"How Do I…" Answered From One Prompt
Platform Help Agent is the in-product help center as a prompt. It searches platform help articles AND the tenant's own knowledge-base entries in one call, finds the canonical setting location via SettingsRegistryService, and resolves "open my schedule" against the same navigation registry the global lookahead uses. Read-only by design — role-aware on every return.
Why "How Do I X?" Becomes A Ticket Instead Of An Answer
The information already exists — in the help center, in the tenant's KB, in the settings registry, in the nav config. Platform Help Agent is the one tool that knows where to look so the user doesn't have to.
The Answer Is In The KB, But Nobody Found It
Acme Co has 80 KB articles answering 80 common questions. New hires don't know they exist. Tenured employees can't remember the search terms that work. So the message goes to IT anyway, and the article that already answered the question stays unread.
"Where Is The SSO Setting?" Sends People Through Six Menus
Admin Console has hundreds of settings across a dozen sections. "Where do I configure SSO?" requires knowing it's under Security → Identity → SSO Providers. Without the registry, everyone clicks through every section the first time.
Help Articles And Nav Shortcuts Live In Different Places
"How do I file an expense?" is a help article. "Open the expense app" is a nav shortcut. Two different surfaces with two different search behaviors. The user shouldn't have to know which they want — they should just ask.
Admin-Only Paths Get Suggested To Employees Who Can't Follow Them
A non-admin asks "how do I add a new user?" and gets back "Console → Users → New User" — a path they don't have access to. The suggestion is worse than no answer because it sends the employee to a 403 instead of to the right admin.
Help Center Search Returns Generic Articles When The Tenant Has A Local Answer
The tenant's own KB has an article called "How we file expenses at Acme" that overrides the generic platform article. Standard search returns the platform one first because it ranks higher globally. The employee gets a process that doesn't match Acme's policy and submits the expense wrong — generating exactly the ticket the local KB was meant to prevent.
Repeat Questions Don't Surface As Candidates For A KB Entry
The same six questions account for 40% of #ask-it volume every week. Each answer gets retyped by whoever's on shift. Nobody runs a "what are the top unanswered questions employees keep asking?" rollup, so the KB never grows where it would help most — and the support load stays constant week after week.
Platform Help Agent At A Glance
Platform Help AI
Read-only platform documentation, navigation, and settings lookup. Role-aware results.
Inside Platform Help Agent — The Actual Capabilities
Every block below maps to a real tool the agent uses against the platform help center, the tenant's own knowledge base, the canonical settings registry, and the navigation shortcut config. Strictly read-only.
One Search, Two Sources · Platform Help AND The Tenant's KB
search_help_articles searches across platform help articles AND the tenant's own knowledge-base entries (Service Desk KB, HR KB, etc.) in one call. Each result carries a source field — business_kb results inline the full content in summary, so the agent answers directly without a second fetch.
- Search both surfaces — search_help_articles covers platform docs and the tenant's business_kb in a single call.
- Tenant KB content is inlined — business_kb results return full content in summary; no follow-up fetch needed.
- Category filter — optional category param narrows to the relevant section.
- Limit default 5 — agent gets a tight ranked list, not a wall of every article.
Pull A Full Article When The Summary Isn't Enough
When a platform help summary teases the answer, get_article_details fetches the full content. For "what should I read next?", get_related_articles walks the related-article graph maintained by the help center.
- Full article body — get_article_details by article_id; used only when the summary isn't sufficient.
- Related-article walk — get_related_articles surfaces follow-up reading for any platform article.
- Feature guides + tutorials — list_feature_guides filterable by app_slug for app-specific walkthroughs.
- Topic browsing — browse_help_topics by section/topic for users who want to explore, not search.
"Where Is The SSO Setting?" Answered From The Canonical Registry
find_setting_location is the agent's answer to "where is X" questions. Backed by SettingsRegistryService — the canonical registry behind /admin/settings_hub — so the path the agent returns matches what the Admin Console actually exposes. Critically, the response is role-filtered: non-admin users do NOT receive admin-only paths.
- Find a setting by name — find_setting_location for "where is notifications?", "where is SSO?", "how do I configure leave management?".
- Canonical source — backed by SettingsRegistryService, the same registry behind /admin/settings_hub; no drift between agent and console.
- Role-filtered results — non-admins do not receive admin-only paths; they get the employee-accessible alternative or a "ask your admin" pointer.
- Distinct from search — "where" / "how do I get to" questions route here, not to search_help_articles.
"Open My Schedule" Resolved To The Right App, Right URL
find_navigation_shortcut answers "open X" and "take me to Y" by resolving the destination against config/navigation_shortcuts.yml — the same source the global Cmd+K lookahead uses. One config, two surfaces, zero drift.
- Find a navigation shortcut — find_navigation_shortcut by query (e.g. "my tasks", "schedule", "directory").
- Same source as global lookahead — both agent and Cmd+K read config/navigation_shortcuts.yml; one config governs both.
- Returns app + URL + display name — agent can deep-link to the destination, not just describe it.
- Role-aware — destinations the asking user can't access don't appear in the result set.
Outcomes Teams Can Measure
Platform Help Agent's job is to turn "how do I" tickets into self-serve answers, and to make the canonical help and navigation surfaces the first stop, not the last. Measure against the pre-agent baseline so you can see where help volume actually moved.
- "How do I" ticket deflection — share of help-style tickets that no longer reach Service Desk because Platform Help Agent answered first.
- business_kb hit rate — share of answers that come from the tenant's own KB vs platform-generic help (signals which KB articles are actually pulling weight).
- Setting-lookup speed — median seconds from "where is X" to a path on screen.
- Nav shortcut adoption — share of "open X" prompts resolved via the agent vs typed into the URL bar.
- Role-filter precision — share of returned paths the asker actually has access to (canary for registry drift).
Intentionally Read-Only · Same Registry Behind Console UI
Platform Help Agent's RISKY_TOOLS list is empty by design. It retrieves and explains; it never changes a setting, opens an article on someone's behalf, or invokes a shortcut. Every read is role-filtered against the same registries the Console UI reads from, so the agent reflects what the user can already do — never broadcasts paths they can't follow.
- RISKY_TOOLS is empty — 7 tools, zero writes; the agent never changes a record or a setting.
- Canonical registries — SettingsRegistryService for settings, navigation_shortcuts.yml for shortcuts; one source per registry, no drift.
- Role-aware reads — admin-only settings and admin-only nav destinations never surface to non-admins.
- Two KB sources, one tool — platform help + tenant business_kb stitched into a single result list with explicit source on every row.
WHAT TEAMS TRY INSTEAD
The four alternatives — and why none of them combine product help, settings lookup, and a tenant-specific knowledge base
Every workforce platform ships help. The honest gap is that most help surfaces either know the product or know the tenant — never both — and none of them point an admin to the exact settings screen and the exact KB row in one prompt.
Pasting questions into ChatGPT, Claude, or Copilot
General-purpose AI guessing at MangoApps from public docs
- Resolves the canonical settings location through <code>SettingsRegistryService</code> — not a guess that links to last year's URL
- Searches the tenant's own <code>business_kb</code> in the same call as platform help — citations from both
- Resolves open-my-schedule against the same navigation registry the global lookahead uses
Glean / Guru AI / Confluence Copilot
Vendor-trapped help AI on a separate knowledge surface
- Lives inside the platform employees actually use all day — no new tab, no separate sign-in
- Reads tenant KB and platform help in one tool call — not stitched by an enterprise-search vendor
- Honors the same role gates the rest of the platform honors — admin-only settings stay admin-only
Custom RAG over the platform's docs
An engineering team's quarterly project that drifts from the live registry
- Settings locations stay current automatically — agent reads the live registry, not a scraped snapshot
- Tenant KB indexed alongside platform help without a separate ingestion pipeline
- New help topics ship with the platform — no re-embedding job to keep the index fresh
The manual fallback — IT and HR triage how-do-I questions
Two teams answering the same questions every day
- How-do-I questions get a citation-backed answer instantly — IT and HR stop relaying tickets between each other
- Where-is-this-setting answered with the exact admin path, not a tour guide narrative
- Tenant KB and platform help unified for the asker — no who-owns-which-doc routing problem
PLATFORM LEVERAGE
Platform Help inherits the same registries the product uses
A standalone help bot has to keep its index current. Platform Help reads from canonical registries directly — there is no index.
SettingsRegistryService as source
Settings lookups resolve against the live registry the product itself reads. New admin screens are discoverable the day they ship — no re-index.
navigation_shortcuts.yml as source
Open-my-schedule, take-me-to-recognition, show-me-OKRs — all resolved through the same shortcut registry the global lookahead uses.
Two KB sources, one tool
Platform help + tenant business_kb returned in a single list with an explicit source on every row. No model bias toward one source.
Role-aware reads
Admin-only settings and admin-only navigation destinations never surface to non-admins — the agent honors what the platform honors.
Read-only by design
RISKY_TOOLS is empty — 7 tools, zero writes. The agent never changes a record or a setting. Help means help, not action.
Mobile-first delivery
Lives inside Ask AI in the same mobile app frontline employees opened for shifts and pay — no separate help portal to learn.
INDUSTRY FIT
Industries where in-product help removes the most friction
Platform Help is universal — but it shines where the workforce is distributed, multi-lingual, or shift-bound and cannot wait for a help-desk callback.
Retail
Store associates find their schedule, paystub, or training assignment in one prompt instead of bouncing between four menus.
Healthcare
Clinicians find the right policy, the right SOP, and the right consent form from one ask — no separate intranet search.
Manufacturing
Plant-floor workers on shared kiosks get product help in their language without learning a help-portal IA they only touch quarterly.
Hospitality
Property staff find the recognition-program rules, the comp policy, and the leave request flow in one ask between guest interactions.
Financial Services
New hires ramp in days — every how-do-I returns the exact path, the policy citation, and the linked help article from one prompt.
Public Sector
Runs entirely inside FedRAMP-eligible deployment options. Tenant KB stays on-tenant; no third-party help vendor in the data path.
WHY MANGOAPPS WINS
An embedded help agent beats a horizontal AI, a separate enterprise-search vendor, or a custom RAG build on every axis
The argument IT, HR, learning, and frontline ops all share — and the one Glean or Guru structurally cannot answer.
Cheaper than the alternatives
No Glean license, no Guru AI add-on, no engineering team rebuilding a RAG index of the same docs every quarter.
More secure
Role-aware reads, registry-grounded answers, zero writes. AiApiLog captures every call. Nothing leaves the tenant boundary.
Easier to deploy
Already deployed if Ask AI is on. Settings registry and navigation registry already exist. The agent reads them — no ingestion pipeline.
Easier to use
Lives inside Ask AI in the same UI employees use for everything else — no separate help portal to learn or sign into.
Easier to manage
KB stays current because the agent reads the canonical registry, not a scraped index. No re-embedding, no drift, no quarterly refresh project.
Easier to extend
Add a setting to the registry and the help agent finds it. Add a navigation shortcut and the agent can open it. Zero agent code changes required.
AI is actually better
A horizontal AI can describe MangoApps. Only Platform Help can return the exact admin path, the exact tenant KB row, and a citation-backed answer — all read from the live registries the product itself uses.
At our head office within HR, marketing, and finance, we use ChatGPT quite a bit, and MangoApps AI Helper is a useful tool for us. As I go around to the stores, I’ve been showing them how the AI Intranet Helper can find things and answer plain language questions about our resources, and they’re quite excited about that.
Alison van Zijl
Head of Human Resources
Full House
Customer Success
Related Customer Stories
Customer Case Studies
Customer Case Studies
Customer Case Studies
Video Case Studies
Customer Case Studies
Customer Case Studies
Frequently Asked Questions About Platform Help Agent
7 tools — search_help_articles (platform help + tenant business_kb in one call), get_article_details (full article body), list_feature_guides (filterable by app_slug), find_setting_location (canonical settings lookup, role-filtered), find_navigation_shortcut (resolve "open X" to a URL), browse_help_topics (section / topic browsing), and get_related_articles (next-read walks). All read-only.
Both, in one call. search_help_articles returns mixed results across platform help articles AND the tenant's own KB entries (Service Desk KB, HR KB, etc.). Each result carries a source field — business_kb entries inline the full content in summary, so the agent answers tenant-specific questions without a second fetch.
Yes. find_setting_location is role-filtered — non-admin users do not receive admin-only paths. The agent will say "this lives in the admin console — your IT admin can help" rather than handing out a path that 403s. Same for navigation shortcuts that point to admin-only destinations.
Two canonical registries. SettingsRegistryService is the same source the /admin/settings_hub page renders from — when a setting moves, the registry updates and the agent's answer updates with it. config/navigation_shortcuts.yml is the same source the global Cmd+K lookahead reads, so the agent and the keyboard shortcut always resolve "directory" to the same place.
"How do I" ticket deflection (volume that no longer reaches Service Desk), business_kb hit rate (share of answers grounded in the tenant's KB), setting- lookup speed, navigation-shortcut adoption, and role-filter precision (returned paths the asker can actually access). All measured against the pre-agent baseline.
Let's Talk
Since 2008, we've been building the workforce platform — earning the trust of 2 million+ users and an NPS of 78.
Why Choose Us?
- AI-Powered Platform: The most unified workforce experience on the planet.
- Top Security: HITRUST, ISO & SOC 2 certified.
- Exceptional UX: Delightful on mobile and desktop.
- Proven Results: 98% customer retention rate.
Trusted by Legendary Companies:
