@adminide-stack/yantra-help-browser 12.0.16-alpha.6 → 12.0.16

package/lib/templates/content/platform-features/platform-overview.md

@@ -0,0 +1,58 @@
+# Yantra Platform Overview
+
+Yantra is an **AI-powered workforce platform** that deploys intelligent AI employees into your existing toolchain. Unlike traditional chatbots that wait for prompts, Yantra AI employees operate proactively — monitoring your communication channels, development pipelines, CRM systems, and project management tools around the clock.
+
+## Why Yantra Exists
+
+Modern teams lose significant productivity to repetitive operational tasks:
+
+- **Context fragmentation**: Critical information is scattered across dozens of tools
+- **Manual data entry**: Teams spend hours updating CRMs, logging notes, and syncing systems
+- **Knowledge silos**: When team members leave or switch projects, institutional knowledge disappears
+- **Alert overload**: Engineers, salespeople, and marketers drown in notifications without clear prioritization
+
+Yantra solves these problems by providing AI employees that integrate directly into your workflow and handle operational overhead autonomously.
+
+## Core Philosophy
+
+### Proactive, Not Reactive
+
+Traditional AI assistants answer when asked. Yantra AI employees **watch, learn, and act** without waiting for instructions. They surface issues before they escalate, update systems before data goes stale, and share context before you need to ask.
+
+### Gradual Trust Model
+
+Every Yantra AI employee starts in supervised mode. You review and approve all actions. Over time, as trust builds, you can grant more autonomy — letting the AI employee handle routine tasks independently while escalating only the important decisions.
+
+### Built on OpenClaw
+
+Yantra is built on top of [OpenClaw](https://github.com/openclaw), a battle-tested open-source AI agent framework. This means transparent, auditable AI behavior backed by a thriving community of contributors and enterprise users.
+
+## What Yantra AI Employees Can Do
+
+| Capability                      | Description                                                |
+| ------------------------------- | ---------------------------------------------------------- |
+| **Continuous Monitoring**       | Watch Slack, GitHub, error trackers, CRMs, and email 24/7  |
+| **Contextual Correlation**      | Connect data points across tools to surface insights       |
+| **Automated Actions**           | Update CRMs, create tickets, draft messages, and log notes |
+| **Intelligent Alerting**        | Filter noise and escalate only what matters                |
+| **Institutional Memory**        | Capture and recall every interaction, decision, and fix    |
+| **Multi-Channel Communication** | Interact via Slack, WhatsApp, Telegram, or email           |
+
+## Who Uses Yantra
+
+- **Engineering Teams**: Monitor errors, correlate bugs with deployments, maintain institutional knowledge
+- **Sales Teams**: Auto-update CRMs, prepare call briefs, ensure 100% follow-up on leads
+- **Marketing Teams**: Manage social media presence, draft content from product updates, track brand mentions
+- **Support Teams**: Route issues proactively, surface relevant context for customer problems
+- **Operations Teams**: Keep systems synchronized and teams informed
+
+## Getting Started
+
+Setting up Yantra takes approximately 5 minutes:
+
+1. **Connect your tools** — One-click OAuth for Slack, GitHub, HubSpot, Sentry, Linear, Gmail, and 40+ more
+2. **Configure preferences** — Define what your AI employee should watch for and how to prioritize
+3. **Start supervised** — Review and approve all AI actions to build trust
+4. **Expand autonomy** — Gradually let your AI employee work independently on routine tasks
+
+For detailed setup instructions, see the [Getting Started Guide](/help-center/ai-employees/getting-started-ai-employees).

package/lib/templates/content/platform-features/proactive-alerts.md

@@ -0,0 +1,88 @@
+# Proactive Alerts — Know Before It Escalates
+
+Yantra AI employees don't wait for you to discover problems — they detect issues early and notify you with full context, so you can respond before situations escalate.
+
+## How Proactive Alerts Work
+
+Your AI employee continuously analyzes events across all connected tools, looking for patterns and anomalies that indicate something needs attention. When it detects an issue, it:
+
+1. **Correlates the signal** with data from other tools to understand the full scope
+2. **Checks for existing responses** — whether someone is already handling it or a ticket exists
+3. **Assesses severity** based on your configured thresholds and historical patterns
+4. **Delivers a contextualized alert** with all relevant information and suggested actions
+
+## Alert Types
+
+### Error Spike Alerts
+
+When your error monitoring tool (Sentry, Datadog, etc.) shows an unusual increase in exceptions, your AI employee investigates:
+
+- Identifies the root error and affected components
+- Checks if the spike correlates with a recent deployment
+- Looks for existing tickets or ongoing fixes
+- Notifies the appropriate team with a complete summary
+
+### Customer Impact Alerts
+
+When multiple customers report similar issues or a high-value account encounters a problem:
+
+- Cross-references support channels with technical monitoring
+- Surfaces the customer's history and account importance
+- Suggests response templates based on similar past incidents
+
+### Stale Activity Alerts
+
+When important items stop progressing:
+
+- Deals that haven't been updated in your CRM
+- Pull requests awaiting review beyond your team's SLA
+- Support tickets approaching response time limits
+
+### Anomaly Alerts
+
+When metrics deviate from normal patterns:
+
+- Unusual traffic patterns or usage drops
+- Unexpected changes in deployment frequency
+- Communication volume spikes in specific channels
+
+## Smart Escalation
+
+Not every alert deserves your attention. Yantra uses **smart escalation** to ensure you only get interrupted when it genuinely matters:
+
+### Severity Classification
+
+- **Informational**: Logged for reference, included in daily briefings
+- **Low**: Batched and delivered at your preferred check-in times
+- **Medium**: Delivered promptly but not urgently
+- **High**: Immediate notification through your preferred channel
+- **Critical**: Multi-channel notification with escalation chain
+
+### Escalation Rules
+
+You define what constitutes each severity level:
+
+- Error count thresholds (e.g., 50+ occurrences in 1 hour = Critical)
+- Customer tier triggers (e.g., Enterprise account issues always escalate)
+- Time-based rules (e.g., P0 tickets unresponded after 15 minutes)
+- Custom keyword triggers (e.g., "data loss", "security breach")
+
+## Configuring Alerts
+
+Navigate to your AI employee's preferences to set:
+
+- **Notification channels**: Where alerts get delivered (Slack, email, WhatsApp, Telegram)
+- **Quiet hours**: When to batch non-critical alerts
+- **Escalation chains**: Who gets notified if the primary responder doesn't acknowledge
+- **Alert suppression**: Rules to prevent duplicate or noise alerts
+
+## Alert History
+
+All alerts are logged with:
+
+- Timestamp and severity
+- Full context at the time of alerting
+- Actions taken (by the AI employee or team members)
+- Resolution status and timeline
+
+This history helps refine alert thresholds over time and provides data for post-incident reviews.

package/lib/templates/content/platform-features/smart-escalation.md

@@ -0,0 +1,93 @@
+# Smart Escalation — Only Get Pinged When It Matters
+
+Yantra's smart escalation system ensures you're only interrupted for issues that truly require your attention, filtering out noise while never missing critical events.
+
+## The Noise Problem
+
+Modern development and business teams are overwhelmed by notifications:
+
+- Hundreds of Slack messages daily
+- Dozens of error alerts, most of which are duplicates or low priority
+- CRM notifications for routine stage changes
+- Email threads that don't require immediate action
+
+This constant noise creates **alert fatigue**, causing teams to miss genuinely important issues buried in the flood. Yantra solves this by acting as an intelligent filter between your tools and your attention.
+
+## How Smart Escalation Works
+
+### 1. Signal Analysis
+
+Every event from your connected tools is analyzed for:
+
+- **Severity**: How impactful is this event?
+- **Novelty**: Is this a new issue or a known one being tracked?
+- **Scope**: How many users, customers, or systems are affected?
+- **Trend**: Is this getting better or worse?
+
+### 2. Context Enrichment
+
+Before escalating, your AI employee gathers full context:
+
+- Related events from other tools
+- Historical data on similar issues
+- Current status of any existing responses
+- Who is available and responsible for this type of issue
+
+### 3. Decision Logic
+
+Based on your configured rules and the enriched context, Yantra decides:
+
+- **Absorb silently**: Log it, include in daily summary
+- **Batch for later**: Group with related items for periodic review
+- **Notify gently**: Send a non-urgent message at your next check-in
+- **Escalate immediately**: Push a notification through your preferred urgent channel
+- **Trigger escalation chain**: If no response within the defined window, escalate to the next person
+
+## Escalation Configuration
+
+### Threshold-Based Rules
+
+```
+Error Count Rules:
+  - Less than 5 occurrences → Log only
+  - 5-25 occurrences → Batch notification
+  - 25-100 occurrences → Immediate notification
+  - 100+ occurrences → Critical escalation chain
+```
+
+### Customer Tier Rules
+
+- **Enterprise accounts**: All issues escalated immediately
+- **Pro accounts**: P0/P1 issues escalated, others batched
+- **Free accounts**: Included in daily summaries
+
+### Time-Based Rules
+
+- **Business hours**: Standard escalation flow
+- **Off hours**: Only critical issues escalate; everything else queued for morning
+- **Weekends**: Emergency-only escalation with higher thresholds
+
+## Escalation Chains
+
+Define cascading notification paths for critical issues:
+
+1. **Primary responder**: Notified first via Slack and email
+2. **Backup responder**: Notified after 10 minutes if no acknowledgment
+3. **Team lead**: Notified after 20 minutes if still unacknowledged
+4. **On-call manager**: Final escalation for sustained non-response
+
+## Feedback Loop
+
+Yantra learns from your responses to refine escalation over time:
+
+- If you consistently dismiss certain alert types, it adjusts thresholds
+- If you escalate items that were batched, it increases their priority
+- Monthly escalation reports help you tune your rules
+
+## Getting Started
+
+1. Connect your tools via [Integrations](/help-center/integrations/integrations-overview)
+2. Set your default escalation preferences in the AI employee configuration
+3. Define customer tiers and priority thresholds
+4. Specify quiet hours and escalation chains
+5. Review and refine based on the weekly escalation summary

package/lib/templates/content/product-features/ai-models.md

@@ -0,0 +1,89 @@
+# AI Models
+
+Yantra integrates with multiple state-of-the-art AI models, automatically selecting the best one for each query to maximize quality, speed, and cost-efficiency.
+
+---
+
+## Available Models
+
+### Large language models
+
+| Model             | Provider           | Best for                                              | Speed     |
+| ----------------- | ------------------ | ----------------------------------------------------- | --------- |
+| GPT-4o            | OpenAI             | Complex reasoning, nuanced analysis, multi-step tasks | Medium    |
+| GPT-4o mini       | OpenAI             | Quick factual lookups, simple Q&A                     | Fast      |
+| Claude 3.5 Sonnet | Anthropic          | Long-form writing, careful instruction following      | Medium    |
+| Claude 3 Haiku    | Anthropic          | Rapid responses, lightweight tasks                    | Very fast |
+| Llama 3 70B       | Meta (self-hosted) | On-premise deployments, privacy-sensitive workflows   | Medium    |
+
+### Embedding models
+
+| Model                  | Provider | Use case                             |
+| ---------------------- | -------- | ------------------------------------ |
+| text-embedding-3-large | OpenAI   | Semantic search, document similarity |
+| Cohere Embed v3        | Cohere   | Multilingual content retrieval       |
+
+### Specialized models
+
+- **Code models** — Optimized for programming questions, code generation, and debugging. Built on fine-tuned versions of GPT-4o and Claude.
+- **Classification models** — Custom fine-tuned BERT variants for intent detection, content tagging, and routing.
+
+---
+
+## Automatic Model Selection
+
+By default, Yantra's **Intelligent Router** selects the best model for each query. You don't need to choose manually.
+
+### How routing works
+
+1. **Analyze the query** — The router classifies the intent (search, generate, summarize, analyze).
+2. **Assess complexity** — Simple factual queries route to fast models; complex reasoning tasks route to frontier models.
+3. **Check latency budget** — Real-time interactions prioritize speed; background tasks prioritize quality.
+4. **Apply fallbacks** — If the primary model is unavailable or rate-limited, the router transparently falls back to an alternate provider.
+
+> **Result:** You always get the best balance of quality and speed without managing model selection yourself.
+
+---
+
+## Manual Model Selection
+
+Pro and Enterprise users can override automatic selection:
+
+1. Click the **model selector** icon next to the search bar.
+2. Choose your preferred model from the dropdown.
+3. Your selection persists for the current session, or you can set a default under **Settings > Preferences > AI Model**.
+
+### When to override
+
+- **Consistency** — Use the same model for a series of related queries to get consistent output style.
+- **Experimentation** — Compare responses from different models side-by-side.
+- **Compliance** — Select on-premise models (Llama 3) when data cannot leave your infrastructure.
+
+---
+
+## Model Capabilities
+
+### What every model can do
+
+- Generate natural language responses in 50+ languages
+- Cite sources with inline references
+- Format output as prose, lists, tables, or code blocks
+- Maintain multi-turn conversation context
+- Respect custom system prompts and guardrails
+
+### Advanced capabilities (frontier models only)
+
+- Multi-step reasoning with chain-of-thought
+- Image and document analysis (GPT-4o)
+- 100K+ token context windows for long documents
+- Function calling and structured JSON output
+
+---
+
+## Model Updates
+
+Yantra regularly updates to the latest model versions:
+
+- **Automatic upgrades** — When providers release improved versions, Yantra transitions within 2 weeks after internal quality validation.
+- **Version pinning** — Enterprise customers can pin specific model versions for stability.
+- **Changelog** — All model changes are documented in the [Product Changelog](https://yantra.ai/changelog).

package/lib/templates/content/product-features/collaboration-tools.md

@@ -0,0 +1,88 @@
+# Collaboration Tools
+
+Work together with your team using shared searches, workspaces, and real-time collaboration features.
+
+---
+
+## Shared Workspaces
+
+Every team on Yantra gets a shared workspace where members can collaborate:
+
+### What you can share
+
+| Content type         | How to share                          | Permissions  |
+| -------------------- | ------------------------------------- | ------------ |
+| Search results       | Click **Share** on any result card    | View or Edit |
+| Conversations        | Share a thread link with team members | View only    |
+| Saved searches       | Add to a shared collection            | View or Edit |
+| AI employee insights | Post to a shared channel              | View only    |
+
+### Setting up a workspace
+
+1. Go to **Settings > Team > Create Workspace**.
+2. Name your workspace and add a description.
+3. Invite members via email or shareable link.
+4. Set default permissions (Viewer, Editor, or Admin).
+
+---
+
+## Sharing Searches
+
+### Share a single result
+
+1. Find the result you want to share.
+2. Click the **Share** icon (arrow) on the result card.
+3. Choose how to share:
+    - **Copy link** — Anyone with the link (and workspace access) can view.
+    - **Send to teammate** — Select a team member to notify directly.
+    - **Post to Slack** — Share directly to a connected Slack channel.
+
+### Share an entire search
+
+Click **Share Search** at the top of the results page to share the query and all results as a snapshot.
+
+---
+
+## Real-Time Collaboration
+
+### Live cursors
+
+When multiple team members view the same workspace, you'll see their cursors and selections in real time — no more "are you looking at this?" messages.
+
+### Comments and annotations
+
+- **Inline comments** — Click any result or AI response to add a comment. Team members are notified.
+- **Annotations** — Highlight text and add notes for context. Annotations are visible to all workspace members.
+- **@mentions** — Tag teammates in comments to draw their attention.
+
+### Activity feed
+
+The workspace activity feed shows:
+
+- Who shared or saved what
+- Comments and annotations
+- AI employee actions and recommendations
+- Integration updates
+
+---
+
+## Team Permissions
+
+| Role   | Can view | Can edit | Can share externally | Can manage members |
+| ------ | -------- | -------- | -------------------- | ------------------ |
+| Viewer | Yes      | No       | No                   | No                 |
+| Editor | Yes      | Yes      | No                   | No                 |
+| Admin  | Yes      | Yes      | Yes                  | Yes                |
+| Owner  | Yes      | Yes      | Yes                  | Yes                |
+
+Permissions can be set at the workspace level or per-item.
+
+---
+
+## Integration with Chat Tools
+
+Collaboration extends beyond the Yantra app:
+
+- **Slack** — Share results, receive team updates, and interact with your AI employee directly in Slack channels.
+- **Microsoft Teams** — Same capabilities as Slack for Teams-first organizations.
+- **Email** — Share results and summaries via email with one click.

package/lib/templates/content/product-features/conversation-features.md

@@ -0,0 +1,76 @@
+# Conversation Features
+
+Yantra doesn't just return search results — it lets you have a natural back-and-forth conversation to refine your understanding and go deeper on any topic.
+
+---
+
+## How Conversations Work
+
+Every search on Yantra can become a conversation. After receiving an initial response, you can:
+
+- **Ask follow-up questions** — Yantra remembers the full context of your conversation, so you don't need to repeat yourself.
+- **Refine your query** — If the first response isn't quite right, clarify what you meant or ask for a different angle.
+- **Go deeper** — Request more detail on a specific part of the response.
+- **Switch direction** — Ask about a related topic — Yantra understands the connection.
+
+---
+
+## Context Retention
+
+### What Yantra remembers
+
+Within a conversation thread, Yantra maintains:
+
+| Context type          | Example                                                         |
+| --------------------- | --------------------------------------------------------------- |
+| Previous questions    | Your full query history in the thread                           |
+| Entities mentioned    | People, products, companies, dates you've referenced            |
+| Preferences expressed | "Show me only academic sources" applies to subsequent responses |
+| Response references   | "Expand on point #3" refers to the current response             |
+
+### Context window
+
+- **Standard mode** — Retains the last 10 exchanges (~50,000 tokens of context).
+- **Pro Search** — Extended context with up to 100,000 tokens.
+- **Deep Research** — Full conversation history for the research session.
+
+> **Tip:** If a conversation gets very long, start a new thread for a fresh topic. You can always reference previous threads.
+
+---
+
+## Conversation Actions
+
+### During a conversation
+
+| Action              | How                                 | Description                                          |
+| ------------------- | ----------------------------------- | ---------------------------------------------------- |
+| Copy response       | Click the copy icon                 | Copies the formatted response to your clipboard      |
+| Share response      | Click the share icon                | Share a specific response with your team             |
+| Edit your message   | Click the edit icon on your message | Rephrase your question to get a different response   |
+| Branch conversation | Click "Branch"                      | Create a parallel thread exploring a different angle |
+| Rate response       | Thumbs up/down                      | Help improve response quality over time              |
+
+### After a conversation
+
+- **Save to collection** — Add the entire conversation to a saved collection for future reference.
+- **Export** — Download the conversation as Markdown, PDF, or plain text.
+- **Continue later** — All conversations are saved automatically. Pick up where you left off from the Threads page.
+
+---
+
+## Multi-Language Support
+
+Conversations work in 50+ languages:
+
+- **Auto-detect** — Yantra automatically responds in the language you use.
+- **Switch mid-conversation** — You can change languages at any point. Yantra adapts.
+- **Translate** — Ask Yantra to translate a response into a different language.
+
+---
+
+## Best Practices
+
+1. **Be specific** — Instead of "tell me more," ask "what are the cost implications of option B?"
+2. **Use context** — Reference earlier parts of the conversation: "Going back to the security concern you mentioned…"
+3. **One topic per thread** — Keep conversations focused for better AI context management.
+4. **Rate responses** — Thumbs up/down helps Yantra learn your preferences over time.

package/lib/templates/content/product-features/export-features.md

@@ -0,0 +1,87 @@
+# Export Features
+
+Take your research anywhere. Yantra lets you export search results, conversations, and reports in multiple formats for offline use, sharing, or integration with other tools.
+
+---
+
+## Supported Export Formats
+
+| Format         | Best for                                 | Includes                                        |
+| -------------- | ---------------------------------------- | ----------------------------------------------- |
+| **Markdown**   | Documentation, GitHub, Notion            | Formatted text with headers, lists, code blocks |
+| **PDF**        | Formal reports, presentations, archiving | Styled document with Yantra branding (optional) |
+| **Plain text** | Simple sharing, email                    | Clean text without formatting                   |
+| **JSON**       | API integrations, data processing        | Structured data with metadata                   |
+| **CSV**        | Spreadsheets, data analysis              | Tabular data (results, citations, metadata)     |
+| **HTML**       | Web publishing, email newsletters        | Rendered HTML with styling                      |
+
+---
+
+## What You Can Export
+
+### Search results
+
+Export individual results or entire result sets:
+
+1. Open a search result page.
+2. Click **Export** in the top-right toolbar.
+3. Select results to include (all or specific ones).
+4. Choose your format.
+5. Click **Download** or **Copy to clipboard**.
+
+### Conversations
+
+Export complete conversation threads:
+
+1. Open the conversation you want to export.
+2. Click the **⋯** menu > **Export Conversation**.
+3. Choose the range (entire thread or selected messages).
+4. Select format and click **Export**.
+
+### Collections
+
+Export entire saved collections at once:
+
+1. Open the collection from your **Library**.
+2. Click **Export Collection**.
+3. Choose format and options.
+
+---
+
+## Export Options
+
+Customize your exports:
+
+| Option               | Description                           | Available for       |
+| -------------------- | ------------------------------------- | ------------------- |
+| Include sources      | Add source URLs and citation metadata | All formats         |
+| Include timestamps   | Add dates for each interaction        | JSON, CSV, Markdown |
+| Include images       | Embed referenced images               | PDF, HTML, Markdown |
+| Custom header/footer | Add your own branding or notes        | PDF only            |
+| Strip formatting     | Export raw content without styling    | All formats         |
+
+---
+
+## Bulk Export
+
+For large-scale exports (data backup, compliance, migration):
+
+1. Go to **Settings > Privacy > Export Data**.
+2. Select data categories: searches, conversations, collections, preferences, account data.
+3. Click **Request Export**.
+4. You'll receive a download link via email within 24 hours.
+5. The archive includes JSON and CSV files organized by category.
+
+> **Enterprise customers** can set up automated scheduled exports via the API.
+
+---
+
+## API Export
+
+Use the export API for programmatic access:
+
+```
+GET /v1/export/searches?format=json&from=2024-01-01&to=2024-12-31
+```
+
+The API supports all formats, date ranges, and filtering options available in the UI.

package/lib/templates/content/product-features/follow-up-questions.md

@@ -0,0 +1,82 @@
+# Follow-Up Questions
+
+Yantra intelligently suggests follow-up questions to help you explore topics more deeply without having to think of what to ask next.
+
+---
+
+## How Follow-Up Questions Work
+
+After every response, Yantra analyzes the content and generates 2–4 suggested follow-up questions. These appear as clickable chips below the response.
+
+### What drives the suggestions
+
+| Factor                | How it's used                                                 |
+| --------------------- | ------------------------------------------------------------- |
+| Response content      | Identifies subtopics or details worth exploring               |
+| Common next questions | Based on what other users typically ask after similar queries |
+| Your history          | Considers your recent searches and interests                  |
+| Conversation context  | Adjusts suggestions based on the flow of the current thread   |
+
+---
+
+## Types of Follow-Up Questions
+
+### Deepening questions
+
+Go further into a topic already covered:
+
+- _"What are the specific performance benchmarks for this approach?"_
+- _"Can you explain how the encryption layer works in more detail?"_
+
+### Broadening questions
+
+Explore related but different angles:
+
+- _"How does this compare to the alternative approaches?"_
+- _"What are the security implications of this architecture?"_
+
+### Practical questions
+
+Move from theory to application:
+
+- _"How would I implement this in a Node.js application?"_
+- _"What's the estimated cost for a team of 50 users?"_
+
+### Clarifying questions
+
+Get more precise information:
+
+- _"Does this apply to the free tier or only paid plans?"_
+- _"Is this feature available on mobile as well?"_
+
+---
+
+## Using Follow-Up Questions
+
+### Click to ask
+
+Simply click any suggested follow-up question to instantly submit it. The response will appear in the same conversation thread, maintaining full context.
+
+### Edit before asking
+
+Hover over a suggestion and click the **edit** icon to modify the question before submitting. This is useful when the suggestion is close but you want to be more specific.
+
+### Ignore and ask your own
+
+Follow-up suggestions are optional. You can always type your own question in the input field. Suggested questions won't interfere.
+
+---
+
+## Customization
+
+### Toggle follow-up suggestions
+
+If you prefer a cleaner interface, disable automatic suggestions:
+
+**Settings > Preferences > AI Interaction > Auto follow-up questions** → Off
+
+When disabled, you can still manually request suggestions by typing "Suggest follow-up questions" in the chat.
+
+### Feedback loop
+
+Click the **×** on any suggestion to dismiss it. Over time, Yantra learns which types of follow-ups are useful to you and adjusts accordingly.