Integrate Edit Image with 500+ apps and services

This n8n workflow demonstrates an approach to parsing bank statement PDFs with multimodal LLMs as an alternative to traditional OCR. This allows for much more accurate data extraction from the document especially when it comes to tables and complex layouts. Multimodal Parsing is better than traditiona OCR because: It reduces complexity and overhead by avoiding the need to preprocess the document into text format such as markdown before passing to the LLM. It handles non-standard PDF formats which may produce garbled output via traditional OCR text conversion. It's orders of magnitude cheaper than premium OCR models that still require post-processing cleanup and formatting. LLMs can format to any schema or language you desire! How it works You can use the example bank statement created specifically for this workflow here: https://drive.google.com/file/d/1wS9U7MQDthj57CvEcqG_Llkr-ek6RqGA/view?usp=sharing A PDF bank statement is imported via Google Drive. For this demo, I've created a mock bank statement which includes complex table layouts of 5 columns. Typically, OCR will be unable to align the columns correctly and mistake some deposits for withdrawals. Because multimodal LLMs do not accept PDFs directly, well have to convert the PDF to a series of images. We can achieve this by using a tool such as Stirling PDF. Stirling PDF is self-hostable which is handy for sensitive data such as bank statements. Stirling PDF will return our PDF as a series of JPGs (one for each page) in a zipped file. We can use n8n's decompress node to extract the images and ensure they are ordered by using the Sort node. Next, we'll resize each page using the Edit Image node to ensure the right balance between resolution limits and processing speed. Each resized page image is then passed into the Basic LLM node which will use our multimodal LLM of choice - Gemini 1.5 Pro. In the LLM node's options, we'll add a "user message" of type binary (data) which is how we add our image data as an input. Our prompt will instruct the multimodal LLM to transcribe each page to markdown. Note, you do not need to do this - you can just ask for data points to extract directly! Our goal for this template is to demonstrate the LLMs ability to accurately read the page. Finally, with our markdown version of all pages, we can pass this to another LLM node to extract required data such as deposit line items. Requirements Google Gemini API for Multimodal LLM. Google Drive access for document storage. Stirling PDF instance for PDF to Image conversion Customising the workflow At time of writing, Gemini 1.5 Pro is the most accurate in text document parsing with a relatively low cost. If you are not using Google Gemini however you can switch to other multimodal LLMs such as OpenAI GPT or Antrophic Claude. If you don't need the markdown, simply asking what to extract directly in the LLM's prompt is also acceptable and would save a few extra steps. Not parsing any bank statements any time soon? This template also works for Invoices, inventory lists, contracts, legal documents etc.

This n8n workflow demonstrates how we can use Multimodal LLMs to parse and extract from PDF documents in n8n. In this particular scenario, we're passing a candidate's CV/resume to an AI which filters out unqualified applications. However, this sneaky candidate has added in hidden prompt to bypass our bot! Whatever will we do? No fret, using AI Vision is one approach to solve this problem... read on! How it works Our candidate's CV/Resume is a PDF downloaded via Google Drive for this demonstration. The PDF is then converted into an image PNG using a tool called Stirling PDF. Since the hidden prompt has a white font color, it is is invisible in the converted image. The image is then forwarded to a Basic LLM node to process using our multimodal model - in this example, we'll use Google's Gemini 1.5 Pro. In the Basic LLM node, we'll need to set a User Message with the type of Binary. This allows us to directly send the image file in our request. The LLM is now immune to the hidden prompt and its response is has expected. The example CV/Resume with hidden prompt can be found here: https://drive.google.com/file/d/1MORAdeev6cMcTJBV2EYALAwll8gCDRav/view?usp=sharing Requirements Google Gemini API Key. Alternatively, GPT4 will also work for this use-case. Stirling PDF or another service which can convert PDFs into images. Note for data privacy, this example uses a public API and it is recommended that you self-host and use a private instance of Stirling PDF instead. Customising the workflow Swap out the manual trigger for another trigger such as a webhook to integrate into your existing services. This example demonstrates a validation use-case ie. "does the candidate look qualified?". You can try additionally extracting data points instead such as years of experiences, previous companies etc.

Are you a popular tech startup accelerator (named after a particular higher order function) overwhelmed with 1000s of pitch decks on a daily basis? Wish you could filter through them quickly using AI but the decks are unparseable through conventional means? Then you're in luck! This n8n template uses Multimodal LLMs to parse and extract valuable data from even the most overly designed pitch decks in quick fashion. Not only that, it'll also create the foundations of a RAG chatbot at the end so you or your colleagues can drill down into the details if needed. With this template, you'll scale your capacity to find interesting companies you'd otherwise miss! Requires n8n v1.62.1+ How It Works Airtable is used as the pitch deck database and PDF decks are downloaded from it. An AI Vision model is used to transcribe each page of the pitch deck into markdown. An Information Extractor is used to generate a report from the transcribed markdown and update required information back into pitch deck database. The transcribed markdown is also uploaded to a vector store to build an AI chatbot which can be used to ask questions on the pitch deck. Check out the sample Airtable here: https://airtable.com/appCkqc2jc3MoVqDO/shrS21vGqlnqzzNUc How To Use This template depends on the availability of the Airtable - make a duplicate of the airtable (link) and its columns before running the workflow. When a new pitchdeck is received, enter the company name into the Name column and upload the pdf into the File column. Leave all other columns blank. If you have the Airtable trigger active, the execution should start immediately once the file is uploaded. Otherwise, click the manual test trigger to start the workflow. When manually triggered, all "new" pitch decks will be handled by the workflow as separate executions. Requirements OpenAI for LLM Airtable For Database and Interface Qdrant for Vector Store Customising This Workflow Extend this starter template by adding more AI agents to validate claims made in the pitch deck eg. Linkedin Profiles, Page visits, Reviews etc.

This n8n workflow demonstrates how to automate image captioning tasks using Gemini 1.5 Pro - a multimodal LLM which can accept and analyse images. This is a really simple example of how easy it is to build and leverage powerful AI models in your repetitive tasks. How it works For this demo, we'll import a public image from a popular stock photography website, Pexel.com, into our workflow using the HTTP request node. With multimodal LLMs, there is little do preprocess other than ensuring the image dimensions fit within the LLMs accepted limits. Though not essential, we'll resize the image using the Edit image node to achieve fast processing. The image is used as an input to the basic LLM node by defining a "user message" entry with the binary (data) type. The LLM node has the Gemini 1.5 Pro language model attached and we'll prompt it to generate a caption title and text appropriate for the image it sees. Once generated, the generated caption text is positioning over the original image to complete the task. We can calculate the positioning relative to the amount of characters produced using the code node. An example of the combined image and caption can be found here: https://res.cloudinary.com/daglih2g8/image/upload/f_auto,q_auto/v1/n8n-workflows/l5xbb4ze4wyxwwefqmnc Requirements Google Gemini API Key. Access to Google Drive. Customising the workflow Not using Google Gemini? n8n's basic LLM node supports the standard syntax for image content for models that support it - try using GPT4o, Claude or LLava (via Ollama). Google Drive is only used for demonstration purposes. Feel free to swap this out for other triggers such as webhooks to fit your use case.

This n8n template demonstrates an approach to image embeddings for purpose of building a quick image contextual search. Use-cases could for a personal photo library, product recommendations or searching through video footage. How it works A photo is imported into the workflow via Google Drive. The photo is processed by the edit image node to extract colour information. This information forms part of our semantic metadata used to identify the image. The photo is also processed by a vision-capable model which analyses the image and returns a short description with semantic keywords. Both pieces of information about the image are combined with the metadata of the image to form a document describing the image. This document is then inserted into our vector store as a text embedding which is associated with our image. From here, the user can query the vector store as they would any document and the relevant image references and/or links should be returned. Requirements Google account to download image files from Google Drive. OpenAI account for the Vision-capable AI and Embedding models. Customise this workflow Text summarisation is just one of many techniques to generate image embeddings. If the results are unsatisfactory, there are dedicated image embedding models such as Google's vertex AI multimodal embeddings.

This n8n workflow demonstrates how to automate indexing of images to build a object-based image search. By utilising a Detr-Resnet-50 Object Classification model, we can identify objects within an image and store these associations in Elasticsearch along with a reference to the image. How it works An image is imported into the workflow via HTTP request node. The image is then sent to Cloudflare's Worker AI API where the service runs the image through the Detr-Resnet-50 object classification model. The API returns the object associations with their positions in the image, labels and confidence score of the classification. Confidence scores of less the 0.9 are discarded for brevity. The image's URL and its associations are then index in an ElasticSearch server ready for searching. Requirements A Cloudflare account with Workers AI enabled to access the object classification model. An ElasticSearch instance to store the image url and related associations. Extending this workflow Further enrich your indexed data with additional attributes or metrics relevant to your users. Use a vectorstore to provide similarity search over the images.

Instructions This automation overlays a background image with another image, making it easy to add watermarks or logos. You can use this automation to watermark your images by overlaying them with a transparent version of your logo. If you'd like to place your logo in a specific corner, feel free to adjust the position of the overlay image in the code node. How it Works Both images are downloaded, so we can process binary files (you can modify the source, tho.) We extract metadata, focusing on the dimensions of each image. The position of the overlay image is calculated (default: dead center of the background image). The two images are composited together. Limitations and Optimisation Opportunities The overlay image must be the same size or smaller than the background image for proper alignment. The overlay image does not automatically scale to match the proportions of the background image. Enjoy the workflow! ❤️ let the work flow — Workflow Automation & Development

This n8n workflow shows how using multimodal LLMs with AI vision can tackle tricky image validation tasks which are near impossible to achieve with code and often impractical to be done by humans at scale. You may need image validation when users submitted photos or images are required to meet certain criteria before being accepted. A wine review website may require users only submit photos of wine with labels, a bank may require account holders to submit scanned documents for verification etc. In this demonstration, our scenario will be to analyse a set of portraits to verify if they meet the criteria for valid passport photos according to the UK government website (https://www.gov.uk/photos-for-passports). How it works Our set of portaits are jpg files downloaded from our Google Drive using the Google Drive node. Each image is resized using the Edit Image node to ensure a balance between resolution and processing speed. Using the Basic LLM node, we'll define a "user message" option with the type of binary (data). This will allow us to pass our portrait to the LLM as an input. With our prompt containing the criteria pulled off the passport photo requirements webpage, the LLM is able to validate the photo does or doesn't meet its criteria. A structured output parser is used to structure the LLM's response to a JSON object which has the "is_valid" boolean property. This can be useful to further extend the workflow. Requirements Google Gemini API key Google Drive account Customising this workflow Not using Gemini? n8n's LLM node works with any compatible multimodal LLM so feel free to swap Gemini out for OpenAI's GPT4o or Antrophic's Claude Sonnet. Don't need to validate portraits? Try other use cases such as document classification, security footage analysis, people tagging in photos and more.