Academy
1. What Deep Research Is and When to Use It
Understand what Deep Research does differently from a single chat answer and identify scenarios where a multi-source investigation adds value.
Deep Research is an AI investigation engine that goes far beyond a single chat answer. Instead of relying on one model's internal knowledge, it queries multiple source providers in parallel — web, academic, news, and specialized databases — then cross-checks the findings and produces a single, sourced report. Every claim in the report is cited inline, so you can verify the material yourself. Think of it as dispatching a research team across the internet and getting back one consolidated briefing document.
A standard chat answer draws on whatever the AI model already knows. Deep Research adds three layers on top:
- Breadth — it fans out across many providers simultaneously, covering angles a single source would miss.
- Cross-checking — results from different providers are compared, reducing the risk of relying on one biased or outdated source.
- Inline citations — every key finding links back to its origin, making the report auditable and reusable in Documents, ChatPro, or workflows.
The same engine also powers the Deep Research toggle inside OUPI Studio, but there the research feeds into a larger agent-driven deliverable. In the standalone Deep Research view, you drive the investigation directly and receive the raw report.
When should you reach for Deep Research instead of a quick chat?
• Multi-angle questions — e.g., comparing regulatory frameworks across countries. You need web, academic, and news sources working together. • Current-event analysis — news providers surface the latest coverage while academic sources add depth. • Due diligence or fact-checking — cross-checked, cited findings are far more trustworthy than a single AI response. • Building a knowledge base — the sourced report can be copied straight into Documents or a workflow for future reference.
If your question has a simple, well-known answer, a regular chat is faster and cheaper. Deep Research shines when breadth, recency, or verifiability matter.
Precision drives quality. Instead of asking "Tell me about AI regulation," try "Compare the EU AI Act and the US Executive Order on AI regarding high-risk system requirements, as of 2024." Name the entities, time period, and angle you care about — vague questions dilute results across too many topics.
Fewer, well-chosen sources often beat "Select All." Match providers to the question: academic for scientific evidence, news for recent events, web for general coverage. Selecting every provider maximizes breadth but increases cost and processing time.
Cost and access are worth understanding upfront. Each research session's cost scales with the number of sources queried and the depth of the synthesis. If you need heavier usage or access to providers not included in your plan, you can add your own API keys in Settings — this shifts provider costs to your own accounts and can unlock additional sources. This flexibility lets teams control spend while still getting comprehensive investigations.
Open Deep Research and type a question relevant to your current work — something that genuinely needs multiple perspectives. Select two or three providers that match the topic (e.g., academic + news). Launch the session and watch it run live. When the report arrives, scan the inline citations to see how sources were combined.
Deep Research differs from a single chat answer by querying multiple providers in parallel, cross-checking findings, and delivering a cited report. Use it when your question demands breadth, recency, or verifiability — not for simple lookups. Choose sources deliberately, write precise questions with named entities and time frames, and remember that cost scales with source count and depth. The resulting report is portable: reuse it in Documents, ChatPro, or any workflow.