Activeloop AI Knowledge Agent - Deep Research on Your Multi-Modal Data
Need to find answers to hard questions across multiple sources, including your private data? Use our Knowledge Agents, powered by AI search to scan up to billions of rows of any data - images, PDFs, text, tables and more, and provide a well-researched answer.
Replies
Considering the most valuable data tends to be in-house, this is amazing 🤩 Great work, @mikayel_harut !
Deep Lake - AI Knowledge Agent
@nicozensara thank you so much <3
Your data is your ultimate competitive advantage! Leveraging it effectively isn’t just an option anymore - it’s the key to staying ahead. Exciting to see solutions like Activeloop Agent unlocking its full potential, driving smarter decisions, and creating real impact!
How does it handle data quality and relevance when dealing with diverse sources ?
Deep Lake - AI Knowledge Agent
@gerasimh Thank you for your message! The system is based on a multimodal retrieval system, capable of obtaining the most relevant information in response to the user's query.
Through a process of data analysis and aggregation, it can provide surprisingly accurate answers. All of this is made possible thanks to the performance and flexibility offered by our database, Deep Lake.
Deep Lake - AI Knowledge Agent
@gerasimh one more additional point to Emanuele's - we learn from user queries over time to suggest more relevant information! And additionally, one surprisingly good way of increasing response quality is vision-language models -> OCR pipelines while performing well, are slightly clunky... Having an end-to-end neural search helps to get full context from the data across modalities, increasing response quality.
Wow, this is exciting! At Cloudchipr, we store a vast amount of data in object storages with diverse structures - from CSVs to time series and key-value data. This is a game-changer for us in generating various general statistics and empowering customers to "talk to with their data" without being restricted by data type.
Deep Lake - AI Knowledge Agent
@ashot_ayvazyan1 Exactly! You definitely need to try our tool and let us know what you think. Your feedback would be incredibly valuable!!
Deep Lake - AI Knowledge Agent
@ashot_ayvazyan1 thank you so much for sharing the use case - actually, you can copy the fine-grained access from your cloud provider to Activeloop, making it possible to restrict certain users of asking questions on only specific data.
Bardeen
Deep Lake - AI Knowledge Agent
@artem_harutyunyan out of the box, the AI Knowledge Agent integrates with your favorite cloud storage providers (Azure, GCP, AWS), Dropbox (with more storage integrations underway). We also integrate with OpenAI, AWS Bedrock, and other model providers (coming soon)!
Congrats @david_buniatyan this looks amazing!
I wonder if you tried doing Deep Research on top of images. Like from big pool of images finding those with a specific object or color contrast.
Deep Lake - AI Knowledge Agent
@david_buniatyan @tigranbs yes it's available right now:)
Perplexity
This would be very useful at every company I have worked in the past.
Deep Lake - AI Knowledge Agent
@ohharsen thank you!
Deep Lake - AI Knowledge Agent
@dmitriy_stepanov thank you!
Deep Lake - AI Knowledge Agent
@avag_simonyan hi there! Actually yes - we have a bit technical blogpost talking about diagram-related use case. This doesn't use the AI Knowledge Agent but if it's possible with just code, it would be even easier through the AI Knowledge Agent UI. Handwritten comments wouldn't be an issue (no more than for an OCR). Specifically for architecture, we haven't tested it but should be extensible from other blueprints without an issue!
Would love to chat about the use case with you.
Here are few things that I really liked in my first experiments.
1) Being able to use lots of public data sets before porting my own. Gave use good idea what the usability would be like so it was really helpful.
2) Found the onboard for data access quiet helpful, lots of things were present that were really on point, e.g. CORS configs.
3) Really liked step by step presentation of things before final output would be generated. Would want to see more control there.
Example, when doing a search it allows to see "Generating TQL", which is great, now I want to see what would happen if I were to change TQL itself. Could I see side by side data? Could I see performance metrics. Not all queries are equal in backend side so may be I want to speed things up or help things get to better results.
I can see that the platform does have the ingredients so looking to explore further and give more updates here.
I've been exploring data visualization tools lately, and managing raw, unstructured datasets is always a major hurdle. An agent that can autonomously perform deep research and structure this data is fascinating. How easily can its findings be exported into formats readable by tools like Power BI or a custom React charting library?