Reviews from AWS Marketplace
0 AWS reviews
-
5 star0
-
4 star0
-
3 star0
-
2 star0
-
1 star0
External reviews
External reviews are not included in the AWS star rating for the product.
A Very Powerful and Promising Tool
What do you like best about the product?
I chose Vectara to power the search on docs.qumulo.com after investigating a number of other AI search solutions. The tech is now wired directly into the Qumulo Documentation Portal—in addition to other Qumulo websites—searching every guide published there.
The feature that produces summaries in addition to lists is especially useful (let alone the feature that can produce summaries in the language of the original query—we've yet to implement this). Compared to the previous, keyword-based lookup solution we had, Vectara offers a completely new approach to finding information about deploying and configuring Qumulo Core.
Vectara is a very exciting tool that has incredible potential, and I like the fact that, each time I get stuck or run into the issue with GitHub repos such as `vectara-ingest` or `vectara-answer`, there is always a senior engineer to help resolve a filed GitHub issue, provide example code, or adjust the existing code in the repository—so far, everyone has been very responsive.
I like the fact that Vectara doesn't make me buy into an ecosystem such as AWS, Azure, or GCP. Any resources that Vectara uses are for Vectara to manage, which avoids vendor lock-in and also abstracts all the infrastructural issues from the boots-on-the ground deployment and configuration that the administrator performs.
The feature that produces summaries in addition to lists is especially useful (let alone the feature that can produce summaries in the language of the original query—we've yet to implement this). Compared to the previous, keyword-based lookup solution we had, Vectara offers a completely new approach to finding information about deploying and configuring Qumulo Core.
Vectara is a very exciting tool that has incredible potential, and I like the fact that, each time I get stuck or run into the issue with GitHub repos such as `vectara-ingest` or `vectara-answer`, there is always a senior engineer to help resolve a filed GitHub issue, provide example code, or adjust the existing code in the repository—so far, everyone has been very responsive.
I like the fact that Vectara doesn't make me buy into an ecosystem such as AWS, Azure, or GCP. Any resources that Vectara uses are for Vectara to manage, which avoids vendor lock-in and also abstracts all the infrastructural issues from the boots-on-the ground deployment and configuration that the administrator performs.
What do you dislike about the product?
Vectara has a very steep learning curve for someone who isn't a coder by trade and, while I'm moderately proficient with a variety of development frameworks and tools, I'm a documentarian, not a developer, which means that I prefer a turnkey solution: something to deploy, configure, and adjust within no more than 2-3 days in total. This isn't necessarily a serious concern—as Vectara grows and develops their offering, they'll have a much better definition for the main user personas.
The following might seem like a strange concern, but Vectara currently offers far too many examples and scenarios instead of offering solutions to the most *common* use cases. To close this gap for the time being, I recommend creating a foolproof, end-to-end Getting Started Guide that would give the user everything she needs to implement AI search for the most common scenario: a website on a domain (example.com) or subdomain (docs.example.com) that has a sitemap.
Currently, Vectara's models appear to be based mainly on variously configured flavours of OpenAI's ChatGPT-3.5 and ChatGPT-4. It seems that using the latest AI engines slows down Vectara's queries. While I'm certain that the tuning of the prompts and adjustments given to the models will be improved in short order, it might be nice to combine a variety of engines, especially for the summaries.
The following might seem like a strange concern, but Vectara currently offers far too many examples and scenarios instead of offering solutions to the most *common* use cases. To close this gap for the time being, I recommend creating a foolproof, end-to-end Getting Started Guide that would give the user everything she needs to implement AI search for the most common scenario: a website on a domain (example.com) or subdomain (docs.example.com) that has a sitemap.
Currently, Vectara's models appear to be based mainly on variously configured flavours of OpenAI's ChatGPT-3.5 and ChatGPT-4. It seems that using the latest AI engines slows down Vectara's queries. While I'm certain that the tuning of the prompts and adjustments given to the models will be improved in short order, it might be nice to combine a variety of engines, especially for the summaries.
What problems is the product solving and how is that benefiting you?
Vectara helps Qumulo customers, Customer Success representatives, engineers—and many others—find just the right information about configuring and integrating Qumulo Core with other systems or deploying it on hardware platforms.
- Leave a Comment |
- Mark review as helpful
Great product that delivers results as promised.
What do you like best about the product?
Vectara has enabled my team to build a robust and accurat Chat Assistant tool which provides our users with quick answers and detailed explinations based on our documentation.
Simplicity of Integration - Vectara provides a simple end-to-end solution that is easy to integrate with. This streamlines our focus on improving the Chat Assistant without complex integrations.
True to Its Word - Vectara offers accurate searches focused solely on our documentation, ensuring swift and accurate results with no hallucinations. Their product just works and does not require deep AI/ML knowledge to use.
Simplicity of Integration - Vectara provides a simple end-to-end solution that is easy to integrate with. This streamlines our focus on improving the Chat Assistant without complex integrations.
True to Its Word - Vectara offers accurate searches focused solely on our documentation, ensuring swift and accurate results with no hallucinations. Their product just works and does not require deep AI/ML knowledge to use.
What do you dislike about the product?
Integration with 3rd party Tools - It would be nice to have direct integration with common document tools to directly integrate with our document sources. Integration with complimentory AI tools would also help simplify the GenAI stack.
Detailed Metrics - It would be nice to have performance metrics built into the dashboard beyond basic usage features.
Detailed Metrics - It would be nice to have performance metrics built into the dashboard beyond basic usage features.
What problems is the product solving and how is that benefiting you?
Focused and controlled searches of our companies documenation using a vector database
Removes hallucinations seen in other AI Chatproducts
End-to-end search and summary tool that is easy to use.
Removes hallucinations seen in other AI Chatproducts
End-to-end search and summary tool that is easy to use.
showing 1 - 2