Coming soon on your Homenibox

What if your box answered on its own.

N8N runs locally on your Homenibox. It connects your pro switchboard (SMS, WhatsApp, calls) to any AI. Every message, every call can trigger an intelligent workflow — automatic reply, triage, transcription, follow-up. Without sending a single byte to a third-party service, unless you decide to.

Visual N8N workflows (drag & drop)
AI of your choice: Mistral, Llama, OpenAI, Anthropic
Data processed on your BOX, not in the cloud
Compatible with the existing Homenichat mobile app

Join the private beta See use cases

The concept in one sentence

Your Homenibox stops being just a phone switchboard. It becomes an automation engine wired into your customer channels. Everything stays at home, unless you decide to call an external AI.

1. Your pro channels already connected

VoLTE calls, modem SMS, official Meta WhatsApp, local WhatsApp Business. Everything is already integrated on the box — no need to glue cloud APIs together.

2. N8N as the conductor

N8N is the most widely used open-source visual workflow editor in Europe. We embed it in the box. You (or we) design your automations with a few clicks.

3. Optional AI, local or remote

Plug in Mistral locally on the box for 100% sovereignty, or point to OpenAI/Claude/Gemini for heavier tasks. You decide per workflow.

What it changes, concretely

Real examples we're building with our pilot customers. They all run as N8N workflows, triggered by an incoming event (SMS, call, WhatsApp).

Instant WhatsApp reply

A customer WhatsApps your business number to book. The AI detects the intent (booking, urgent, menu info), answers simple questions itself, or routes to the right teammate when it's complex. Round the clock.

Call transcription & summary

After every call (with legal consent), the box transcribes the audio locally, sends the text to the AI, and you receive an email: "Patient Dupont — request to reschedule appointment from Thursday to Tuesday 10am. No urgency detected."

Smart SMS triage to team

All incoming SMS are classified by AI: urgency (leaking roof), quote (cost request), info (opening hours). The right teammate gets the notification — not everyone, just whoever owns that scope.

Automatic sales follow-up

Property viewing finished? The box sends a satisfaction SMS 2 hours later. No reply within 48 hours? Auto follow-up. Positive reply? Notification to the agent + a WhatsApp thread opens for the next steps.

Weekly AI summary

Every Monday morning, you get a summary email: number of requests handled by category, average response time, files awaiting action, alerts flagged by the AI. Without lifting a finger.

Intelligent switchboard (IVR + AI)

Incoming call? The synthetic AI voice picks up: "Hello, you're calling practice X. Who would you like to leave a message for?" The message is transcribed, forwarded to the right contact, and the caller receives a confirmation SMS.

Auto-language detection

A tourist WhatsApps in English. The AI detects the language, replies in the same one, and translates it to French for you in the app. No more language barrier at the agency counter or restaurant.

Prescription photo understood

The patient sends a photo of their prescription on WhatsApp. The AI reads it, identifies the molecules, checks stock in your software, prepares the order. You approve with a single click.

Why self-hosted, not in the cloud

Because a medical practice, a law firm, a restaurant, doesn't send its customer conversations to obscure third-party servers.

❌ Classic cloud SaaS solutions

All your conversations transit through their cloud.
Your data trains their model.
If their AI provider raises prices, you pay.
If you cancel, you don't get your workflows back.
Customer data = their servers.

✅ Homenichat AI self-hosted

Everything runs on your Homenibox, at your premises.
You choose the AI: local Mistral (zero external send) or a third-party API (OpenAI, Claude, Gemini).
Your N8N workflows are exportable (JSON), you stay the owner.
Your customer conversations stay inside your BOX.
GDPR-compliant, fit for medical/legal professional secrecy.

The code runs as an encrypted binary on your Homenibox. N8N, the local AI model, your workflows: everything is on the box. You stay in control. You only send out what you decide to send.

Roadmap

We're building this module with pilot customers. Here's where we stand.

Available

Phase 0 — The foundation

Local SMS/WhatsApp/phone APIs already exposed by the box. Pro switchboard + multi-pickup + WhatsApp Business 3-way: all in production today.

Private beta Q3 2026

Phase 1 — N8N embedded

N8N runs on the box, accessible via a dedicated interface in the Homenichat app. 8 ready-to-activate workflow templates (WhatsApp auto-reply, SMS triage, call transcription, weekly summary). OpenAI/Claude/local Mistral connector.

Roadmap Q4 2026

Phase 2 — Real-time voice AI

Interactive voice switchboard driven by AI (natural synthetic voice, voice understanding, free-form conversation). The IVR evolves from "press 1, press 2" toward a real conversation.

Under study 2027

Phase 3 — Workflow marketplace

A shared library of N8N workflows by trade, created by Homenichat users. Install in one click what a fellow osteopath has already built.

AI & automations questions

Will it be included in Premium or a paid add-on?

The N8N module + local connectors will be included in Premium (EUR 19/month excl. VAT) at no extra cost. For external AIs (OpenAI, Claude…), you pay your own subscription directly with the provider — Homenichat doesn't bill for AI. For local AIs (Mistral 7B on the box), it's 100% free.

Do I need to know how to code to use N8N?

No. N8N is a visual drag-and-drop editor. You connect "nodes" together: "SMS received" → "Analyse with AI" → "If urgent → SMS to manager". 8 pre-made templates cover most use cases. For specific cases, our team can build the workflow with you (separate service).

Which local AI will actually run on the box?

The Homenibox's 4-core processor with 4 GB of RAM can run compact models (quantised Mistral 7B, Llama 3 8B, Phi-3). Speed: ~5-10 tokens/s locally. Enough for triage, short summaries, language detection. For heavy tasks (long-form writing, complex reasoning), switch to an external API.

What about GDPR / professional secrecy compliance?

In 100% local AI mode: no data leaves your box. Full GDPR compliance, medical secrecy, lawyer/notary professional secrecy. In external AI mode: you know what's sent, you validate it per workflow, and you can exclude specific data (social security numbers, patient records) from the payload before sending.

If I don't like it, can I disable it?

Yes, at any time. It's an optional Premium module. Toggle it off in the app and the box reverts to a simple pro switchboard. Your N8N workflows stay saved in case you change your mind later.

When exactly does the beta start?

Private beta announced for Q3 2026 (July-September). 50 pilot customers will be selected first among Premium subscribers who join the waitlist below. Public beta: Q4 2026. GA release expected early 2027.

Do you have concrete examples of workflows already built?

We have 3 templates currently being tested with our pilot customers: (1) Construction SMS triage (quote vs urgent vs info), (2) Property viewing follow-up + auto reminder, (3) Osteo appointment confirmation + D-1 reminder via WhatsApp. If you join the beta, you'll get direct access. And we're working on more as requests come in.

Join the private beta.

50 Premium pilot customers will help shape this module. Early access, exclusive templates, direct line with the product team. No extra cost.

Request beta access See the Premium plan

Private beta Q3 2026 · Premium customers only · No extra cost