Understanding Grounding
What is “Grounding”?
Section titled “What is “Grounding”?”Grounding is the mechanism that ensures ADD CREATE generates content based strictly on your uploaded documents, not on the AI’s general training knowledge.
Without grounding, AI models can “hallucinate” — confidently producing incorrect facts, outdated figures, or invented statistics. With grounding, every piece of content the platform generates is tied directly to passages retrieved from your files. You can always trace a claim back to its source document.
How It Works
Section titled “How It Works”When you trigger a generation job, the platform doesn’t simply hand your documents to the AI and ask it to summarise. Instead, it:
- Identifies the key themes in your project by scanning a sample of your uploaded content.
- Searches for the most relevant passages across all your documents for each of those themes.
- Assembles an evidence pack — a curated set of the most pertinent excerpts, each tagged with its source filename and page reference.
- Generates content strictly from that evidence pack. The AI is instructed to use only what is in the evidence, not its general knowledge.
- Verifies every factual claim in the generated output against the evidence. Claims that cannot be traced back to a source passage are flagged or removed.
- Saves a grounding report alongside the artifact so you can audit the output.
Why This Matters
Section titled “Why This Matters”- Accuracy: Statistics, product names, pricing, and technical details come from your data — not the AI’s training set.
- Brand safety: Generated materials contain only facts from your verified documents.
- Auditability: Every statement in an artifact can be traced to a specific document and page.
Focusing on Specific Files
Section titled “Focusing on Specific Files”When launching a generation job, you can optionally choose to scope it to specific files within your project. This is useful when you have a large project but want an output focused on just one report or document, without the AI drawing from the rest of the knowledge base.
Advanced Usage Notes
Section titled “Advanced Usage Notes”Use these patterns to get more predictable results:
- Keep inputs specific and scoped to one intent per generation request.
- Save intermediate outputs often, especially before running major edits or regenerations.
- Use review and assignment workflows early to reduce rework late in the lifecycle.
Troubleshooting
Section titled “Troubleshooting”If results are not what you expected:
- Verify source quality first (missing context, inconsistent structure, or conflicting inputs reduce output quality).
- Re-check selected options and audience/profile settings before re-running.
- If a job fails repeatedly, wait briefly and retry once; then escalate with the artifact or job context.
Frequently Asked Questions
Section titled “Frequently Asked Questions”Why did quality change between runs?
Section titled “Why did quality change between runs?”Different input context, settings, and model/runtime conditions can produce different outputs.
Should I regenerate or manually edit?
Section titled “Should I regenerate or manually edit?”Regenerate when the core structure is wrong; manually edit when only wording, ordering, or style needs adjustment.
What is the best way to reduce iteration cycles?
Section titled “What is the best way to reduce iteration cycles?”Start with clear scope, run smaller batches, and use review checkpoints after each major step.