Skip to main content

​
Overview

This guide shows how a Portfolio Manager with a Claude subscription can use Claude Cowork to automate a daily research workflow. A scheduled task runs early each morning to identify the top 5 topics that could impact the financial market, assess their relevance to every company in your portfolio, and deliver individual PDF briefs, all before the trading day begins.

​
Requirements

​
What you will build

By the end of this guide you will have a fully automated daily task that: ✅ Searches for the top 5 financial news topics from the last 24 hours using Bigdata.com
✅ Assesses how each topic could impact each company (impact score and confidence level)
✅ Generates a PDF report per company with detailed analysis and inline attribution
✅ Updates your portfolio spreadsheet with a summary of the results

​
Step 1: Portfolio spreadsheet

Create an Excel file listing the companies you want to monitor. You can download the template spreadsheet from the link below:

Download portfolio template

my_portfolio_YYYY_MM_DD-HH_MM.xlsx
You only need to fill in the Ticker and Company Name columns. Claude will automatically populate the remaining columns (Topics, Impact score, Confidence, and Daily brief status) with the results of the workflow. Portfolio template Upload this file to your Claude Cowork project folder so the scheduled task can read it.

​
Step 2: Create the scheduled task

Open Claude Cowork and create a new scheduled task by clicking + New task: New task button in Claude Cowork Configure the task with the following settings:
  • Name: Portfolio impact brief
  • Frequency: Daily at 09:00
  • Model: Claude Opus 4.6 or any latest model available in your subscription.
  • Folder to work in: Point to your Claude Cowork project directory where you stored the spreadsheet.
In the Prompt field, provide the following prompt: 
The task folder contains a spreadsheet called my_portfolio_YYYY_MM_DD-HH_MM that lists the companies in my portfolio, with each company identified by its ticker, name, and optionally its RavenPack Entity ID.

If the column "RavenPack Entity ID" is not populated. Please use the Bigdata tools to find those companies and obtain their Entity ID. Immediately after finding them, write the IDs back to the spreadsheet so we can use them in future task executions.

Each time this task runs, I want to create a folder named YYYY_MM_DD-HH_MM within the working folder to store the task output reports.

I want to create a PDF report for each company in the portfolio spreadsheet, analysing how the latest news topics could impact them. Use Bigdata tools and add inline attribution with hyperlinks to the original document.

- First, search for the top 5 topics of the last 24 hours that could impact the financial market.

- Then search for information and assess how each topic could impact the companies in my portfolio, along with the impact score and confidence.

- Create the PDF reports in the created folder.

- Create a copy of the spreadsheet - adding the timestamp to the name - into the created folder, and update the columns with the calculated values for "Impact score" and "Confidence", and mark the "Daily brief status" column as done.
You can run a test and customize the prompt to your needs. Scheduled task created

​
Step 3: Run the workflow

When the scheduled task triggers, Claude Cowork executes the following steps autonomously: Scheduled task running
1

Search for top 5 financial news topics

Uses the Bigdata.com connector to search for the most significant financial news topics from the last 24 hours.
2

Resolve all portfolio companies

Calls find_companies through the Bigdata Knowledge Graph to resolve each ticker in your spreadsheet to its canonical entity.
3

Assess impact of each topic on each company

Cross-references the discovered topics against every company in the portfolio, determining relevance, impact direction, and confidence.
4

Create PDF reports

Generates an individual PDF brief for each company with detailed analysis and source attribution.
5

Update spreadsheet with scores

Writes the topic mappings, impact scores, and confidence values back into a new copy of the portfolio spreadsheet.
6

Verify outputs and finalize

Confirms all files were created successfully in the timestamped output folder.

​
Step 4: Review the results

Once the task completes, navigate to your project folder. You will find a root-level structure with your original template and timestamped daily folders: Folder structure

​
Updated spreadsheet

Inside each daily folder, the updated spreadsheet contains the analysis summary. Each row now shows:
  • Topics: Which of the top 5 topics (T1-T5) are relevant, with their individual impact direction (positive or negative)
  • Impact score: A net score aggregating the impact of all relevant topics
  • Confidence: The model’s confidence level in the overall assessment
  • Daily brief status: Done once the PDF report has been generated
Portfolio analysed

​
Individual PDF reports

Each company receives its own detailed PDF brief named DailyBrief_{TICKER}_{DATE}.pdf, containing the full analysis with inline attribution linking back to the original sources indexed by Bigdata.com. Reports generated

​
Tips and customization

Modify the instructions to search for more or fewer topics (e.g. “top 10 topics”) depending on how broad you want the coverage to be.
Add constraints to the search prompt such as “focus on US technology sector” or “topics affecting European markets” to narrow the scope.
Set the task to run multiple times per day (e.g. morning and afternoon) to capture intraday developments, or only on weekdays to skip weekends.
Extend the template with columns like Sector, Region, or Portfolio Weight to give Claude additional context for more nuanced impact assessments.