blog: Add Deep Research Piece

[danielcwq]

Describe Your Changes

Added new deep research piece for the blog!

Fixes Issues

• Closes blog: scope out search and deepresearch piece #5870

Other comments

• Might want to remove the "Pipeline Flow" Column for the table so that everything is rendered on screen!

---

Important

Adds a blog post on replicating Deep Research using Jan, detailing methodology, testing, and performance findings compared to commercial offerings.

• New Blog Post:
  • Adds deepresearch.mdx to docs/src/pages/post/.
  • Title: "Replicating Deep Research in Jan".
  • Describes methodology for replicating Deep Research using Jan.
• Content Details:
  • Explains Deep Research methodology and its core features: exhaustive search and report generation.
  • Compares Deep Research implementations across platforms like OpenAI, Grok, Claude, Gemini, Perplexity, and Kimi.
  • Details testing of Jan's capabilities using Model Context Protocol with Jan-Nano, GPT-4o, and o3 models.
  • Provides performance findings and analysis of Jan's output quality compared to commercial offerings.
• Technical Implementation:
  • Utilizes Model Context Protocol (MCP) for search capabilities.
  • Uses Serper API for google_search and scrape tools.
  • Includes assistant prompt for conducting research and writing reports.

^{This description was created by for 700af88. You can customize this summary. It will automatically update as commits are pushed.}

[github-actions]

Preview URL: https://366e1401.docs-9ba.pages.dev

[danielcwq]

@eckartal

[ellipsis-dev]

Caution

Changes requested ❌

Reviewed everything up to 714a9ba in 1 minute and 49 seconds. Click for details.

• Reviewed 129 lines of code in 1 files
• Skipped 8 files when reviewing.
• Skipped posting 6 draft comments. View those below.
• Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.

1. docs/src/pages/post/deepresearch.mdx:68

• Draft comment:
  Typo: Duplicate phrase 'gets as' found. Please remove the extra word for clarity.
• Reason this comment was not posted:
  Marked as duplicate.

2. docs/src/pages/post/deepresearch.mdx:83

• Draft comment:
  Typo: Duplicate 'that' in 'Ensure that that is loaded'. Please remove one occurrence.
• Reason this comment was not posted:
  Decided after close inspection that this draft comment was likely wrong and/or not actionable: usefulness confidence = 10% vs. threshold = 50% While this is a real grammatical issue, it's a very minor one. The sentence is still perfectly understandable. The rules say not to make comments that are obvious or unimportant. Grammar fixes in documentation, unless they significantly impact readability, probably fall into this category. The double "that" could potentially cause confusion for non-native English speakers. Documentation should be held to a high standard of clarity. While documentation quality is important, this particular issue is too minor to warrant a PR comment. The meaning is clear even with the duplicate word. This comment should be deleted as it addresses a grammatical issue that is too minor to warrant a PR comment.

3. docs/src/pages/post/deepresearch.mdx:122

• Draft comment:
  Best practice: Add a trailing newline at the end of the file.
• Reason this comment was not posted:
  Decided after close inspection that this draft comment was likely wrong and/or not actionable: usefulness confidence = 10% vs. threshold = 50% While having a trailing newline is generally considered good practice, this is an extremely minor issue. It's unlikely to cause any real problems and would be automatically fixed by most editors. The comment doesn't point out any significant code quality or functionality issues. The suggestion is technically correct according to common coding standards. Not having a trailing newline could potentially cause issues with some tools or text processing. However, this is such a minor issue that it doesn't warrant a PR comment. Most modern editors and tools handle this automatically, and it doesn't impact functionality or readability. Delete this comment as it's too minor of an issue to warrant a PR comment. Let automated tools handle trailing newlines.

4. docs/src/pages/post/deepresearch.mdx:11

• Draft comment:
  Inconsistency: Twitter title ('Replicating Deep Research with Jan') differs from page title ('Replicating Deep Research in Jan'). Consider unifying the phrasing.
• Reason this comment was not posted:
  Decided after close inspection that this draft comment was likely wrong and/or not actionable: usefulness confidence = 10% vs. threshold = 50% While technically correct that there's an inconsistency, this is an extremely minor difference in preposition that doesn't affect meaning or clarity. Both prepositions are grammatically correct and convey essentially the same meaning. This feels like nitpicking that doesn't add value. The inconsistency could potentially affect brand consistency and SEO if exact title matching is important. Some style guides do recommend exact consistency across all metadata. The difference is so minor that it's unlikely to impact SEO or user experience in any meaningful way. The core message and keywords remain identical. This comment should be deleted as it focuses on an insignificant difference in preposition choice that doesn't materially impact the content or user experience.

5. docs/src/pages/post/deepresearch.mdx:44

• Draft comment:
  Suggestion: Consider removing or adjusting the 'Pipeline Flow' column in the table for improved on-screen rendering, as noted in the PR description.
• Reason this comment was not posted:
  Decided after close inspection that this draft comment was likely wrong and/or not actionable: usefulness confidence = 10% vs. threshold = 50% 1. This is a new file being added, so the table is new content. 2. The Pipeline Flow column appears to be an intentional part of the comparison. 3. The comment references the PR description but we don't have access to that. 4. The table formatting looks reasonable in the markdown. 5. UI/styling suggestions are explicitly out of scope per the rules. I could be wrong about the table rendering - maybe there is a real issue visible in the PR description. The Pipeline Flow column could be causing horizontal scrolling issues. Even if there are rendering concerns, UI and styling feedback is explicitly out of scope per the review rules. The content itself appears intentional and valuable. This comment should be deleted as it relates to UI/styling concerns which are out of scope, and the table content itself appears intentionally designed.

6. docs/src/pages/post/deepresearch.mdx:41

• Draft comment:
  Typographical/grammatical note: "only OpenAI and Kimi does this currently" should use "do" instead of "does" for subject-verb agreement.
• Reason this comment was not posted:
  Comment did not seem useful. Confidence is useful = 0% <= threshold 50% This comment is purely informative and focuses on a typographical/grammatical issue, which does not align with the rules provided. It does not provide a code suggestion or address a potential issue in the code logic.

Workflow ID: wflow_VSnaHJI71IAdXCZl

^{You can customize by changing your verbosity settings, reacting with 👍 or 👎, replying to comments, or adding code review rules.}

[ramonpzg]

Good job so far. I would like to see a few of the comments I highlighted implemented before approving it :)

[ramonpzg]

Cut out the MCP details, this is not a tutorial but rather a blog post. Highlight the results first and then go to the details, and then to the conclusion of what you tried. Mention

• how long it took?
• what was the quality of the output?
• how does it compare to other deep research functionalities from other providers?

Don't leave the reader hungry

[danielcwq]

Got it -- removed the tutorial like version in place of this, would love your comments!

[ellipsis-dev]

Important

Looks good to me! 👍

Reviewed e38d653 in 1 minute and 56 seconds. Click for details.

• Reviewed 168 lines of code in 1 files
• Skipped 0 files when reviewing.
• Skipped posting 5 draft comments. View those below.
• Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.

1. docs/src/pages/post/deepresearch.mdx:41

• Draft comment:
  Good simplification by removing the 'Pipeline Flow' column. Confirm this aligns with design requirements for improved on-screen rendering.
• Reason this comment was not posted:
  Comment did not seem useful. Confidence is useful = 0% <= threshold 50% The comment is asking the PR author to confirm alignment with design requirements, which is against the rules. It does not provide a specific code suggestion or ask for a specific test to be written.

2. docs/src/pages/post/deepresearch.mdx:54

• Draft comment:
  Review the sentence on line 54 for subject–verb agreement ('23 pages that reads…').
• Reason this comment was not posted:
  Decided after close inspection that this draft comment was likely wrong and/or not actionable: usefulness confidence = 10% vs. threshold = 50% While this is technically correct grammar feedback, our rules say not to make purely informative comments or obvious corrections. Grammar/spelling fixes are not substantive code changes that require review comments. The author can catch this themselves or use a spell checker. The grammar error could impact readability and professionalism of the documentation. Documentation quality matters. While documentation quality matters, this is a minor grammar fix that doesn't warrant a PR comment. It's the kind of thing that should be caught by basic proofreading or tools. Delete this comment. Minor grammar corrections don't warrant PR comments and should be caught through normal proofreading.

3. docs/src/pages/post/deepresearch.mdx:75

• Draft comment:
  Detailed setup instructions were removed. Ensure the new high-level overview meets the intended scope of the blog post.
• Reason this comment was not posted:
  Comment did not seem useful. Confidence is useful = 0% <= threshold 50% This comment is asking the PR author to ensure that the new high-level overview meets the intended scope, which is against the rules. It doesn't provide a specific suggestion or point out a specific issue with the code.

4. docs/src/pages/post/deepresearch.mdx:102

• Draft comment:
  For consistency, use standard hyphen '-' for bullet points instead of an en dash if needed in the 'The Reality' section.
• Reason this comment was not posted:
  Comment looked like it was already resolved.

5. docs/src/pages/post/deepresearch.mdx:105

• Draft comment:
  Consider adding a trailing newline at the end of the file for better compatibility.
• Reason this comment was not posted:
  Comment was on unchanged code.

Workflow ID: wflow_i41Hk4NgMJxpY69K

^{You can customize by changing your verbosity settings, reacting with 👍 or 👎, replying to comments, or adding code review rules.}

[ellipsis-dev]

Caution

Changes requested ❌

Reviewed 7154d10 in 1 minute and 54 seconds. Click for details.

• Reviewed 206 lines of code in 1 files
• Skipped 0 files when reviewing.
• Skipped posting 3 draft comments. View those below.
• Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.

1. docs/src/pages/post/deepresearch.mdx:75

• Draft comment:
  Consider adding a language identifier (e.g., ````plaintext```) to the code block for improved syntax highlighting.
• Reason this comment was not posted:
  Confidence changes required: 33% <= threshold 50% None

2. docs/src/pages/post/deepresearch.mdx:131

• Draft comment:
  Typo: "limiations" should be "limitations".
• Reason this comment was not posted:
  Decided after close inspection that this draft comment was likely wrong and/or not actionable: usefulness confidence = 0% vs. threshold = 50% While this is a valid spelling error, spelling and grammar fixes are typically very minor issues that don't require a PR comment. These kinds of small fixes can be made directly by the PR author during their own review or caught by automated spell checkers. The comment doesn't add significant value. The spelling error does exist and fixing it would improve the document quality. Maybe it's worth pointing out to ensure it gets fixed? Simple spelling fixes don't need PR comments - they just add noise to the review. The author can catch this themselves or use a spell checker. Delete this comment as it's too minor of an issue to warrant a PR comment.

3. docs/src/pages/post/deepresearch.mdx:157

• Draft comment:
  Typographical error: "platfoms" should be "platforms".
• Reason this comment was not posted:
  Decided after close inspection that this draft comment was likely wrong and/or not actionable: usefulness confidence = 0% vs. threshold = 50% While this is technically correct, our rules state we should not make purely informative comments or obvious corrections. Spelling errors are typically caught by basic spell checkers and don't require a PR comment. This kind of minor issue doesn't warrant a review comment. The spelling error could potentially cause confusion for readers if left uncorrected. It's an objective, clear issue in the text. While true, this is still too minor an issue for a PR comment. The author can use spell check or their editor to catch these kinds of basic issues. Delete this comment. Spelling errors should be caught by basic tools and don't warrant PR comments.

Workflow ID: wflow_n7cu1fuqBP0xj1G5

^{You can customize by changing your verbosity settings, reacting with 👍 or 👎, replying to comments, or adding code review rules.}

[ellipsis-dev]

Important

Looks good to me! 👍

Reviewed b423d72 in 52 seconds. Click for details.

• Reviewed 13 lines of code in 1 files
• Skipped 0 files when reviewing.
• Skipped posting 1 draft comments. View those below.
• Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.

1. docs/src/pages/post/deepresearch.mdx:72

• Draft comment:
  Verify the updated link to prompt experiments (https://github.com/menloresearch/prompt-experiments) is intended as the new repository for generated reports.
• Reason this comment was not posted:
  Confidence changes required: 0% <= threshold 50% None

Workflow ID: wflow_rQ7OYAf73CPPwsOC

^{You can customize by changing your verbosity settings, reacting with 👍 or 👎, replying to comments, or adding code review rules.}

[ellipsis-dev]

Important

Looks good to me! 👍

Reviewed 700af88 in 1 minute and 33 seconds. Click for details.

• Reviewed 38 lines of code in 1 files
• Skipped 0 files when reviewing.
• Skipped posting 6 draft comments. View those below.
• Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.

1. docs/src/pages/post/deepresearch.mdx:80

• Draft comment:
  Good job converting provider names to markdown links for interactivity. One note: verify that the label “Google's generated report” correctly points to the Gemini 2.5 Flash Report PDF. If that is intentional, it’s fine.
• Reason this comment was not posted:
  Confidence changes required: 33% <= threshold 50% None

2. docs/src/pages/post/deepresearch.mdx:86

• Draft comment:
  Markdown link for OpenAI is consistent with the other changes. No issues found.
• Reason this comment was not posted:
  Confidence changes required: 0% <= threshold 50% None

3. docs/src/pages/post/deepresearch.mdx:89

• Draft comment:
  Perplexity now has a clickable link—this enhances usability. Looks good.
• Reason this comment was not posted:
  Confidence changes required: 0% <= threshold 50% None

4. docs/src/pages/post/deepresearch.mdx:93

• Draft comment:
  The link for Claude is implemented correctly. The comma after the link remains, which is acceptable.
• Reason this comment was not posted:
  Confidence changes required: 0% <= threshold 50% None

5. docs/src/pages/post/deepresearch.mdx:97

• Draft comment:
  Grok’s provider name is now a clickable link which is consistent with the rest of the document.
• Reason this comment was not posted:
  Confidence changes required: 0% <= threshold 50% None

6. docs/src/pages/post/deepresearch.mdx:100

• Draft comment:
  Kimi now uses a markdown link as well. This is clear and improves navigation.
• Reason this comment was not posted:
  Confidence changes required: 0% <= threshold 50% None

Workflow ID: wflow_9xvju4AxRToARvZK

^{You can customize by changing your verbosity settings, reacting with 👍 or 👎, replying to comments, or adding code review rules.}