-
Notifications
You must be signed in to change notification settings - Fork 13
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add Search improvement blog #617
base: main
Are you sure you want to change the base?
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The build is failing because of the gif too. Looking into fixing it right now
This pull request is automatically being deployed by Amplify Hosting (learn more). |
src/content/blog/2024-08-28-search-improvement-tembo-cloud/index.mdx
Outdated
Show resolved
Hide resolved
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
First, I reviewed for grammar review and small suggestion, then took a higher-level look at it.
In my opinion, there is greater opportunity in this blog to make a case for why using Tembo's feature for search makes sense, and emphasize that this is a simple and effective way to do search in comparison to the status quo of UI search. We should help people understand why using pg_vector and LLMs for search is the new and cool way to do it, making the content more sharable within engineering teams that may be looking to implement search. I also think that can make this blog more discoverable to those searching online for how to build a search bar using Postgres and pg_vector, which may be something people are interested in.
So overall, I think the audience is more interested in "why do it this way" and "how to do it this way" than what this feature does and the fact that there is a search bar on Tembo. i.e. we are trying to sell the vector and AI features of tembo as compelling options for users rather than trying to sell tembo cloud by saying now we have a search bar.
src/content/blog/2024-08-28-search-improvement-tembo-cloud/index.mdx
Outdated
Show resolved
Hide resolved
src/content/blog/2024-08-28-search-improvement-tembo-cloud/index.mdx
Outdated
Show resolved
Hide resolved
src/content/blog/2024-08-28-search-improvement-tembo-cloud/index.mdx
Outdated
Show resolved
Hide resolved
src/content/blog/2024-08-28-search-improvement-tembo-cloud/index.mdx
Outdated
Show resolved
Hide resolved
src/content/blog/2024-08-28-search-improvement-tembo-cloud/index.mdx
Outdated
Show resolved
Hide resolved
src/content/blog/2024-08-28-search-improvement-tembo-cloud/index.mdx
Outdated
Show resolved
Hide resolved
src/content/blog/2024-08-28-search-improvement-tembo-cloud/index.mdx
Outdated
Show resolved
Hide resolved
src/content/blog/2024-08-28-search-improvement-tembo-cloud/index.mdx
Outdated
Show resolved
Hide resolved
src/content/blog/2024-08-28-search-improvement-tembo-cloud/index.mdx
Outdated
Show resolved
Hide resolved
src/content/blog/2024-08-28-search-improvement-tembo-cloud/index.mdx
Outdated
Show resolved
Hide resolved
I almost completely agree with everything @sjmiller609 has said. Let's focus on the underlying Postgres based tech. The blog should be structured more to emphasize on: "Improved search in Tembo powered by Postgres and AI" (See: https://tembo.io/blog/tembo-data-warehouse for a template) My proposed structure would be something like: Intro to the functionality we've addedSome cool / example searches you can do and Ask functionalityBuilding search and ask on postgresExplain why we feel building this on Postgres is a good idea. Avoiding Tool sprawl etc. Powered by Postgres, pgvector and Tembo AIExplain how we thanks to Tembo AI, pgvector and Postgres building all these apps is super straightforward. Explain the architecture and maybe an API demo Upcoming featuresLet's also say we'll have a follow up blog on how you can just add search and ask functionality to any docs you have powered by Postgres |
…ex.mdx Co-authored-by: Steven Miller <sjmiller609@gmail.com>
|
||
Our Search is underpinned by the search and command widget tool [cmdk](https://github.com/pacocoursey/cmdk), fuzzy-search library [fuse.js](https://www.fusejs.io/), and vector search. We use vector search for the Docs, dogfooding our [pgvectorize extension](https://pgt.dev/extensions/vectorize), embeddings and [pgvector](https://pgt.dev/extensions/pgvector). | ||
We’ve also introduced an AI agent for more complex queries. _Ask Tembo_ uses 2 custom [Tembo Apps](https://tembo.io/docs/product/cloud/apps/overview), deployed and hosted on Tembo Cloud. It uses a LLM hosted on Tembo. If you do a search in the Cloud UI, you're logged in and we understand your context, your results are bound to be most relevant. | ||
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Now that we changed from product update to "here's what we did and how you can do it too" I'm going to need a lot more meat here about which apps used and how someone can replay the same script for their application @aishwaryaborkar @ChuckHend
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I can write that blog but I think we should sync up about timing for it and not push that content into this PR. There's a lot to it and we will likely need to cut a lot of content to fit into a single blog. Or maybe a better idea would be to make it a series of blogs on 'building chat applications on Postgres'.
@samay-sharma, @FloorD, what are your thoughts on pushing out this as a 'product update blog' and then working on something bigger to cover the 'here's how you can do it too'?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
If we'd do that, will we push search for the blog as well at the same time?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Cause then it'd just be:
Improved search experience for Tembo Cloud
When you’re in Tembo’s Cloud UI you might need to look up a thing you’re trying to accomplish. Rather than always having our Docs open in a second window, you can now use our improved search functionality to search our Docs, blog and website straight from cloud.tembo.io.
With our improved search capabilities you can also search within your Instances (in your currently selected Organisation - if you have multiple Organisations, you must first navigate to the appropriate Organisation).
To top it off we’ve introduced an AI agent for your more complex queries.
Search within Tembo Cloud looks like so:
GIF
You can use Cmd+K or Ctrl+K to open the search window as well.
Toggle Ask Tembo to ask our AI assistant a question:
GIF
Tembo-powered search for your application
“In the olden days” we used Algolia for Docs search, but unsatisfied with the results we wanted to try and roll our own. We will share our setup as an open source Postgres-powered alternative for search in an upcoming blog post.
What’s next?
In terms of further improvement we want to offer suggestions for queries (FAQs), and show users their recent search queries. We’re also thinking of introducing a feedback mechanism if the selection doesn't match what you were looking for.
We could also make an attempt at showing relevant results for Ask Tembo immediately while processing the query in the backend, and then show more relevant results once the query is completed.
As always, we love to hear from you about functionality that will make your life easier. Please suggest and upvote features via [roadmap.tembo.io/](https://roadmap.tembo.io/roadmap)!
Adds search blog for review.