We are Seaplane and this is what we believe…
We, the humble members of the Seaplane team, believe it is important to share our core beliefs with you and provide you with Seaplane’s vision of building an inclusive community of empowered AI developers. We aim to uphold our beliefs and values while creating a space where AI developers can join us in this exciting journey in 2024.
Over the next 60 days, we will be expanding on our thinking on each of these rules with blogs of their own, but here’s the quick articulation of the full list - we’ve spent a lot of time thinking about these and a lot of time talking to hundreds of AI professionals (from data scientists and ML researchers, to ML engineers and ML Ops teams) to distill this set of core beliefs we see as essential to AI applications successfully reaching truly mass-scale.
To plant our flag in the AI landscape, the following is a set of principles we've crafted to serve as a quick litmus test for our community. It will help you ascertain if you, too, are itching for a change. Are you exhausted from investing extra hours in non-differentiating cloud infrastructure tasks for your company? Are you frustrated with the need to hire expensive resources in pursuit of promises unfulfilled by traditional MLops and LLMops platforms? I know we are. Over the past few years, viable alternatives have been scarce – that is, until now.
Rule 0: We believe every app will be an AI-infused app in the future
AI is here, and even if you think AI can’t beat you, deep down you know someone using AI probably can. We use the term infused since we know AI (this includes Machine Learning (ML) and Data Science (DS)) is becoming part of every system. Co-pilots, chatbots, and computer vision are already part of daily life, and without including these in your product, it fall behind. Unfortunately, the large cloud providers were built to serve legacy applications and are not equipped to onboard the vast number of people who need to innovate in the emerging AI space. They (CSPs) have simply made it too hard or too complex for the majority of developers out there to keep pace with innovation. We know there is a simpler way. We built it.
Rule 1: We believe the Developer Experience (DX) is all that matters
We have listened. You have been loud and clear. Seaplane ensures your app is "always on" but without excessive costs during idle times. We offer global deployment with strategically located stateful partitions for your users worldwide. All this is achieved without you needing to delve into the minutiae of infra/platform micromanagement. By avoiding the complexity of writing and managing an endless bundle of micro-services and fragmented functions, we’re not only enhancing reliability but also cutting down your development and operational costs. Our customers, the ones crafting these AI-infused apps, get to sidestep that entire chaotic world.
Rule 2: We believe in "Region: Earth” and “Region: Regulatory Zone"
Businesses can achieve global scale while respecting local data-sovereignty.<NameYourCSP> East vs <NameYourCSP> West is a broken, outdated approach that modern AI developers should not be forced into. Where do you want your application to live? Everywhere. I want it to live everywhere my users care about. Don’t even get us started on multi-cloud, multi-region as we don’t want to hear it and we know you don’t either. Seaplane will never expose you to that pain train, we’d rather you fly First Class.
Rule 3: We believe that mixing business logic with configuration is a sub-optimal tradeoff
That's why, for three years now, our core focus has been on delivering modern apps globally. With our Python SDK and AI-infused runtime, we've opened up a new realm for AI creators to tap into the power of Seaplane. Our aim has always been to keep things at a high-level, an application-oriented vibe, ensuring that as many folks as possible can build and run things. Our abstraction level is much higher than your typical PaaS without compromising any functionality – that's always been our approach!
Rule 4: We believe cloud infrastructure configurations have become too complex
Especially at scale, AI Developers are tired of fighting with multi-cloud, multi-region, lack of GPUs where you need them and all things that make your life miserable when trying to deploy and operate in production. The Seaplane Platform unveils simply and operates for you a global edge network, stateless user compute, and a suite of stateful services like SQL, Object Stores, and Vector Databases; all delivered with a page of Python. Now, add all your favorite LLMs, stateless edge GPUs and a global no-cost to egress data network to the mix, and you've got the complete AI package. It is that simple and there are not thirty or more training sessions you have to take to get “Hello World” to work. With Seaplane, write a page of Python and go to production, today.
Rule 5: We believe in a Total Cost of Ownership pricing model
After you’ve managed to configure a large CSP, the pricing might not be clear. Ever received a surprise Cloud bill because you left a GPU machine running for a weekend (or month!)? Has anyone on your team executed a multi-join select * dataset and egress fees punched you in the face? Us neither 😅😬(NOT). Well Seaplane allows you to define your peak monthly expense for an AI App and handles the rest. Only paying for what you use and focusing on TCO are core themes you will experience here at Seaplane.
Rule 6: We believe developers should profit from their work in a fair Marketplace.
Thousands of professional services firms are “<NameYourCSP> Certified” but these Boutique Data Science Firms only get paid by the hour by their clients to make the CSP rich in perpetuity. Seaplane is building a marketplace for apps (think Huggingface but for more than models, entire applications or parts of an application). Seaplane’s Partner team will work with you to get that app on the platform, available to use by the customer base, and put revenue streams in place to help you grow your small business.
Rule 7: We believe depending on one Foundational Model (FM) for your business is risky business.
On the Seaplane Platform, the Foundational Models (FM) and top open-source generative AI models are at your fingertips with a simple Python function. Even more so, we have reference designs for auto-switching models when your preferred LLM is down, slow, or too busy for you. Our customers achieve resilience and provide their clients and users the best experiences possible—eliminating the helpless feeling that many have had when they only rely on one not-so “Open” solution.
Rule 8: We believe in Hybrid AI solutions
We believe in the evolution of hybrid AI. The trajectory is set to converge on a balanced approach between expensive foundational models such as ChatGPT or Gemini and the integration of smaller, finely-tuned open source AI models. There is a collaborative alliance between these Large and Small models, where the extraordinary reasoning capabilities of larger models need to harmonize seamlessly with the agility of smaller models which are fine-tuned and/or optimized by our clients to ensure the best cost, data privacy, and increased user experience (low-latency responses). Dialing in these ratios will require increasing collaboration between product and data science teams.
Rule 9: We believe data scientists shouldn’t have to wait for someone else to deploy their apps.
According to the 350 AI practitioners we’ve interviewed, 90% of data science and machine learning projects fail and the top issues boil down to projects fail due to complexity in getting from notebook to production. (Data quality / wrangling are the other top issues). Clearly, there is a lack of people, process, and technology enabling producers to experiment, iterate, and deploy insights quickly. The elephant in the room? DevOps teams are just too busy or too expensive and must be dedicated to low-risk, high-reward core business features. With Seaplane, your MVP is your production deployment. Just set a feature-flag.
Rule 10: We believe the AI Landscape will rapidly change, and your deployment platform should keep up.
You’ve heard this before, “keeping up with the pace of innovation is a challenge”, but in the AI era, it is practically impossible! It seems every day brings a new LLM challenger to the throne—the “next ChatGPT” is here and the world (and your boss) expect you to know about it. In other parts of the stack, GPU technology is developing rapidly. Nvidia leads the pack with options for A100, H100, H200, B100, R100, (whew! trying to keep up like 😥), but direct competitors AMD, Qualcomm, and Intel are not taking that lying down, and the CSPs are making in-house GPUs (and/or TPUs). Each comes with its own interfaces, software stacks, and APIs, compounding technical potholes for you geometrically! Being a “<NameYourCSP> shop” is becoming an anchor not a propeller. What is needed is a ‘supercloud’ that abstracts all of that complexity away and lands you in the central spot of the Venn Diagram, the best of the best of the best; that’s the space where Seaplane resides and welcomes you.
Rule 11: We believe AI-infused apps need to deliver provable ROI (asap).
As we said before, 90% of machine learning projects fail. Once deployed, AI practitioners have challenges associating their insights to business metrics and positively correlating experiments to business outcomes. Experimentation to demonstrate solutions quickly with full visibility into results alongside cost projections are needed to help AI teams build, decide, and deploy applications that exceed their minimum ROI requirements. What we know is that higher ROI is based on these factors 1) Value Created; 2) Scale of usage; 3) Efficiency of Time; 4) Flexibility or Adaptability; 5) First Mover Advantage; 6) Price or Investment. We focus on the success of each of these items while we mainly focus on your success.
Rule 12: We believe anyone with a great idea (and minimal python chops!) should be able to get an app in front of a user anywhere in the world, today.
Jack or Jills of all Trades is a Master of None is a phrase that helps explain focus, or lack thereof. If you are reading this, it is likely that being a Kubernetes (K8S) expert is not what makes your company unique. It is also likely that running and maintaining a large suite of technologies like DNS, Load Balancers, YAML configs, code, networks and the needed additional software stacks like databases, vector databases, GPU software, and more will NOT help you win the race against your AI competitors. Time-To-Value is what will win. Thus, we’ve created a very high level of abstraction as close to the Application layer as possible so you can just create, test, deploy knowing only Python. (Which we’ve learned from our hundreds of research interviews is a comfort-zone for most data scientists.) We take care of the rest: all the things you’d need a team of cloud experts to do otherwise. We will even carry the pager for you day and night.
Rule 13: We believe all AI should be delivered at the EDGE, everywhere in the world
For reasons of performance, efficiency of data movement, and regulatory compliance, running your AI-infused application at the edge, close to your end-user just makes sense. And when your end-users are all over the world, there’s still no reason you can’t serve them all: locally, dynamically and magically scaling as they appear and disappear. Sometimes though it’s more cost efficient to run where the traffic isn’t - Seaplane helps optimize for you either way, without you having to wrangle machines, data, compute, storage and networks in regions all over the globe. Giddy Up!
Rule 14: We believe models are the start, not the be-all, end-all of AI
While models - be it foundational LLMs, open-source models optimized on a particular domain, or even your own trained models on your proprietary data - are critical to ML and AI offerings, we believe they are only one element to delivering value. As everyone experimenting in 2023 has discovered, it’s easy to get terrible results from a model (like GPT) if you don’t have your workflow and your application suitably and efficiently configured. Seaplane is designed to remove complexity from this process meaning less surface-area to have to manage and less ways for things to go wrong. Delivering value quickly also means making your data scientists and ML teams more productive, allowing them to focus on core competencies and deploy faster. Finally, when you get to scale, efficiently and cost-effectively delivering to users wherever they are maximizes ROI. Seaplane is designed to be the simpler, faster, better onramp to all of this.
Summary:
This is what is needed to succeed in 2024. It is what we believe. It is what we built. It is what will empower you to win the AI race, while others will have 90% of their projects fail and remain stuck on a whiteboard waiting for deployment.
From a seamless Developer Experience (DX) and transparent pricing structures to making global edge GPU infra simple, we are pulling back the curtain on what drives our approach. By openly sharing these core tenets or rules, we aim to connect with like-minded individuals who share our vision and values. At the heart of it all, we're not just offering a product – we're inviting those who believe what we believe to join us on a journey where simplicity, transparency, and shared values pave the way for a more efficient and collaborative future. If you're passionate about aligning with a company that shares these convictions, we’d ask you to sign up for the beta and get started today.
Bring Your Apps to the World