--- title: "Search the web and run Deep Research" description: "Find current information with citations, or plan and run a longer investigation that produces a sourced report." canonical: "https://bearly.ai/docs/features/web-search-deep-research/" markdown: "https://bearly.ai/docs/features/web-search-deep-research.md" --- # Search the web and run Deep Research HTML: [https://bearly.ai/docs/features/web-search-deep-research/](https://bearly.ai/docs/features/web-search-deep-research/) > Find current information with citations, or plan and run a longer investigation that produces a sourced report. Bearly offers two ways to work with public web information: - **Web Search** is part of an ordinary chat reply. Bearly searches when current information or outside sources would improve the answer. - **Deep Research** is a separate, longer investigation. You review an editable research plan before starting a paid run, then receive a sourced report and PDF. Both work in Bearly on the web, desktop, iOS, and Android when they are available to your account. A managed-team policy can set or lock Web Search and can disable Deep Research. ## Get a current, sourced answer With Web Search set to **Auto**, ask for the result you want in ordinary language. It helps to name the date range, geography, comparison, or kind of source that matters. ```text copy Compare the three most recent quarterly results from these companies. Use their investor-relations releases and cite every figure. ``` Bearly decides when a search is useful, may open relevant pages for more context, and includes citations in its answer. Select a citation to open its source. A row of source icons may also appear with the response; point to an icon to see its title or select it to open the page. To control Web Search for the current chat: 1. Select the sliders control beside the chat input. 2. Under **Search**, select **Web**. 3. Choose **Auto** or **Off**. **Auto** makes web tools available; it does not force every message to search. **Off** removes web search and page retrieval from the current chat, so answers may rely on the conversation and the model's existing knowledge instead. To set the starting mode for future chats, open the sliders control, select **Chat Settings**, then choose **New Chats** → **Web Search** → **Auto** or **Off**. This default does not rewrite existing chats. Project chats may use their project's chat settings instead. > **Note — When a setting is managed** > If **Web Search** is missing from **This Chat**, or changing **Web** does not change the active mode, a project or team policy may be setting it for you. A team administrator can review the assigned policy under **Teams** → team name → **Policies**. ## Run Deep Research Use Deep Research for a question that deserves an explicit scope, a reviewable plan, and a substantial report—not for a quick fact check. 1. **Start with the Deep Research command** Type `/deep-research`, choose **Deep Research**, and enter the topic. On wider screens, you can instead select the **Deep Research** magnifying glass beside the chat input. ```text copy /deep-research Assess the US market for community solar software. Focus on 2024–2026 adoption, buyers, regulation, and credible market-size estimates. ``` 2. **Review the research brief** Bearly prepares a **Deep Research** card with an editable **Query**. Clarify the time period, geography, definitions, exclusions, and comparison criteria before selecting **Create Plan**. Creating the plan is free. An empty query cannot be submitted. 3. **Shape the plan** When **Plan** is ready, read it before spending anything. You can edit the plan directly. To ask for a newly drafted version, add guidance under **Notes**, then select **Revise Plan**. **Revise Plan** remains unavailable until you add notes. Direct edits in **Plan** are included in the eventual run even if you do not request another revision. 4. **Choose the depth** The card shows the selected tier and estimated cost: - **Standard** is estimated at $3 and requires the **Analyst** plan or higher. - Turn on **Ultra** for a deeper run estimated at $7. Ultra requires the **Research** plan. Planning and revision remain available before an upgrade. If the chosen tier is not included in your plan, **Run Research** opens the upgrade flow instead of starting the run. 5. **Start the investigation** Select **Run Research** only after the query, plan, tier, and estimate look right. The approved plan locks and the research begins. The research card shows its phase and elapsed time. You can open individual research-step chips while work is underway. A completed report hides those steps by default; select **Show research steps** to revisit them. 6. **Read and verify the report** The finished report appears in the chat with numbered citations and a collapsed source card. Select the source count to open **Sources**, then select any source to inspect it. Bearly also prepares a PDF. When delivery succeeds, the card shows **Sent to your inbox** and **View PDF**. If email delivery fails but the PDF was created, **View PDF** remains available in the chat. > **Caution — Review before you run** > The current Deep Research card does not offer a cancel button after a paid run begins. Treat **Run Research** as the final confirmation. The ordinary chat **Stop** control can stop a normal response, but do not rely on it to cancel a Deep Research run that has already started. ## Leave and return to a long run Deep Research is designed to recover an in-progress run after a page refresh, app restart, or handoff to another signed-in device. Reopen the same chat to see the current status. Recovery can take a moment while Bearly reconnects to the investigation. On the web, a long-running report may offer **Get the app**. The desktop app can keep the live status visible while you work elsewhere, but it is not required to receive the emailed report. On mobile, the operating system may pause the app in the background; reopen the same chat to resume its live status. ## If research does not go as expected | What you see | What to do | | --- | --- | | Web Search returns no useful sources | Rephrase the request with distinctive names, a date range, location, or preferred source type. Ask Bearly to search again. | | A page cannot be read | Ask Bearly to use another primary source. Some sites block automated retrieval or require a sign-in. | | **Deep Research** does not appear after typing `/deep-research` | Check whether your team policy permits Deep Research. If it should be available, reopen or update Bearly and try again. | | **Create Plan** is unavailable | Add a non-empty question in **Query**. | | Planning or revision reports an error | Keep the useful text, narrow the query or notes, and select **Create Plan** or **Revise Plan** again when the control is available. | | **Run Research** opens billing | The selected tier is not included in the current plan. Use the indicated upgrade or choose Standard if your plan includes it. | | The run shows **Research failed.** or **Research cancelled.** | The existing plan cannot be run again. Type `/deep-research` to begin a new investigation, reusing the useful scope from the earlier card. | | The report email does not arrive | Return to the chat and select **View PDF** if it is available. Also check the inbox for the email address on your Bearly account. | ## Continue through the documentation - [Documentation index](https://bearly.ai/docs.md): Browse every Bearly guide. - [Previous: Research your documents](https://bearly.ai/docs/features/document-research.md) - [Next: Summarize a document](https://bearly.ai/docs/features/deep-summary.md)