Last week I had the opportunity to attend MongoDB.local NYC and help out at the MongoDB Community Booth. We had a fun “scavenger hunt” around the MongoDB Community website to get a MongoDB t-shirt which saw a lot interest (everyone loves a free t-shirt!)

It was great to catch up with many members of the MongoDB Community including the MongoDB Champions and Creators.

Hi again if I saw you there!

Make sure to checkout the full announcements post from MongoDB, but here are some of my thoughts and experiences at the MongoDB Developer Day before the conference, the AI Build Together after, and the .local NYC itself!

Key Note: Tech Adoption Happens in S-curves

The overriding theme of the event was everything AI, and how MongoDB is making working with AI easier than ever. The key note really clarified what using AI with MongoDB can be all about, and how typically these types of technologies see adoption in waves. At first there is a lot of interest, and then a period of what seems like a “fizzle” as the “hype” dies down. However for really core technological changes (MongoDB clearly sees AI as one) there will be a constant upwards trend over time in an S-curve.

Some of the key ways MongoDB can help you leverage AI have seen additions and improvement this last year including improvements to Vector Search, the $search operator and more. Not only that but both Vector Search and $search will soon be coming to the Community Addition of MongoDB!

That last bit was an exciting announcement for a number of developers I spoke with at the conference as many of them are still in the first phases of testing out messing around with AI. Being able to take advantage of these features within the Community Addition is great for tinkerers, but also don’t forget you can get a free Atlas cluster which enables most of these features too.

MongoDB AI Application Platform

Speaking of making AI powered apps, also announced was the development of the MongoDB AI Applications Platform (MAAP), a partnership between MongoDB and leading AI companies to help organizations of all sizes to use AI in their apps:

With MAAP, we give customers the blueprints and reference architectures to easily understand how to build AI applications. We also take on the heavy lifting of integrating MongoDB’s developer data platform with leading AI partners like Anthropic, Cohere, Fireworks AI, Langchain, LlamaIndex, Nomic, Anyscale, Credal.ai, and Together AI, all running on the cloud provider of your choice. MAAP will be available to customers in early access starting in July.

Once developers get to building AI apps, they’ll find that MongoDB allows them to speak the data “language” of AI. Our developer data platform unifies all different data types alongside your real-time operational data—including source data, vector embeddings, metadata, and generated data—and supports a broad range of use cases.

You can find a lot more AI resources for MongoDB on the Community Resources site.

MongoDB Developer Day NYC and AI Build Together Event

Before and after the .local I was also able to attend two different AI focused events for developers: The MongoDB Developer Day NYC and a AI “build together” event the evening of .local.

You can checkout all the things we hacked on during the developer day here, and try some of it out yourself if you want!

For the build together event we split up into small groups at tables around themes we cared about. I chose team “Education” and we discussed about how we could use AI to help reduce teacher “burn out” as well as better tailer individual success plans for students.

It was a blast to meet a lot of great people from a number of different backgrounds and discuss ideas to solve our problem statements. In the end we came up with a couple of ideas:

• Leverage LLMs with a Constitutional AI (like Claude AI) to help teachers craft lesson plans and curriculum that is personalized yet also makes sure to include required aspects, terms and so on required by school systems.
• Create daily short “quiz” for each student using AI, which would be personalized and adjusted each day based on the previous days results.
• This way each individual child can be quizzed and receive personalized homework each night. Using a vector database and AI / RAG this system can offer multi-modal resources to help each child learn, bringing togethers textbooks, videos, library books and more which we’d feed into our vector database and use an LLM to query.

We had a lot of great advisers to help us along the way as well. The event was very well received from everyone I spoke with!

MongoDB 8.0 Preview

Also announced were a number of improvements, speed enhancements and quality of life changes coming in MongoDB 8.0!

MongoDB 8.0 is all about stepping up the game for modern and AI apps, making them faster, more scalable, and tougher. They’ve really put the focus on making life easier for developers while still giving apps the muscle and security they need.

Key Updates in MongoDB 8.0

Here’s my takes:

• First off, they’ve turbocharged performance by making data queries and transformations lightning-fast, thanks to upgraded memory tricks and slick batch write commands.
• Then, they’ve got your back when things get crazy busy with better queuing and smarter ways to handle those sudden spikes in demand.
• Plus, they’ve made scaling up a breeze, so your app can grow without hitting any walls.
• And let’s not forget about time series data – super important for IoT and analytics – they’ve souped that up too, making it faster and more efficient with nifty compression tricks.

You can read more in the release notes for the MongoDB 8.0 Preview …

• Sharding configuration servers can be embedded in sharded clusters, which simplifies the architecture and lowers the overall infrastructure cost without impacting scale and performance.
• Unsharded collections in databases can be moved to different shards without the need for additional configuration such as sharding the collections and selecting a shard key. This makes it easier than ever to horizontally scale and distribute collections geographically for compliance reasons.
• Balancing data across shards via resharding is now 4x faster based on internal tests, and this capability now extends to time series collections, which reduces the operational costs and complexity of maintaining the database architecture.

Just remember, while this preview version looks promising, not all the bells and whistles might be here yet. And hey, there’s always a bit of risk in the mix, so keep your eyes peeled for any surprises.

Conclusion

Overall the event felt like a real success and the events before and after the conference where devs could get much more hands/head on were a real hit. I’m excited to see what will be built with all these new tools!