WhatsApp Digital Workforce
Deploy specialized AI experts, configure intelligent intent-based chat routing, and define global fallback parameters for smooth human-AI collaboration.
Interactive operations logs simulator demonstrating natural intent matching, confidence evaluation, fallback triggers, and instant routing bypass.
1Multi-Agent Slot System
Instead of relying on a single generalist AI, SyncFlo allows you to deploy a digital workforce of specialized agents. Each agent acts in its own sandbox slot with custom behaviors, tones, and target offerings.
Workforce Slots & Plan Gating
Your total active agent slots are governed by your subscription plan. You can extend your workforce limits by purchasing extra slots as add-ons.
Includes 1 active agent slot. Best for simple FAQ bots or small single-purpose assistants.
Includes 3 active agent slots and unlocks Intelligent Intent Routing.
Supports unlimited/custom agent slots built for large organizations.
Core Slot Behaviors:
- Primary Expert Status: One slot is always designated as the Primary Expert. If routing fails or is turned off, this primary agent handles the conversation by default.
- Active Toggles: Quickly spin up or deactivate agent slots without losing their configuration history.
- Locked Slots: Slots exceeding your plan limit will be visually locked. You can buy add-ons in the dashboard to immediately unlock them.
2Intelligent Intent Routing
When multiple agent slots are active, the system routes incoming messages to the most qualified agent. This is controlled by the **Intelligent Intent Routing** toggle.
Intent Analysis Engine
The Overseer AI performs Advanced Intent Analysis on the customer's message, matching their request against the Assignment Rules of all active agents.
If the Overseer detects an intent change mid-chat (e.g. transitioning from asking product specs to complaining about a refund), it hot-swaps the active agent slot seamlessly.
Writing Routing Rules
Under each agent's config, write a clear, descriptive Assignment Rule outlining when this agent should be triggered.
3AI Rules & Handoff Configuration
Global AI Rules establish guardrails on the conversations, determining exactly when the AI should attempt to help and when it must hand the customer over to human support agents.
Operating Hours & Availability
Control when the AI can request automatic human takeover. The AI continues responding 24/7, but human transfer is only triggered during specified hours (e.g., 09:00 - 18:00) on active days. Customizable timezone (default: Asia/Kolkata) keeps routing aligned with your team.
Keyword Triggers
Specify comma-separated keyword patterns (e.g., help, support, urgent). When a customer sends these matching terms, the conversation triggers immediate routing escalation, routing the user to a live team member.
Auto-Transfer & Failed Attempts
Toggle auto-transfer to allow the AI to release the conversation to a human. Configure a failed-attempts threshold (between 1 and 5 attempts) after which human takeover executes automatically.
Low Confidence Behavior
Define what the system should do if the AI's confidence in understanding is low. Choose between Ask once, then transfer (clarify once, and transition on next failure) or Transfer immediately.
Review Solicitation
Toggle post-delivery reviews. If enabled, set a review delay from 1 to 30 days. SyncFlo automatically follows up with the customer via WhatsApp asking for review feedback after the set duration.
Sensitive Topics & Instant Bypass
Define comma-separated sensitive topics (e.g., refunds, payment disputes, legal questions, complaints). Any incoming messages matching these topics bypasses the AI completely and triggers immediate human routing.
Control who receives the chat when transferred to human agents:
- Available Team Member: Routes the conversation to active teammates using round-robin distribution.
- Account Owner: Transfers the chat directly to the primary account owner's inbox.
- Manual Review Queue: Places the conversation into a triage folder for managers to assign manually.
Customize the template message sent to customers right before the AI pauses. (e.g., "I'm connecting you with a team member who can help with this.").
4Detailed Agent Settings Editor
When configuring or creating an individual Agent, settings are organized across four tabs in the Dashboard configuration panel:
1. Identity (Persona)
- Agent Name: Internal tracking name (e.g., Sales Assistant).
- Conversational Tone: Dropdown selection between 🤵 Professional, 👋 Friendly, ❤️ Empathetic, or 💰 Sales-Oriented.
- System Instructions (Behavioral Rules): Large text area for detailed instructions defining personality, boundaries, and formatting (see Prompting Guides below).
2. Routing Rules (Mission)
- Internal Purpose Description: A short, clear summary explaining what this specialist does (e.g., "Handles pricing and technical spec inquiries").
- Assignment Rule (Trigger): Declarative text block evaluating when the Overseer router should hand over the chat (e.g., "I handle all queries about pricing, discounts, and enterprise quotes.").
3. Business Context
- Business Name: Custom name of the company or unit.
- Industry Mode: Selection including Ecommerce, Real Estate, Healthcare, Education, and General Services to apply specialized context parsing.
- Industry Niche: Highly specific focus area details.
- Target Audience & Key Offerings: Core product descriptions, buyer personas, and service boundaries injected directly into the LLM system context.
4. Parameters & Tuning
- Response Capacity (Max Tokens): Limits the output length of responses. High-end recommended default is 500 to 1000 tokens to ensure concise WhatsApp messages.
- Creativity (Temperature): Interactive slider from 0.0 (Strict / Precise) to 1.0 (Creative / Varied). Use 0.2 to 0.4 for strict support, and 0.7 to 0.9 for sales copy.
5Prompting & Instruction Guides
Learn how to write effective System Instructions and Assignment Rules. Better inputs directly result in higher routing accuracy and more professional customer responses.
Writing System Instructions
1. Define Persona & Mission
Instruct the model on its exact job role. Make sure it knows what it is authorized to do.
"You are the Support Specialist. Your job is to resolve technical bugs and guide users on feature setup."
2. Keep Formatting WhatsApp-Friendly
WhatsApp messages should be highly readable on mobile screens. Instruct the AI to avoid long paragraphs.
"Use short sentences. Format details with bullet points. Bold key terms using asterisks (*)."
3. Enforce Strict Out-of-Scope Rules
Prevent your agent from hallucinating or answering off-topic questions by defining clear boundaries.
"If asked about pricing plans or billing, state that you cannot access billing details and refer them to the Sales agent."
Writing Assignment Rules
1. Use First-Person Intent Declarations
The Overseer router evaluates these rules as semantic descriptions of what the agent is capable of handling.
"I handle questions about item availability, order shipping updates, and return logistics."
2. Delineate Clear Boundaries
Explicitly mention topics that this agent should NOT cover, helping the Overseer avoid false-positive routing.
"Do NOT route billing complaints, pricing inquiries, or refund requests to me."
3. List Specific Keywords and Milestones
Include concrete terms that are highly likely to appear in customer queries related to this agent's mission.
"Route queries containing words like: tracking number, parcel status, delivery duration, shipping address."
Step-by-Step: Go Live in Minutes
Open New Agent Slot and set up their Identity, Mission, Context, and Parameters settings.
Toggle Intelligent Intent Routing ON to allow the Overseer AI to handle routing between slots.
Click AI Rules to adjust the failed attempt thresholds, sensitive keywords blacklist, and timezone parameters.