Customer Dashboard Portal

Customer-facing dashboard for models, playground, observability, usage, billing, API keys, and projects

1. Description

The Customer Dashboard portal is the end-user console for Bud customers to discover models, test prompts, monitor inference traffic, manage API credentials, and track usage and billing in one workspace. It provides a guided, drawer-based experience for viewing model details, project metadata, and API key security while surfacing observability and audit insights for operational trust.

This portal balances self-service and governance: teams can browse models and pricing, generate secure credentials scoped to projects, review usage and billing alerts, and trace inference requests down to raw gateway payloads. Deployments published from Bud AI Foundry appear in the models listing so teams can immediately use them without leaving the portal.

2. USPs (Unique Selling Propositions)

1. One console for discovery, execution, and governance

From model discovery to inference observability, users manage the entire lifecycle in a single dashboard without context switching.

2. Playground-first validation

Instant, in-portal testing means teams can validate prompts, latency, and responses before shipping to production.

3. Foundry-controlled pricing and quotas

Bud AI Foundry admins can set model token pricing (including zero-cost access) and define user-level custom token and cost quotas to control usage in the Customer Dashboard.

4. Actionable observability with deep request detail

Metrics charts, request filtering, and trace-level detail pages let teams troubleshoot latency, token usage, and failures with precision.

5. Secure-by-default API key workflows

API key creation, revocation, and encrypted display workflows are built into project context to minimize credential sprawl.

6. Usage transparency with proactive billing alerts

Real-time spend and token analytics plus threshold-based alerts provide immediate clarity on costs and guardrails.

7. Role-friendly navigation and shortcuts

A consistent sidebar, quick keyboard shortcuts, and collapse/expand views keep navigation efficient for daily operations.

3. Features

3.1 Global navigation & layout

• Collapsible left sidebar with shortcuts for Models, Playground, Logs, Usage & Billing, Audit, API Keys, and Projects.
• Theme toggle in the header area for light/dark switching.
• Task Island widget for notification and user profile with logout for quick operational control.

3.2 Models catalog

• Deployments published from Bud AI Foundry appear in the models catalog.
• Search by model name and browse a grid of cards with description, tags, and pricing hints.
• Model detail drawer with General and Use Model tabs.

3.2.1 Model detail: General tab

• Description with expand/collapse, basic metadata, provider type, and model size.
• Modalities shown for input/output (text, image, audio).
• Supported endpoints list with endpoint paths.
• Model URI with copy-to-clipboard action.
• Tags, strengths, and limitations sections.

3.2.2 Model detail: Use Model tab

• Code snippets for Curl, Python, and JavaScript with endpoint-aware payloads.
• One-click copy to clipboard for the selected code snippet.

3.3 Playground

• Embedded playground experience for testing prompts and model interactions.
• Open the new chat window workflow to place two or more models side-by-side.
• Send the same prompt to each model and review outputs together to judge quality or latency differences.

3.4 Logs (Observability)

• End-to-end metrics include request volumes, latency distributions, token usage, throughput (RPS), TTFT, and geographic breakdowns rendered as cards and charts.
• Dual tabs for Metrics and Requests to separate aggregated insights from raw inferences.

3.4.1 Metrics tab

• View-by segmentation (Model, Project) with time-range selection.
• Preset time ranges (Last 1h, 6h, 24h) and custom date range picker.
• Refresh button to fetch updated metrics.
• Multi-chart analytics for latency, tokens, success/failure rates, request volume, and geographic distribution.

3.4.2 Requests tab

• Rich filters for date range, endpoint type, model, token thresholds, and max latency.
• Quick time buttons (last 1h/6h/24h/7d/30d) and success-only toggle.
• Sortable inference table for timestamp and response time.
• Refresh and Export (CSV) actions for operational workflows.
• Click-through to inference request detail page for full request context.

3.4.3 Requests detail page

• Overview summary with inference ID, model, endpoint, project, timestamp, cached status, finish reason, and success/failure state.
• Copy-to-clipboard actions for inference IDs and request payloads.
• Request metadata (client IP, method, path, device, location, auth method, user agent) when available.
• Performance metrics (latency, TTFT, token counts) and error details for failed requests.
• Conversation view for chat requests with copy and download actions per message.
• Raw request and response panels with copy/download actions.

3.5 Usage & Billing

• Summary cards for total spend, total tokens, and total requests against quotas.
• Current plan visibility with billing period metadata.
• Project and time-range selectors (1d/7d/30d/90d).
• Export usage data to JSON.
• Billing alerts list with enable/disable toggles and delete actions.
• Drawer-based flow to create new billing alerts with type and threshold validation.
• Usage table with paginated daily records for tokens, requests, and cost.

3.6 API Keys

• Table with sortable columns: credential name, project, expiry, created, last used.
• Pagination controls with page-size selection.
• Create new key flow (credential name, project selection, expiry).
• Row actions: revoke key, open detail drawer.

3.6.1 API key detail drawer

• View key metadata (project, created, expiry, last used).
• Show/hide key with copy-to-clipboard action.
• Edit or delete key actions via menu.

3.7 Projects

• Search across projects and view cards with API key counts.
• Create new project drawer.
• Edit and delete actions via per-card menu.

3.7.1 Project detail drawer

• Project metadata (name, description, tags, project type).
• API key count and list of available keys with last-used timestamps.

3.8 Audit

• Summary cards for total events, failed actions, and resources modified.
• Search by resource name/user/ID, action filter, resource filter, and date range filter.
• Clear filters control and export-to-CSV action.
• Paginated audit table with action tags, resource details, IP address, and status.

4. How-to Guides

4.1 Create client users in Bud AI Foundry

1. In Bud AI Foundry, open User Management.
2. Select Create User and choose the client user role.
3. Save the user and share login credentials for Customer Dashboard access.
4. (Optional) Set custom token quota and cost quota for the user.

4.2 Access the Customer Dashboard

1. Log in to the Bud Customer portal with your credentials.
2. Use the side menu to navigate between Models, Playground, Logs, Usage & Billing, Audit, API Keys, and Projects.
3. Expand or collapse the sidebar using the chevron control to optimize workspace space.

4.3 Use a model in your app

1. From the model detail, select Use Model.
2. Choose Curl, Python, or JavaScript from the snippet dropdown.
3. Copy the snippet and replace the API key placeholder.
4. Send the request to validate connectivity and model responses.

4.4 Playground

1. Open Playground from the side menu.
2. Select a model or endpoint you want to test.
3. Enter a prompt and run it to review responses.
4. Copy the output or iterate on the prompt before moving to production.
5. Click the + icon in the toolbar to open a new chat window.
6. Load the desired model in the new pane; each window retains its own messages and settings.
7. Use the kebab menu to close or delete a chat.
8. Set presets, temperature, response length limit, repeat penalties, stop strings, optional notes and structured output options as needed in the settings panel.

4.5 Inspect logs and metrics

1. Open Logs from the side menu.
2. Use the Metrics tab to select View by (Model/Project) and a time range.
3. Click Refresh to update analytics.
4. Charts for request volume over time, latency distribution, geographic origin, token usage, requests per second, and TTFT with series colored per selected view dimension.
5. Switch to Requests to filter by date, endpoint type, model, tokens, or latency.
6. Inference request table with timestamp, project, deployment, prompt preview, response time, tokens, and success/failure status columns plus sortable fields for timestamp and response time.
7. Export CSV to share results or click a row to open Request Details.

4.6 Review an inference request detail

1. Click an inference request row to open the request detail page.
2. Copy the Inference ID or raw request/response as needed.
3. Review request metadata, performance metrics, and error details.
4. Download conversation or payload JSON for offline analysis.

4.7 Usage and Billing

1. Open Usage & Billing from the side menu.
2. Pick a project and time range to update charts and the usage table.
3. Click Export to download the usage JSON.
4. Select Add Alert in the Billing Alerts section.
5. Set the alert name, type (Cost Usage Alert or Token Usage Alert), and threshold.
6. Toggle alerts on/off or delete them from the list.

4.8 API keys

1. Open API Keys from the side menu.
2. Click Create New Key.
3. Enter a credential name, choose a project, and set expiry.
4. Open a row to view the key and copy it securely.
5. Use Delete to remove a key.

4.9 Manage projects

1. Open Projects from the side menu.
2. Click Project to create a new project record.
3. Enter a project name, tags and description.
4. Submit to add the project to the catalog.
5. Click on to edit or delete actions if needed.
6. Click on a project to open the project detail page.
7. Review tags and available API keys.

4.10 Monitoring & Audit

1. Open Audit from the side menu.
2. Review cards for total events, failed actions, and resources modified.
3. Filter by action, resource, or date range.
4. Search for users or resources to isolate events.
5. Export CSV when you need a compliance-ready activity log.

4.11 Task Island

1. Click Task Island in the left sidebar to open the notification panel.
2. Review notification cards with a title, message, timestamp, status, and icon.
3. Use the list to track success confirmations, and failure alerts.
4. Select Clear all to mark notifications as read and collapse the list.

5. FAQ

Q1. What can I do from the Customer Dashboard portal?

Browse models, test interactions in the playground, manage API keys and projects, monitor usage, configure billing alerts, and review observability and audit logs in one place.

Q2. Which filters are available in Logs?

You can filter by date range, endpoint type, model, token thresholds, max latency, and success-only status. Quick presets speed up time-based filtering.

Q3. How do I access a model’s API snippet?

Open the model detail page and use the Use Model tab to copy Curl, Python, or JavaScript examples tailored to the model’s supported endpoint.

Q4. How are API keys secured?

Keys are generated per project, shown once in detail page with show/hide controls, and can be revoked immediately when needed.

Q5. Can I export usage data and logs?

Yes. Usage exports download JSON data, and Logs/Observability exports provide CSV or JSON for detailed analysis.

Q6. Where do I see request-level troubleshooting details?

Open Requests tab in the Logs page, select an inference, and review the Observability Details view for metadata, errors, and raw payloads.

Q7. Can token pricing be set to zero for shared access?

Yes. Admins can set input and output token costs to 0 so teams can use a published endpoint without cost limits.