Search-to-Report Unified Workflow

New

A guided research pipeline that takes you from source discovery through investigation to report generation — with quality gates and RSI intelligence preserved at every step.

What is Search-to-Report?#

Search-to-Report is a unified research workflow that connects five systems — Search, Notebooks, Report Generation, Prompt Engineering, and Trend Analysis — into a single guided path. Instead of manually copying sources between tools and re-entering context at every stage, the workflow carries your research intelligence automatically from discovery all the way to your final output.

The workflow has two core phases: Collect Sources on the /search page, and Investigate Sources inside a notebook. From the notebook, you choose one of four output actions — Generate Report, Pull from Prompt Folder, Generate Prompt, or Generate Trend Config — and the system pre-fills the destination with everything it knows from your investigation.

Context carried automatically

Your curated sources, RSI scores, investigation findings, topic, audience, and domain travel with you through every phase. You never re-enter context that the system already captured.

Collect Sources

Curate a high-quality source set on the /search page. The quality gate ensures you proceed with at least 10 sources averaging RSI ≥ 50.

Investigate Sources

Dig into your sources in a notebook. Chat with sources, run Deep Research, and explore connections — all auto-saved.

Generate Report

One click pre-fills the report builder with your title, summary, sources, audience, and domain — no re-entry needed.

Generate Trend Config

Turn your investigation into a structured trend monitoring configuration, pre-loaded into the Trend Analysis setup page.

Generate Prompt

Save a research prompt generated from your investigation directly to a prompt folder, then jump to the Prompt Playground.

Archive Integration

Archived sources carry RSI scores computed at ingest time. Redesigned archive cards display quality signals alongside source metadata.

The Research Source Index (RSI)#

RSI is a composite quality score (0–100) computed for every source you add in the search phase. It measures six dimensions of source quality and governs whether your source set is strong enough to proceed with investigation.

Dimension	Weight	What It Measures
Authority	22%	Domain reputation and author credentials
Recency	18%	How fresh the source publication is
Methodology	18%	Research quality and data rigor
Corroboration	18%	Agreement with other sources in your set
Relevance	14%	Topical match to your research query
Access	10%	Source availability and accessibility

RSI scores are computed when you add a source, preserved through the bulk save API, displayed on notebook source cards, stored at archive ingest, and shown again on the Generate page when your sources are pre-loaded. The score follows your research from first discovery to final output.

RSI Explainer Modal

Any time you see an RSI badge, click the info icon next to it — or click "What is RSI?" in the guidance banner — to open the RSI Explainer Modal with a full breakdown of each dimension and tips for improving your scores.

Phase 1: Collect Sources#

The /search page is your starting point. It now operates as a structured Collect Sources phase with a quality gate that must be satisfied before you can proceed to investigation.

Search for sources

Enter your research topic in the search bar. Results appear with RSI scores shown on each card, so you can evaluate quality before adding.

Curate your source set

Click the bookmark icon on high-quality results to curate them. The guidance banner below the search bar shows your progress: "7/10 sources curated" and your current average RSI.

Reach the quality gate

Once you have 10 curated sources with an average RSI ≥ 50, the Investigate button pulses to signal it is ready. Sources with RSI below 50 drag the average down — swap them for higher-quality alternatives.

Click Investigate

The system creates a notebook from your curated source set, preserving RSI metadata for every source, and navigates you to the Investigate Sources phase.

Quality gate is a hard requirement

The Investigate button cannot be clicked until both conditions are met: 10 curated sources AND average RSI ≥ 50. This prevents proceeding with source sets that are too small or too low-quality to produce a reliable report.

Live RSI on every source card

See the composite RSI score on each source card in search results before adding it to your curated set.

Guidance banner with progress

The banner shows exactly how many sources you have, your current average RSI, and what you need to unlock the next phase.

Investigate button pulse

When the quality gate is met, the Investigate button pulses with a subtle animation to signal you are ready to proceed.

RSI Explainer Modal

Click any RSI info icon or 'What is RSI?' in the banner to open a full educational overlay on all six RSI dimensions.

Phase 2: Investigate Sources#

When you click Investigate, you are navigated to a notebook pre-loaded with your curated sources. The notebook header displays "Investigate Sources" as a subtitle, confirming you are in Phase 2.

Every source in the notebook's source panel displays its RSI badge from Phase 1 — green for high quality (≥ 70), amber for acceptable (50–69). You can chat with your sources, run Deep Research, verify claims, and build a complete picture of the topic before generating any output.

RSI badges on source cards

Each source card in the notebook shows its RSI score, so you always know which sources are your strongest while investigating.

Adaptive chat suggestions

Before turn 4, suggestions focus on research exploration. After turn 4, they shift to include generation actions — surfacing output options when they become relevant.

Deep Research auto-save

Deep Research results are automatically saved to the notebook. You will not lose your findings when you navigate to generate a report.

Four-button Generate panel

The studio panel shows four clearly labeled output actions: Generate Report, Pull from Prompt Folder, Generate Prompt, and Generate Trend Config.

Investigate before you generate

The more you investigate before generating, the richer the pre-fill context will be. The system captures your chat findings, source analysis, and topic exploration as part of the prefill record sent to the report builder.

Generate Report#

Clicking Generate Report in the notebook studio panel creates a prefill record from your notebook context and navigates you to /generate with the form already populated:

Field	Populated From
Report title	Notebook topic / investigation title
Executive summary	Key findings from your investigation
Sources	Your curated source set with RSI badges and favicons
Target audience	Audience set during notebook creation
Domain / industry	Domain captured at source collection

The seed sources appear in the source picker on the Generate page — each one displaying its RSI badge, domain favicon, and metadata. You can add more sources, remove sources you no longer want to include, or proceed directly to generation.

Pull from Prompt Folder#

The Pull from Prompt Folder button opens the Prompt Folder Picker — a modal that lists all your prompt folders and the prompts inside them. Select any prompt to use it as the generation template for your report.

After selection, you are navigated to /generate?promptId={id} where the selected prompt content is pre-loaded into the generation form alongside your notebook sources. This lets you apply a carefully crafted prompt structure to the sources from your current investigation.

Combine prompt templates with live research

Use Pull from Prompt Folder when you have a standard report format (executive brief, competitive teardown, investment memo) and want to apply that exact structure to the sources you curated in this session.

Generate Prompt#

Generate Prompt uses your notebook investigation to produce a reusable research prompt. The LLM generates a structured prompt based on the topic, findings, and source set from your investigation session.

Click Generate Prompt

The system analyzes your investigation context and generates a research prompt that could recreate a similar investigation.

Prompt saved to your folder

The generated prompt is automatically saved to your default prompt folder. A success toast shows the prompt name and folder location.

Jump to Prompt Playground

The toast includes a direct link to open the prompt in the Prompt Playground, where you can refine it, test it, or add variables before saving a final version.

Auto-saved to folder

No manual save needed. The generated prompt lands in your default prompt folder immediately.

One click to Playground

Jump directly from the save confirmation to the Prompt Playground to refine your prompt before reuse.

Generate Trend Config#

Generate Trend Config turns your investigation into a structured trend monitoring configuration. The system analyzes your sources and findings to extract the keywords, signals, and monitoring parameters most relevant to your research topic.

Config Field	Generated From
Primary keywords	Key terms from your investigation and source titles
Monitoring phrases	Patterns and signals identified in your sources
Data source recommendations	Source types most represented in your curated set
Alert thresholds	Suggested sensitivity based on topic volatility
Cadence	Recommended monitoring frequency (daily/weekly/monthly)

After generation, you are navigated to /trend-analysis/setup?configId={id} where all fields are pre-filled from the generated configuration. Review the suggestions, adjust anything that does not fit, and activate your trend scan in seconds.

Archive Integration#

Sources archived through the accumulation sidebar now receive RSI scores at ingest time. This makes the archive a quality-aware source library — you can see at a glance which archived sources are high-authority and which should be used with caution.

RSI at ingest

Every source archived after this feature was deployed carries a composite RSI score computed at the moment of archiving.

Redesigned archive cards

Archive source cards now show RSI badge, source type icon, domain, publication date, author, and word count — bringing them to visual parity with search and notebook cards.

Historical archive sources

Sources archived before this feature was deployed will show no RSI badge. To get an RSI score for an older source, re-archive it through the search flow.

Quality Gate Reference#

The quality gate on the Collect Sources page enforces two simultaneous conditions before you can proceed to investigation:

Condition	Threshold	How to Meet It
Curated source count	10 sources minimum	Bookmark sources from search results
Average RSI score	≥ 50 (out of 100)	Add higher-authority, more recent sources

Both conditions must be true simultaneously. Adding 20 sources does not help if their average RSI is 42. Adding high-RSI sources does not help if you have fewer than 10. The guidance banner shows which condition is blocking you and what to do about it.

Tips for improving RSI

To raise your average RSI: (1) prefer .edu, .gov, and established trade publication domains over blog posts; (2) use sources published within the last 12 months for the Recency dimension; (3) look for sources that cite each other or share common references for the Corroboration dimension; (4) use narrow, specific search queries to improve Relevance scores.

Frequently Asked Questions#

Can I start investigating with fewer than 10 sources?

No. The quality gate is a hard floor. The 10-source minimum ensures the notebook has enough material for meaningful investigation and report generation. For quick one-off research, use the notebook directly without going through the Search-to-Report workflow.

What happens to my sources after I click Investigate?

Your curated sources remain in your search session. They are also copied into the newly created notebook. You can continue adding sources to the search session and create a second notebook if needed — the original notebook is independent once created.

Can I use Generate Report more than once from the same notebook?

Yes. Each click of Generate Report creates a new prefill record and navigates you to a fresh Generate page. Previous prefill sessions are not affected.

What if my RSI scores are wrong?

RSI is computed automatically from source metadata. If a score seems off, it is usually because the source URL lacks metadata (no publication date, unknown domain, paywalled content). In these cases, the affected dimensions default to neutral scores rather than zero, so the impact is limited.

Does the Trend Config I generate create an active scan?

No. Clicking Generate Trend Config creates the configuration and takes you to the setup page with fields pre-filled, but the scan is not activated until you confirm it on the setup page. You have full control over what gets activated.

Where are generated prompts saved?

Generated prompts are saved to your default prompt folder. You can change which folder is your default in Prompt Settings. After saving, you can move the prompt to any other folder from the Prompt Library.

PreviousWorkflow Templates

NextWebhooks