ApertureDB
The Vector + Graph Database for Multimodal AI
Start new thread
Vishakha Gupta

ApertureDB Multimodal AI Workflows - Automate common AI tasks for multimodal data

by
How do you easily generate embeddings, detect objects, infer new attributes, or query your multimodal data? Stop wrestling with your datasets - use ApertureDB Multimodal AI workflows instead! Ingest or enrich complex datasets, run Jupyter notebooks, and more.

Add a comment

Replies

Best
Vishakha Gupta
Maker
📌
Hello, Product Hunt Community 👋, I’m Vishakha Gupta, CEO and Co-founder of ApertureData. I am excited to do my very first Product Hunt Launch, for ApertureDB Multimodal AI Workflows. If we want AI to think like humans, it needs all the signals a human brain processes to form our stories, which means multimodal data. Having worked with AI and data for years, I understand the immense challenge users face when trying to leverage multimodal data. Real world data is messy and hard to access. Why? It’s scattered in silos, there are different databases or data tools built based on the modality of data, and how on earth do you talk to all those data types through one language? Teams spend countless hours wrangling images, text, and other data types, trying to build AI applications that truly deliver value. This leads to delayed deployments, increased costs, and missed opportunities. Sounds familiar? 🙄 That’s exactly why we built ApertureDB — a purpose-built database that unifies multimodal data management, vector search, and knowledge graphs into a single backend, delivering up to a 10x productivity boost and accelerating AI deployments by 6-9 months on average. Building on this foundation, we’re excited to launch ApertureDB Multimodal AI Workflows today — the easiest way to manage multimodal datasets with just a few clicks on a scalable cloud platform. 🚀 What are ApertureData Multimodal AI Workflows? Workflows provide a way to automate the execution of a commonly-performed task, when managing multimodal datasets for AI. You can use preconfigured sample sets or your own data when executing the workflows. Here’s a breakdown of the workflows for today’s launch: ✅ Ingest Datasets: Ingest datasets into ApertureDB, from a selection of preconfigured public datasets and explore how it can be used with real data. ✅ Automated Embedding Generation: Generate embeddings for images, then use these embeddings to search for similar images or classify them based on the embeddings. ✅ Detect Faces: Add bounding boxes for faces detected within images in ApertureDB. ✅ Detect Object: Add labelled bounding boxes for objects detected within images in ApertureDB ✅ Direct Jupyter Notebook Access: Run Jupyter notebooks with seamless access to your ApertureDB instance. 🔥 Why ApertureData Multimodal AI Workflows stand out? To deliver a meaningful AI solution built on rich data, today’s infrastructure requires stitching together disparate a number of tools. ApertureDB workflows offer a unified approach. 💸Use Cases From e-commerce, smart retail, to inspection and life sciences, whether you are building classic ML, analytics or Generative AI applications, these workflows empower users to build advanced AI applications with ease. 😍 Want to Try it Out? Sign Up for a free 30-day trial of ApertureDB - https://cloud.aperturedata.io/si... We’d love your feedback — share your thoughts at team@aperturedata.io Have questions or need help – we are available on Slack A special perk for the Product Hunt community: Join our Virtual Lunch and Learn today (March 13th, 2025 @ 9AM PST ) for a deep dive into Workflows. There is still time to Save Your Spot - https://lu.ma/vnabtolp! What’s Coming Next? We’re continuously expanding our workflow library (available to the community on Github) to cover more use cases and integrations. 📢 Soon, you’ll be able to: * Integrate with more data sources and AI models * Launch embedding, graph, and attribute classification for other data types beyond images * Access advanced customization options for specific business needs. Huge thanks to this amazing community — your support is invaluable! Let’s build the future of multimodal AI together. 🚀 Best Regards, Vishakha Gupta CEO, ApertureData
Vishakha Gupta
Maker

If you want to see a live demo of how you can use workflows, do join us for our lunch & learn in the morning at 9am PT
https://lu.ma/vnabtolp

Gautam Saluja
Maker

@vishakha_gupta4 @hamza_afzal_butt The best part is that what the workflows do is open source. While the workflows on the cloud UI are a subset of possibilities, this repository has the all the detailed workings of workflows under the hood.


With this repository as a reference guide, following are the possibilities:

  1. You may refer to what those scripts are doing to get a blue print for building your own workflow.

  2. You may submit a PR. A PR for any custom workflow would be highly encouraged. TIA.

  3. If it is a general enough workflow, it would eventually get published on the cloud UI too!


Vishakha Gupta
Maker

@hamza_afzal_butt  do join in the lunch & learn happening now - it's one of the things Luis can answer showing how to from the repo as Gautam described : https://lu.ma/vnabtolp


Eoin McMillan

Hi Vishakha – How does ApertureDB compare to alternatives in terms of read/write speed and query performance on both small and large datasets? Additionally, does it have any unique optimizations or "special sauce" for faster token processing?

Vishakha Gupta
Maker

@mceoin great question - we have some recent benchmarking results summarized here: https://docs.aperturedata.io/category/benchmarks--comparisons

Mainly, for vector search, we are anywhere between 2-10X faster in terms of KNN throughput and offer sub-10msec latencies on service side. For graph search, our prior evaluations against Neo4j put us sometimes over 30X faster. Mainly, ApertureDB continues to scale for very large workloads (Billion scale graphs so far and 10s and millions of embeddings per search space). We have optimizations when we load data - so far we have tested it more on parallel load of large number of blobs or images - we can extend that to faster token processing though we are yet to test it.

Eoin McMillan

@vishakha_gupta4 30x Neo4j is very impressive. Will have to check it out!

Vishakha Gupta
Maker

@mceoin let's set up time to chat - would love to understand your use case and see if we can collaborate.

Michael Vandi

This is a game-changer for AI developers! Congrats on the launch @ApertureDB

Vishakha Gupta
Maker

@michael_vandi thanks a lot. We are happy to be working with you all!

David Aronchick

This is the hidden missing piece in SO MANY ML workloads. Great work by the ApetureDB team!

Vishakha Gupta
Maker

Thank you @aronchick we look forward to our collaborative examples coming in the near future to demonstrate how everyone can use these end to end even starting from edge to query

Mahima Manik

Love this! Super useful for devs. Congrats on the launch!

Vishakha Gupta
Maker

@mahima_manik thank you for your support. Looking forward to integrating this with Datahawk!

Peter Bordes
A well deserved congratulations from all of us at Trajectory. This is such a much needed platform that moves AI forward globally.
Vishakha Gupta
Maker

@peterbordes thank you ! we are seeing more and more people realize the need for the combined solution that we offer. It is hard to do vector in one, graph in another , data in a third place. Starts to wear people out as they scale and try to keep up with the rate at which AI is evolving

Deniece Moxy
Maker

👋 Hey Product Hunt,

We’re kicking off Summer of Workflows — a 12-week series of ApertureDB Multimodal AI Workflows designed to accelerate how developers build their AI agents and GenAI applications with multimodal data.

Release #1 is here: 📥 Croissant Ingestion Workflow — A powerful, ready-to-run workflow that ingests datasets defined using the MLCommons Croissant format directly into ApertureDB.

▶️ Watch the 2-min Demo

🛠️ What it does:

  • Parses Croissant metadata Downloads referenced assets (images, video, text, etc.)

  • Ingests them into ApertureDB — keeping structure, attributes, and relationships fully intact

  • Sets you up for faster AI/ML experimentation, production pipelines, and even RAG agents.

This workflow works out of the box with many datasets hosted by MLCommons, Hugging Face, Kaggle, and other public sources.

💡 Why it matters:

Multimodal data is everywhere — but it’s messy, inconsistent, and scattered. Croissant (by MLCommons) solves the format problem. This workflow solves the ingestion problem.

🔗 Try it now:

→ Run it in the cloud
→ Read the docs

💬 Tell us what you think about this workflow and what you'd like to see us build next. Hit the comments with your ideas—We're listening!

This is just the beginning. One new workflow will launch every Wednesday this summer. Follow along and let us know what you build!

-Team ApertureData

nina k

Great team + really interesting space — multimodal feels like one of the key themes this year.

Vishakha Gupta
Maker

If you couldn’t make it to our lunch & learn session demonstrating how to Launch Your AI Project Today with ApertureDB Multimodal AI Workflows—you can watch the full session on-demand now!


​Learn how to:

  • ​Launch AI workflows in minutes – No complex setup required

  • ​Use pre-built solutions for image embeddings, classification & more

  • ​Accelerate AI development – Focus on building, not infrastructure

​Bonus: Get access to sample code and resources to kickstart your AI project!

WATCH ON DEMAND NOW

View PDF Slides


You can of course try it out and tell us what other workflows you would like to see on https://cloud.aperturedata.io

Vishakha Gupta
Maker

A few months ago, I found myself deep in a conversation about the future of retrieval—how we move beyond simple keyword or vector search into something more connected and context-aware. That’s when GraphRAG came up.The idea of combining graph and vector data isn’t just theoretically elegant—it’s powerful and now pretty practical. It lets you go from a sea of unstructured chunks to a semantically linked knowledge web. That’s why I’m excited to share this introductory blog on GraphRAG with ApertureDB, where we blend the strengths of knowledge graphs with the precision of vector search. It’s the first step in showcasing how ApertureDB makes this possible, and there's a more advanced example coming soon.


Check it out here:
Enhanced Retrieval with GraphRAG and ApertureDB


If you’re curious about the mechanics under the hood—like how schema and knowledge graphs work in ApertureDB—here’s the relevant documentation:
Knowledge Graphs and Schema in ApertureDB


Would love to hear your thoughts—have you tried blending graphs and vectors yet?

Deniece Moxy
Maker

Hi Product Hunt Community,

You Asked, We Delivered: ApertureDB Cloud Passwordless Login Is Here!

We heard you - it was complicated to setup and login to your ApertureDB instance. You can now use passwordless login to ApertureDB, generate API tokens, and it's particularly easy on ApertureDB Cloud!

No more passwords at instance creation

You no longer need to set or remember passwords when spinning up a new instance. Instead, simply generate a secure access token during setup.

  • Users can generate secure access tokens directly from the console.

  • These tokens grant immediate access to the Web UI—no login prompt required.

  • Tokens are also used to authenticate via the Python SDK and other client tools.

Benefits

  • Improved developer workflo

  • API keys can be auto-generated for use in scripts, automation, or third-party services.

  • Eliminates password reset issues and user lockouts

  • Reduces credential sprawl across projects

  • Enables more secure, scalable integration patterns

  • Consistent experience across UI and API access

Learn more: https://docs.aperturedata.io/Setup/client/configuration

Start your free 30-day trial of ApertureDB today!

Deniece Moxy
Maker

Hi Product Hunt,

📢 New ApertureDB Workflow Release Alert!

Summer of Workflows rolls on! We are excited to release MCP Server Workflow — a MCP (Model Context Protocol) server that connects directly to your ApertureDB Cloud instance.

🎬 See It In Action

This workflow gives your Generative AI models and AI agents live, multimodal memory—enabling real-time access to images, text, video, embeddings, and more.

🔍 Why it matters:

Static context limits what AI agents can do. With MCP + ApertureDB, your LLMs can now query fresh, contextual information as they reason, plan, and act.

✅ What’s included:

  • A deployable MCP-compliant server - Zero glue code needed

  • Works out-of-the-box with ApertureDB Cloud

  • Built-in authentication for secure, production-ready deployment

👉 Try ApertureDB MCP Server Workflow


We are building the memory layer for Generative AI. Let us know in the comments what you would build with real-time LLM memory!


Additional Resources

Docs

GitHub Code

Vishakha Gupta
Maker

@deniece_moxy this is great! Anyone using ApertureDB can easily set up their MCP server for agent memory