Your clients ask the same questions. Deploy an AI agent on all your channels.

A SkaLean agent can be deployed on up to 10 channels simultaneously with the same configuration, knowledge base, and personality:

1. WhatsApp Business: the preferred communication channel in Quebec for B2C interactions.
2. Phone (AI voice): via SIP/WebRTC, compatible with all modern PABX/IP systems and Twilio. Voice response time under 1 second.
3. Website: chat widget embeddable with 2 lines of JavaScript code on any website.
4. Incoming email: automatic analysis and response to emails based on content.
5. SMS: for reminders, confirmations, and short communications.
6. Instagram Direct: for businesses active on Instagram.
7. Facebook Messenger: for Facebook audiences.
8. Slack or Teams (internal): an agent for your team's internal questions.
9. Direct API: for integration into your own applications.

Conversational memory is shared across channels: if a client starts on WhatsApp then calls, the agent already knows the WhatsApp conversation history.

For a basic agent (client FAQ, appointment booking, human escalation): 5 to 7 business days from the start of onboarding. Step breakdown:

• Days 1-2: agent configuration (personality, name, tone, system instructions, allowed and forbidden topics).
• Days 2-3: connection to your knowledge base (indexing your existing FAQ documents, service procedures).
• Days 3-4: channel integration (WhatsApp Business API, website widget, telephony).
• Days 4-5: internal testing with your team (client question simulation, escalation testing, adjustments).
• Days 5-7: progressive production launch (activated for 10% of traffic, then 100%).

For a more complex agent (multi-channel, CRM and telephony integrated, complex escalation workflows): 10 to 15 days. The main delay is often waiting for WhatsApp Business account approval (Meta can take 2 to 5 business days) — initiate this process in parallel from day one.

The agent isn't "trained" in the model training sense — it's configured. Five configuration elements define its response quality:

1. Knowledge base (Studio AI): your internal documents (FAQ, procedures, product sheets, pricing policies) constitute the primary source of truth. The agent responds based on these documents, not general knowledge.
2. System prompt: precise instructions on personality, allowed topics and forbidden ones.
3. Pre-validated responses: for sensitive or frequent questions, you write responses in advance. The agent uses them as-is, without reformulating.
4. Confidence threshold: for each response, the agent calculates a confidence score. Below 70%, it transfers to a human rather than "guessing."
5. Continuous improvement loop: every conversation is accessible in the dashboard. You identify questions the agent answers poorly, add content to the knowledge base, and quality improves progressively.

SkaLean implements a 3-level uncertainty management system, configurable to your preferences:

1. Human-in-the-Loop transfer: if the confidence score is below your threshold (configurable from 50% to 90%), the agent automatically transfers the conversation to an available human agent, with a context summary. The client sees "I'll connect you with an advisor."
2. Transparent response without invention: if no human agent is available (outside business hours), the agent explicitly states it doesn't have the information and can note the request for follow-up within 24 hours. The agent never "guesses."
3. Unanswered question collection: all questions the agent couldn't answer are automatically compiled into a weekly report. This lets you identify knowledge base gaps and address them progressively.

Anti-hallucination protection: responses are anchored in your document base. If a document doesn't contain the information, the agent cannot invent it.

Yes — and it's often the best approach for organizations with diverse needs. SkaLean supports deploying multiple distinct agents on your account, each with: its own identity (name, photo, bio), a configured personality (formal, warm, technical), a different knowledge base (or a subset of your main base), different channels, and different escalation rules.

Multi-agent usage examples:

1. A "client service" agent on WhatsApp for clinic patient questions, an "internal HR" agent on Teams for employee questions.
2. A "sales" agent that qualifies prospects and books appointments, a "support" agent handling after-sales issues.

Each agent is managed separately in the dashboard but shares the common knowledge base.

Both modes are supported.

Reactive mode (default): the agent responds to incoming client messages, 24/7.

Proactive mode (outreach): the agent can initiate conversations based on defined triggers:

1. Time trigger: appointment reminder 24h before, follow-up on an unanswered quote after 3 days, payment reminder at D+30.
2. Event trigger: a new lead enters the CRM, the agent sends a personalized welcome message. An order ships, automatic notification to the client.
3. Behavioral trigger (Enterprise plans): a client hasn't interacted in 90 days, the agent sends a reactivation message.

Important legal constraints: proactive mode on WhatsApp and SMS requires prior opt-in from the recipient (CASL regulation in Canada). SkaLean includes automatic compliance checks before any proactive sending.