# How I Replaced My RAG Pipelines With Claude Projects | No Code

Source: https://qvantx.com/insights/how-i-replaced-my-rag-pipelines-with-claude-projects-no-code/
Author: Ani Björkström — QvantX Sweden AB, Stockholm
Published: 2026-04-19
Updated: 2026-09-20
Video: https://www.youtube.com/watch?v=VSF94NSdCYM (7 min)
Topic: AI Tools

License: free to quote and cite with attribution to https://qvantx.com

---
_DATA ENGINEERING_

## Can Claude Projects Replace a Traditional RAG Pipeline?

#### Key takeaways

- A Claude Cowork project answered questions correctly across five file formats — PDF, DOCX, XLSX, CSV, and JPEG — with no embeddings, no vector database, and no code.

- Claude found a $612 million net income figure on page 5 of a PDF annual report and summed $106,005.65 across 48 CSV rows without being told which file to search.

- Projects connect directly to a folder on your computer, so they accept far larger document sets than a normal chat upload allows.

A traditional retrieval-augmented generation pipeline means chunking documents, generating embeddings, running a vector database, and maintaining all of it as files change. Ani Björkström, a tech consultant based in Stockholm, tested whether Claude Cowork projects can deliver the same result with none of that infrastructure, using a folder of mixed-format finance files.

### What is the difference between a Claude project and a normal chat?

Projects accept much larger document sets, because a normal chat has a smaller limit on document sizes. A project also connects to a folder path on your computer, either created from scratch or pointed at an existing catalog, so every file in that folder becomes queryable without individual uploads.

The test project was connected to an existing folder containing five files: a PDF annual report, a DOCX professional services agreement, an XLSX quarterly sales report, a CSV transaction log, and a JPEG invoice. Creating the project and linking the folder takes a few clicks, with no ingestion step to configure.

### How accurate is Claude across different file formats?

Claude answered all five test questions correctly, one per file format, without ever being told which document held the answer. In each case it identified the right file on its own and cited where the figure came from, down to the page, section, or sheet.

| File format | Question asked | Claude's answer (verified correct) |
|---|---|---|
| PDF annual report | Net income for 2025 | $612 million, cited from page 5 |
| DOCX services agreement | Termination terms | 60 days prior written notice, per section 2.2 |
| XLSX quarterly sales data | Monthly active users for 2025 | 892,000, from the key KPI sheet |
| CSV transaction log | Sum of the training category | $106,005.65 across 48 rows |
| JPEG invoice | Invoice due date | November 14, 2025 |

The CSV case stands out: Claude aggregated an amount column filtered by category and reported the 48-row count, which matched a manual check of the file. The JPEG invoice shows images work too, with the November 14, 2025 due date read correctly.

### Do you still need embeddings and a vector database?

For a document set like this, no — Claude retrieved correct information from every format without any RAG infrastructure to build or maintain. That removes the ongoing maintenance burden of a traditional pipeline: no re-chunking when documents change, no embedding model to manage, no vector store to operate.

For finance teams, this means a controller or analyst can point a project at a folder of reports, contracts, spreadsheets, and scanned invoices and start asking questions in minutes.

#### FAQ

**Do I have to tell Claude which document contains the answer?**

No. In all five tests Claude scanned the connected folder, picked the correct document, and cited the location — for example page 5 of the annual report or section 2.2 of the agreement.

**Can it handle images and spreadsheets, not just text?**

Yes. It read a due date of November 14, 2025 from a JPEG invoice, pulled 892,000 monthly active users from an XLSX KPI sheet, and summed $106,005.65 across 48 CSV rows.

**Why use a project instead of pasting files into a chat?**

Chat has a smaller limit on document sizes, while a project connects to a whole folder on your computer. That makes projects the practical choice for document sets of any real size.


## Full video transcript

Chapters: 0:00 Catalog overview: 5 file types, one Claude project · 0:06 PDF annual report Q&A (extracting net income for 2025) · 1:02 DOCX agreement lookup (finding the termination clause) · 1:55 XLSX KPI extraction (monthly active users from Excel) · 2:33 CSV transaction summation (sum by category, 48 rows) · 3:32 Invoice image due date (reading a JPEG scan) · 3:52 Wrap up: is traditional RAG still worth it?

[0:00] Hi, my name is Annie. I'm a tech consultant based in Stockholm. In this tutorial, I will help [music] you to understand how you can extract information from different documents or images without creating a traditional drug or being a developer and understanding how [music] embedding or vector databases work. Let's get started. So, for this tutorial, I will use Cloud Cohort and in here I will come to projects and I will create a new project. Difference between creating a project and using just a chat is that in the project, you can upload more files. However, if you use just a general chat, then you can't upload big of documents there because there is a smaller limit for document sizes. So, I will create a new project and from here I will use an existing folder.

[0:49] However, you can even start from scratch. When you click on start from scratch, then you will create a new folder in your computer or you will use an existing folder in your computer. I will go with use an existing folder because I already created a catalog with all the information that I want to use for asking questions. And in here, I will click on documents and under documents, I will click on cloud and then projects and then from here I will use Cloud Cohort practical tips. And in here, you can see that I have sample documents. I have quarterly sales data, then invoice which is in JPEG format, then I have transactions data which is in CSV format, another document that is in docx format and last but not least, I have another document that is annual report and it is in PDF format.

[1:43] So, I will just choose this catalog. We click on open and then I will click on create. So, I created just a new project and I connected this project with a catalog where I have all the files from where I want to retrieve information. So, here you can see that this is the project that I created and it is connected to this path. So, right now we can go and get this path and read first all the documents that are in this catalog and then after that, we can come back and try Cloud and see how good it is to retrieve information from those documents. If we now open our catalog, we can see that we have different files in different formats. The first one is in PDF format and it is in annual report.

[2:28] Let's try to open this document and ask questions about this document and see if Cloud can answer. So, this is the actual annual report and in here we have different numbers and let's say I want to know what is the net income for 2025. So, I will copy net income from here and I will go back to Cloud and I will ask. Pay attention that I am not specifying from which document it should find this information and I'm expecting Cloud to scan all the documents to get the correct document and the correct number. So, here it says from page five, net income for 2025 is 612 million, which is absolutely correct. And now let's go to the next files.

[3:16] So, the second file in this catalog is the one called professional services. Let's try to open this one. We can see that this is a professional services agreement. We have different information, but under 2.2, we have either party may terminate this agreement for convenience upon 60 days. Let's go back to Cloud and ask. And now Cloud will think about this, will try to identify the correct documents and here we have the answer. As stated in section 2.2 of the professional services agreement, either party may terminate the agreement for convenience upon 60 days prior written notice. This is correct. Cloud could even get information from the second file which was in docx format.

[4:04] Now, let's go ahead and ask questions about the other document that is in xlsx format. I will open this document. You can see that it has information about monthly revenue, regional breakdown and key KPIs. Now, if we want to understand how many monthly active users do we have for 2025, we will go back to Cloud and we'll ask that question. And here is the answer. According to the key KPI sheet in quarterly sales data report we had 892,000 monthly users, which was definitely correct. So, Cloud is doing great so far, but we have two more formats. The other document is in CSV format and it involves a lot of transactions. If we open this document, we can see that we have categories and amount USD.

[4:53] And right now, I want to know what is the sum of amount USD for category training. Let's go to Cloud and ask. It is actively thinking right now and here we can see that it says that it's 106,005.65 dollars. And we can see that it is getting the information from the correct CSV file and let's go and see if we actually had 48 rows as it states. If we open the same document, can see 48, which means that we actually had 48 rows in this document. This means that Cloud was correct once again. And the last file in this folder is an invoice image.

[5:41] Here we can see that it says October 15th is the invoice date and November 14th is the due date. Now, let's go and ask Cloud. I'm expecting Cloud to find the invoice, then to extract information for invoice due date and here it says it is November 14th, 2025, which is definitely correct. As we saw in this tutorial, Cloud could extract correct information from all types of documents without any difficulties. So, I want to understand what do you think? Will you still use the traditional drug and then do maintenance for drug or you feel like Cloud projects are a lot better alternative at this point. Let me know, please, in the comments and if you like this video, give me thumbs up and see you in the next one.

[6:32] Bye.