude code · verboo mcponline

Where your weekends go

You're maintaining infra instead of building product.

n8n

self-hosted

03:47

✕ ERR Cannot read property "message" of undefined

One node updates, the flow breaks.

6h

weekend

Chatwoot

self-hosted

16:22

$ docker compose ps

chatwoot_puma.webup · 12d

chatwoot_sidekiqup · 12d

chatwoot_postgresexited 1

chatwoot_rediswaiting

chatwoot_nginxup · 12d

!postgres dropped connection pool

One service goes down, everyone waits.

3h

to get back online

Custom stack

DIY

23:04

stderr · agent.pyTraceback

File "main.py", line 42, in run

result = llm.parse(msg.content)

File "parser.py", line 18, in parse

return payload["choices"][0]["text"]

KeyError: 'text'

# LLM changed schema last week

!retry queue: 247 stuck jobs

LLM changed the schema. Rebuild everything.

15

open PRs

Two platforms, one account

Build on Lab.
Your client operates on Chat.

Platform

Verboo Lab.

Build the agent's brain in one place: system prompt, model, tools, and live playground. Create via MCP or the panel without losing state between the two.

Open Lab

verboo.ai/labLivia · B2B

Publicado

Nome

Livia · B2B

Gemma 4 31B

System prompt

Você é a Livia, SDR da Verboo.

Qualifica leads B2B em pt-BR, tom direto

e consultivo. Nunca mente sobre preços.

When you detect buying intent,

dispare o trigger purchase_intent.

Tools

create_leadsend_whatsappschedule_callsearch_kb

Temp

0.3preciso

verboo.ai/chat/boo-admin

291 leads· 12 quentes

C

Carla Mendes

+55 11 9•••• 2847 · WhatsApp

Auto

Oi! Queria saber quanto custa o plano pra 3 atendentes.

Oi, Carla! Pra 3 atendentes o plano Growth atende bem. R$ 637/mês com modelos em pt-BR inclusos.

Posso testar antes?

Pode sim — 7 dias grátis, sem cartão. Vou te mandar o link em instantes.

Take over the chat

Platform

Verboo Chat.

Your client's inbox: real-time conversations, hot leads at the top, handoff to human agents, and each contact's journey. You hand it off to the client's team and they run it themselves.

Open Chat

If you prefer direct code.

inicio-rapido.sh

# Uma chamada, um agente conversando. Endpoint do Generative API.

curl -X POST https://generative-api.verbeux.com.br/session/one-shot \

-H "api-key: $VERBOO_KEY" \

-F "assistant_id=865" \

-F "message=Quero saber sobre os planos" \

-F "channel=API"

# Resposta

# { "responses": [{ "type": "TEXT", "data": "Temos 3 planos ..." }] }

Three pt-BR models · no key required

Three brains. You choose.
We pay.

Verboo runs and pays for the models. You don't need an OpenAI or Anthropic key, and you don't manage a separate LLM invoice. Switch models in one line via MCP.

denseGeneralist

qwen-3.5-27b

Qwen 3.5

27B

The best balance of speed and quality. This is where we start.

p50 Latency

320ms

Context

128k

Languages

pt-BR · en · es

Start here

denseShort instruction following

gemma-4-31b

Gemma 4

31B

Doesn't stall on short instructions. Favorite for quick WhatsApp exchanges.

p50 Latency

280ms

Context

64k

Languages

pt-BR · multi

denseDense context

gpt-4.1-mini

GPT-4.1

mini

For when the history gets deep or the question becomes a rabbit hole.

p50 Latency

410ms

Context

128k

Languages

pt-BR · multi

Dense beats MoE on WhatsApp

Fast exchanges
need a dense brain.

MoE excels at long context. But on WhatsApp, messages are two lines, and expert routing adds latency. Dense models deliver the first token on time and follow instructions consistently. That's what converts.

p50 Latency · pt-BR · WhatsApp

Dense 27-31B (Verboo)310ms

GPT-4.1 mini410ms

MoE 45B-A13B typical720ms

Internal test · 10k msgs pt-BR · p50

Direct comparison

Verboo vs what you have now.

Criterion	Verboo	n8n	Chatwoot	DIY
Time to first agent live	5 min	2 to 6 h	2 to 8 h	1 to 3 days
Official WhatsApp Business		separate BSP	manual	you build it
Server to maintain	none	Docker + PG + Redis	Docker + PG + Redis + Rails	everything
Models included
MCP native		plugin		you implement
Human inbox				you build
Update breaks production		sometimes	sometimes	depends
Commercial license	clear	SUL restricted	paid	free
Export without lock-in	via API	JSON	DB dump	your code
Server cost per month	$ 0	$ 20–100	$ 40–150	$ 50+

Use cases

Three ways to plug in
to what you already do.

Native integrations for common cases, your own MCP for everything else. Click a case to see how it works in practice.

Come in to code again.
Leave whenever you want.

Open SDK on GitHubMCP · open standardReal-time webhooksExport via APISelf-hosted availableLGPD · data in Brazil

Total cost, on the table

Verboo isn't cheap.
Maintaining infra is expensive.

Hours / week maintaining n8n, Chatwoot or custom stack6h

0h30h

Your hourly rateR$ 150

R$ 30R$ 500

Clients / projects with chatbot today3

120

Compare against plan

You burn / month

R$ 11.691

26.0h × R$ 150 × 3 clients

Verboo Growth

R$ 637/ mês

Models, infrastructure, and support included.

Savings R$ 11.054 / month

You pay 18.4× more in time than Verboo costs

Migration

Twelve nodes become
one prompt.

The same flow you wire by hand in n8n, Verboo orchestrates in a single conversation.

antes

n8n self-hosted

12 nodes · 4h setup

workflow quebrou às 03:14

depois

Verboo MCP

1 prompt · 5 min

Gemma 4

RD Station

Postgres

Retry

Verboo

agente no ar · 100% uptime

complexidade

12 nodes→1 prompt

setup

4 horas→5 min

infra

R$ 400/mês→R$ 0

Need help migrating? We'll do it together →

Technical FAQ

Questions you were going to ask.

Full export via API: agents, prompts, conversations, leads. MCP is an open standard. Any MCP client talks to any server that implements the protocol.

No. We run the models for you, by design. This eliminates configuration, duplicate billing, and the paralysis of choosing a model. The three included models cover 99% of cases.

Yes. Direct BSP with Meta. Pre-approved templates, multiple numbers on the same account, handoff to a human agent. We also support Evolution API and Z-API.

A panel with the timeline of each session, real-time event webhooks, and conversation export. Everything inspectable via MCP.

Yes, on the Enterprise plan. Docker and Kubernetes on your own cloud. LGPD, SOC2, data sovereignty.

Plans

Models included.
Zero separate LLM billing.

7-day free trial on all plans. No card. Cancel anytime.

Starter

First tests and MVPs

R$ 247/mo

Conversations

300/mo

Agents

3

3 pt-BR models included
REST API + Go SDK
MCP integrated
Webhooks
Community support

Growth

Production with traction

R$ 637/mo

Conversations

1.000/mo

Agents

5

Everything in Starter
Official WhatsApp Business
On-the-fly model switching
Advanced triggers
Email support

Scale

Full-scale operation

R$ 1677/mo

Conversations

2.500/mo

Agents

10

Everything in Growth
All channels
Native CRM integrations
Advanced analytics
Priority support

Enterprise

Critical scale, on-premise, SLA

Custom

Conversations

Unlimited

Agents

Unlimited

Everything in Scale
On-premise Docker+K8s
99.9% SLA
SSO / SAML
24/7 support

Deploy your first Assistant
in 5 minutes.

Free 7 days. No card. Models included.

Build an AI Assistanton WhatsApp in minutes.

You're maintaining infra instead of building product.

Build on Lab.
Your client operates on Chat.

Verboo Lab.

Verboo Chat.

If you prefer direct code.

Three brains. You choose.
We pay.

Qwen 3.5

Gemma 4

GPT-4.1

Fast exchanges
need a dense brain.

Verboo vs what you have now.

Three ways to plug in
to what you already do.

Catalog and sales on WhatsApp

Clinical scheduling with Feegow

Integration with your system

Come in to code again.
Leave whenever you want.

Verboo isn't cheap.
Maintaining infra is expensive.

Twelve nodes become
one prompt.

Questions you were going to ask.

Models included.
Zero separate LLM billing.

Starter

Growth

Scale

Enterprise

Deploy your first Assistant
in 5 minutes.

Build an AI Assistanton WhatsApp in minutes.

You're maintaining infra instead of building product.

Build on Lab.Your client operates on Chat.

Verboo Lab.

Verboo Chat.

If you prefer direct code.

Three brains. You choose.We pay.

Qwen 3.5

Gemma 4

GPT-4.1

Fast exchangesneed a dense brain.

Verboo vs what you have now.

Three ways to plug into what you already do.

Catalog and sales on WhatsApp

Clinical scheduling with Feegow

Integration with your system

Come in to code again.Leave whenever you want.

Verboo isn't cheap.Maintaining infra is expensive.

Twelve nodes becomeone prompt.

Questions you were going to ask.

Models included.Zero separate LLM billing.

Starter

Growth

Scale

Enterprise

Deploy your first Assistantin 5 minutes.

Build on Lab.
Your client operates on Chat.

Three brains. You choose.
We pay.

Fast exchanges
need a dense brain.

Three ways to plug in
to what you already do.

Come in to code again.
Leave whenever you want.

Verboo isn't cheap.
Maintaining infra is expensive.

Twelve nodes become
one prompt.

Models included.
Zero separate LLM billing.

Deploy your first Assistant
in 5 minutes.