Hypermode Launch: Kevin Van Gundy

00:00:00.000 | Nick DiBona: Before hypermode, I worked at Vercel.

00:00:16.800 | We had an office down the street above a pizzeria,

00:00:19.320 | and we had three big problems.

00:00:21.760 | One, we were losing to other JavaScript frameworks.

00:00:25.520 | Two, we were losing badly to other hosting providers.

00:00:29.320 | And three, I was losing to my diet of exclusively pepperoni

00:00:32.560 | pizza.

00:00:34.980 | Eventually, we started to win, not because we were smart

00:00:38.560 | or we knew all the right answers,

00:00:40.400 | but because we dealt this core competency of iterating

00:00:42.640 | really, really, really quickly.

00:00:46.340 | We didn't know the optimum strategy,

00:00:47.760 | but we figured if we just tried more things faster

00:00:50.020 | than everyone else, we'd eventually

00:00:51.700 | be able to adapt and figure out the right products and strategies

00:00:54.620 | to figure out what the market wanted.

00:00:57.960 | Iteration is the compound interest of software.

00:01:00.960 | Keep doing it long enough, and eventually really good stuff

00:01:03.460 | starts to happen.

00:01:05.460 | Because we tried a lot of things really quickly,

00:01:08.160 | we eventually figured out two things.

00:01:10.200 | One, developers want to incrementally adopt new technologies.

00:01:14.080 | And two, they don't want to commit to architectural patterns

00:01:17.080 | before they know how their application is actually going to work.

00:01:19.580 | But iteration can't happen if you're afraid of getting it wrong.

00:01:26.580 | The same thing that has held back web is also holding back AI.

00:01:31.080 | And if I'm honest, there are even more things for us to get wrong about Gen AI.

00:01:34.580 | When I think about it, I'm grossly overwhelmed.

00:01:37.080 | What's the right hardware?

00:01:38.580 | What's the right model?

00:01:39.580 | What's the right prompt?

00:01:40.580 | How do I integrate?

00:01:41.080 | How do I monitor?

00:01:42.080 | How do I improve?

00:01:43.080 | Everyone here knows a horror story of someone with a runaway bill,

00:01:48.080 | a hallucinating chatbot, a project that took months and months

00:01:52.080 | and never delivered any value.

00:01:53.580 | And in the end, we need to build systems that de-risk getting it wrong.

00:01:58.080 | Because we are going to get it wrong a lot.

00:02:02.080 | Picking the wrong model doesn't matter if there's no friction to switching it out.

00:02:06.580 | Integration is simple when your classical systems and your AI systems use the same APIs.

00:02:11.580 | You can fearlessly make changes to prompt strategies, data mixes,

00:02:15.580 | if you can trace that inference step by step by step.

00:02:18.580 | At HyperMode, we care deeply about making AI approachable.

00:02:23.580 | Everyone here should be able to put AI in their apps without specialized skills.

00:02:27.580 | At its core, HyperMode is a runtime.

00:02:31.580 | It allows you to easily integrate models and data into AI functions.

00:02:35.580 | We then surround that runtime with a bunch of tools that make it easy for you to rapidly iterate

00:02:41.580 | and observe those AI functions in prod.

00:02:43.580 | We make it easy to get started, incrementally adopt AI as appropriate,

00:02:48.580 | and then as your team develops those skills, reimagine those applications as AI native.

00:02:54.580 | First and foremost, we want to make the developer experience of developing with AI a lot less terrible.

00:03:01.580 | When it comes to adding a new model to your service,

00:03:04.580 | you probably don't want to read a bunch of pages of docs to figure out the temperatures on a 0 to 2 rather than a 0 to 1 or a 1 to 10.

00:03:11.580 | With HyperMode, we provide you type ahead and your favorite code editor right out of the box.

00:03:15.580 | No SDKs, nothing to download.

00:03:19.580 | Then when you do ship to prod, we give you strong defaults just to get started.

00:03:23.580 | Or if you have your own stack, bring it along.

00:03:26.580 | In either case, we'll remove a lot of that complexity for you.

00:03:28.580 | For example, traditional RAG requires N+1 requests.

00:03:32.580 | You need to make an additional call to embed the inputs.

00:03:35.580 | Go talk to your vector store with HyperMode.

00:03:38.580 | You can do that all in one request.

00:03:41.580 | We've built an in-memory embedding and search service that will allow you to do that and save a couple of milliseconds per request.

00:03:47.580 | Finally, building intuition around non-deterministic systems is hard.

00:03:53.580 | Each model has its own personality, and we make it really easy for you to quickly compare different inferences, different tunes, different models.

00:04:01.580 | And you can then export this data set to fine tune.

00:04:05.580 | On Monday, your boss is going to ask you, what did you learn at AI World Fair?

00:04:10.580 | If you come by our workshop after lunch, I'll prove to you that you can make iteration velocity of core competency.

00:04:18.580 | The team that built all this amazing stuff will be there.

00:04:21.580 | We'll show you how to build natural language search, intelligently sort every data list in your product, detect outliers, catch bad guys.

00:04:28.580 | You'll walk over the demo that you're proud of and a plan to put something like it in prod by the end of next month.

00:04:33.580 | And if seeing my happy face again and building something really cool is not enough, we'll give you $1,000 in HyperMode credits to get started.

00:04:41.580 | Thank you all so much.

00:04:42.580 | Thank you all so much.

00:04:43.580 | Thank you all so much.

00:04:43.580 | Thank you.

00:04:44.580 | Thank you.

00:04:45.580 | Thank you.

00:04:46.580 | Thank you.

00:04:47.580 | Thank you.

00:04:48.580 | Thank you.

00:04:49.580 | Thank you.

00:04:50.580 | Thank you.

00:04:51.580 | Thank you.

00:04:52.580 | Thank you.

00:04:53.580 | Thank you.

00:04:54.580 | Thank you.

00:04:55.580 | Thank you.

00:04:56.580 | Thank you.

00:04:57.580 | Thank you.

00:04:58.580 | We'll see you next time.

Hypermode Launch: Kevin Van Gundy

Chapters