Current version: 1.16.0 ·Instagram available ·More channels in preparation

Never miss an enquiry again.

KISSA_AI watches your company’s social channels, tells real prospects from spam and praise, drafts a reply from your own knowledge, and queues it for approval. Optionally the agent replies on its own.

Instagram Multi-tenant Approval or full autonomy Self-hosted
Placeholder: inbox

Image placeholder for the inbox hero shot. Replace with hero-inbox.png.

What KISSA_AI does

👀

Detect

Every inbound message and comment is classified: prospect, support, praise, criticism, spam.

🧠

Understand

Intent, urgency and purchase interest are scored and stored with a rationale.

💬

Reply

Drafts use only your knowledge base. What is not stored is not claimed.

Approve

On level 1 nothing leaves without a human. Raise autonomy only when quality is proven.

🔔

Follow up

A watchdog flags stalled work before reply windows expire.

🤝

Escalate

Complaints, legal topics, press and aggressive messages never get an automated reply.

🧾

Prove

Every step is logged immutably — from intake to the message actually sent.

🏢

Isolate

Multi-tenant by design. Separation is enforced in the database.

Two ways of working

With approval

Recommended start. Every new account begins here.

Autonomous

Replies without asking, under your rules — once drafts consistently pass unchanged.

You can switch per account at any time. Test mode publishes nothing; an emergency stop halts all sending.

Channels

Instagram available

Comments and DMs. Professional account linked to a Facebook Page required.

Facebook Page in preparation

Same Meta access as Instagram. Adapter not built yet.

WhatsApp Business in preparation

Only channel that can reopen a conversation after days via approved templates.

TikTok access unclear

Comment endpoint scope still unverified. No promises until checked.

YouTube in preparation

Feasible; limited by Google daily quota.

LinkedIn on request

Paid partner programme only.

That a machine is replying is stated upfront

Since 2 August 2026, Article 50 of the EU AI Act requires disclosing that a person is talking to an AI system. KISSA_AI enforces this: the first reply in a conversation is prefixed with your wording. A reply claiming to be a human is rejected and never sent — even with approval.