Workforce Bots

Each bot has a shareable public portal page, an embeddable iframe snippet, and a floating chat-widget script for external sites. Bots can also run internal-only for authenticated employees, in OneMango, Ask AI and mobile. Access mode is configurable per bot. Slack and Microsoft Teams aren't available for Workforce Bots yet.

Answers are grounded in your uploaded documents via retrieval-augmented generation and include source citations (top 3 source chunks). When the knowledge base has no match, the bot can fall back to web search (if enabled) or its configured fallback message, and low-confidence answers are logged to the Low-confidence answers queue for an admin to follow up. The asker still receives the answer — the queue is a review trail, not a gate, and nothing holds a question back. On the portal, a low-confidence answer also offers the asker a "Route to a human" button, which files a Service Desk ticket (or, without Service Desk, notifies the bot's admin).

Workforce Bots lets you stand up multiple named, purpose-built bots — HR Policy, Process/SOP, Compliance, Recruiting, or Custom — each with its own knowledge base, welcome message, and system prompt, and each deployable to outside audiences (candidates, contractors, customers) via a public portal, iframe, or widget. A single business-wide "Ask" box then routes every question to the best-matching bot automatically, so users never have to pick the right portal.

Yes. Each bot can be pinned to a specific language or set to auto-detect the asker's language and reply in kind, across 16 languages including English, Spanish, French, German, Portuguese, Italian, Japanese, Chinese, Korean, Hindi, Arabic, Dutch, Polish, Russian, Turkish, and Vietnamese.

White-label, yes: a bot's public portal can be custom-branded with your theme and MangoApps branding removed. Your own sign-in, not yet: for portal deployments you can set a per-bot SSO shared secret, and the portal checks a signed JWT (HS256, with an expiry claim); the asserted name and email are not yet attached to the conversation, which still shows as an anonymous visitor.

Every message keeps an audit trail of the retrieved sources and confidence score. You can enable optional regex-based PII redaction (SSN, credit card, email, phone, date of birth are masked before messages are stored), purge conversations for GDPR/incident response, and fire outbound webhooks on events like low-rating, unanswered, or conversation-complete. Low-confidence answers are listed in the Low-confidence answers queue for review after they are delivered, and the asker can use "Route to a human" to reach a person.

Grading now lives in the apps that own the content: the Internal Comms Effectiveness Grader is in Communications Hub, and the Compliance Requirements Checker is in Policy Hub. The Handbook, Job Description, SOP, and Intranet graders are a development preview and are not offered in the app's menus.

Yes, on bots where the bulk Q&A capability is enabled (e.g. Compliance and Custom bots). Upload a question list, the bot answers each from its knowledge base, and you can export the results to XLSX, CSV, or Markdown — useful for compliance audit questionnaires and RFP-style Q&A.

There is no hard question cap by default. Admins can set an optional monthly question quota per bot or across the business for metering; once a configured quota is reached the bot stops answering everywhere — the portal, the Ask page, the mobile app, Ask AI and the chat panel — until the next reset. Leaving the quota blank means questions aren't capped.