L o a d i n g

ContentIQ Documentation

Enterprise-grade, AI-powered content intelligence platform built in collaboration with IBM watsonX Orchestrate (wxO)

Table of Contents

What is ContentIQ?

ContentIQ is an enterprise‑grade, AI‑powered content intelligence platform built in collaboration with IBM watsonX Orchestrate (wxO). It ingests knowledge from websites, Microsoft OneDrive, and local documents, then powers watsonX Orchestrate agents with grounded, source‑cited answers. ContentIQ lets you create new wxO agents or extend existing ones, orchestrate multiple agents, deploy chat widgets, and monitor usage and ingestion with unified analytics.

1) Prerequisites

  • An IBM watsonX Orchestrate account and wxO API key.
  • Content sources you can access: public/internal websites, OneDrive, or local docs (PDF, DOCX, PPTX, XLSX, TXT, etc.).
  • Appropriate enterprise permissions to connect sources and deploy agents (RBAC supported).
  • Modern browser (Chrome, Edge, Safari) and network access to target sites.

2) Five‑Minute Quick Start

  1. Sign in to ContentIQ and open Settings.
  2. Paste your wxO API Key and API URL, then Save Settings.
  3. Go to Connections → Website Connect (or Document/OneDrive Connect), add your source.
  4. Choose Agent Type: use an existing wxO agent or create a new one.
  5. Ingest and watch progress in Ingest Analytics.
  6. Test in Chat Playground.
  7. Deploy via Embed (copy snippet) or Deploy Agent (form‑based push).
  8. Track impact in Usage Analytics.

3) Authentication & Setup

Setup Steps:

  1. Navigate to Settings in the left sidebar.
  2. Enter:
    • User Name (optional display name)
    • Orchestrate API Key
    • Orchestrate API URL (e.g., https://api.orchestrate.com)
  3. Click Save Settings.
Tip: API keys are stored securely; rotate them per your enterprise policy.

4) Connect Content & Ingest Knowledge

ContentIQ supports Website Ingestion, Document Uploads, and Microsoft OneDrive. Each path can attach knowledge to a new or existing wxO agent.

A. Website Ingestion

Step 1:

Go to Connections → Website Connect.

Enter the Website URL you want to ingest and click Continue.

Step 2: Choose Agent Type

  • Use Existing Agent (recommended if you already have a wxO agent)
  • Create New Agent (provisions a new agent in wxO via ContentIQ)

Step 3:

If using an existing agent, Select an Existing Agent from the list and Finish & Connect.

ContentIQ starts ingestion using an enterprise‑grade scraper with parallel fetching and retry logic.

Monitor live progress and post‑run health in Ingest Analytics.

What's handled automatically

  • Parallel crawling with backoff & retries
  • De‑duplication and chunking for high‑quality embeddings
  • Automated versioning and scheduled re‑ingestion to keep knowledge fresh

Best practices

  • Prefer a sitemap or a hub page with links to ensure coverage.
  • Ensure robots.txt / auth policies allow your crawl.
  • Use stable URLs for evergreen docs and release notes.

B. Document Uploads

Process:

  1. Go to Connections → Document Upload.
  2. Drag‑and‑drop or select files (PDF, DOCX, PPTX, XLSX, TXT, etc.).
  3. Choose the target agent (existing or new) and start ingestion.
  4. Use batch uploads for large libraries; ContentIQ processes in parallel.
Notes:
  • Versioning keeps the newest copy active; historical traces support analytics.
  • Non‑text PDFs should include a text layer (OCR if needed) for best results.

C. Microsoft OneDrive

Process:

  1. Go to Connections → OneDrive Connect.
  2. Authenticate via Microsoft OAuth.
  3. Select files or folders to ingest; start batch processing.
  4. Monitor progress in Ingest Analytics.

Additional connectors (SharePoint, Box, Google Drive) follow a similar flow.

5) Create or Extend Agents (Orchestration)

ContentIQ works natively with wxO's agent orchestration layer.

  • Create new agents: Provision directly in ContentIQ; they appear in wxO.
  • Extend existing agents: Attach ingested knowledge to any wxO agent.
  • Multi‑agent orchestration: Route queries or tasks across multiple domain agents.
  • Grounded answers: Responses are sourced only from verified content (minimizing hallucinations) with source citations.
  • RBAC: Apply role‑based permissions for who can connect sources, deploy, or view analytics.
Where to work with agents:
  • During connection flows (Website/Document/OneDrive) when you choose the agent.
  • In Chat Playground for active agent selection and testing.

6) Chat Playground & Conversations

A. Chat Playground

Usage:

  1. Open Chat Playground in the sidebar.
  2. Pick the active agent from the dropdown.
  3. Ask questions; observe streaming responses and contextual memory.

B. Conversations

Features:

  • Navigate to Conversations to see session history across agents.
  • View a conversation, or Download it for audits and training feedback.
  • Threaded history offers continuity across sessions (ContentIQ + wxO).
Quality tips:
  • Use clear prompts; leverage agent‑specific language (e.g., "Based on policy docs, what's…").
  • If an answer is incomplete, add missing docs and re‑ingest.

7) Deployment & Embedding

Option 1: Embed the Agent (Widget)

Steps:

  1. Go to Deploy Agent → Embed Agent.
  2. Select agent to embed.
  3. Copy the code snippet shown for that agent.
  4. Paste the snippet before the closing </body> tag (or where your widget container lives).
  5. Refresh your site; the agent widget appears.

Branding & UX

  • Customize colors, fonts, and placement to match your design system.
  • Configure default opening message and suggested prompts.

Option 2: Deploy the Agent (Form Push)

Steps:

  1. Go to Deploy Agent → Deploy Agent.
  2. Select the agent instance.
  3. Enter the target URL (destination website or portal).
  4. Click Deploy.

Integration hooks

  • REST APIs allow advanced embedding and application integrations.
  • Handoffs from ContentIQ agents to wxO workflows are supported.

8) Analytics & Monitoring

ContentIQ provides unified analytics across ingestion and usage for both ContentIQ and wxO activity.

A. Ingest Analytics (Pipeline Health)

Key Cards & Metrics

  • Total Documents Ingested and Success Rate
  • Ingest Time (with trend vs. prior runs)
  • Embedding Coverage (% of chunks successfully embedded)
  • Duplicate Content (% filtered duplicates)
  • Average Chunk Length & Chunk Overlap Ratio
  • Failures (by type)

Failure Types

  • Content Fetch Failures: 404, 403, timeouts, SSL, rate limits
  • Preprocessing/Chunking Failures: e.g., missing structure, block length, encoding anomalies
  • Parsing Errors: corrupted files, unsupported file types, JS‑heavy pages
  • Embedding/Storage Errors: LLM embedding timeout, chunk size limit exceeded, embedding rate limit hit, vector store insert failure, vector schema mismatch

Each card includes a trend indicator (↑/↓) to highlight regressions or improvements.

Operator actions

  • Fix 404/403 by updating URLs or allowlisting the crawler.
  • Address parsing with alternate formats (e.g., export HTML → PDF with text).
  • Reduce chunk size or increase limits for embedding timeouts.
  • Re‑run ingestion after fixes; schedule re‑ingestion for freshness.

B. Usage Analytics (Adoption & Answer Quality)

Engagement

  • Total Questions (with trend)
  • Active Users (weekly/monthly)
  • Average Session Length
  • User Feedback Rate and Helpful Answer Rate

Answer Quality

  • Answer Success Rate and Fallback Rate
  • Hallucination Rate and Citation Accuracy
  • Query Completion Rate and Prompt Latency

Insights

  • Most Queried Content Types (FAQ, blogs, docs, etc.)
  • Topic Clustering Insights and Unsuccessful Topic Clustering
  • Languages Used (English, French, Spanish, Other)
  • Top Referred Sources (Support, Docs, FAQ, etc.)

Optimization playbook

  • Low Helpful Answer Rate → add missing docs; refine chunking; provide more examples.
  • High Fallback Rate → broaden coverage; tune retrieval; add synonyms.
  • Elevated Hallucination Rate → verify sources; tighten grounding; enforce citations.
  • High Latency → reduce context size; cache FAQs; scale infra; review concurrency.

9) Governance, Security & Compliance

  • RBAC: Limit who can connect sources, change agents, deploy widgets, or view analytics.
  • Encryption: TLS 1.3 in transit; secure key handling; JWT/OAuth 2.0 with wxO.
  • Enterprise Readiness: SOC 2, GDPR, HIPAA‑ready deployments available.
  • Auditability: Downloadable conversations and ingestion logs for reviews.
  • Data Freshness: Scheduled re‑ingestion and versioning.

10) Troubleshooting

Ingest Analytics shows "Failed to fetch"

  • Check network/firewall connectivity and API credentials.
  • Refresh the page; verify the analytics service is reachable.

High 403/404 or timeouts

  • Confirm the crawl domain and paths; allowlist the crawler; use stable URLs; retry.

Parsing or JS‑heavy page errors

  • Prefer static render paths or server‑side rendering; export to PDF/HTML with text.

Embedding timeouts or size limits

  • Reduce chunk size; stagger large batches; confirm embedding service quotas.

High hallucination rate

  • Add authoritative sources; enable citations; review agent grounding configuration.

Slow prompt latency

  • Reduce retrieved context; pre‑cache hot content; scale concurrency; check region proximity.

OneDrive auth fails

  • Re‑authenticate via OAuth; confirm tenant consent and folder permissions.

11) Best Practices

  • Start with one high‑value agent (e.g., Support or Policy) before expanding.
  • Curate a canonical source set (docs, KB, policies); avoid noisy pages.
  • Use clear titles and headings to help chunking and retrieval.
  • Schedule regular re‑ingestion after product releases or policy changes.
  • Review Usage Analytics weekly; feed gaps back into ingestion.

12) FAQ

Q: Can I use my existing wxO agents?

A: Yes. Select Use Existing Agent during connection and choose from the list.

Q: How do I deploy to an internal portal?

A: Use Embed Agent to copy the widget snippet into your portal's HTML, or integrate via REST APIs.

Q: Can I export conversations for audit?

A: Yes. Go to Conversations and use Download.

13) Support & Resources

  • API Documentation & Integration Guides: (available from symplistic.ai)
  • Knowledge Base & Troubleshooting: (available from symplistic.ai)
  • Email Support: info@symplistic.ai
↑ Top