The GenAI Stack Landscape for Enterprises in 2024
As companies wrap up their fiscal year and plan for the next, AI becomes an integral part of every conversation. ChatGPT emerged seemingly "out of nowhere" for many at the end of 2022, catching most off guard, too abrupt for most to react and act in 2023.
In 2024, it will be a different story; a conscious choice will be made whether to embrace AI in your stack or wait.
With Goldman Sachs forecasting $200B in AI investments by 2025, companies are already exploring how to use their budget.
I certainly won’t be the first—or last—venturing to assist in demystifying the landscape of the GenAI tech market for enterprises. It's a fascinating exercise with much to explore and learn in the process.
Two Predictions for 2024
We're still in January, so why not kick off with two predictions.
Prediction #1: GenAI Evaluations Must Translate into Clear Benefits
Many organizations will focus on testing, experimentation, and evaluation of GenAI. However, it won’t be long before these investments need to demonstrate tangible value in production.
Prediction #2: GenAI Will Be Everywhere!
This implies that every application you use today—and will adopt tomorrow—will integrate some GenAI capabilities in one way or another. If you don’t go to GenAI, don't worry; GenAI will come to you!
However, GenAI within applications you purchase is only one side of the story—one where you have limited control, as these capabilities will be "built-in."
Where your organization will have more control is over GenAI capabilities and applications built internally.
This post focuses on the tools that will play a role in building an enterprise GenAI tech stack.
The Decisions To Build This GenAI Stack Market Guide
To avoid “boiling the ocean” 🌊 in an exercise that would never be finished, I’ve made some early decisions:
No (Business) Applications: I won’t be representing GenAI applications explicitly. As predicted earlier, GenAI is anticipated to become omnipresent in the future. Listing every company would be no different from preparing an infinite loop. However, I acknowledge the existence of applications specially crafted around GenAI!
No Vertical/Industry Category: Companies and products listed are predominantly industry-agnostic. This means you can consider them, regardless of the industry you operate in.
No “Format” Category: Some models and tools specialize in text, voice, image, and even video—these being considered the most common formats today. However, this distinction is not accounted for in this exercise.
While I don’t dismiss the value of digging deeper into categorization and considering the points mentioned above, maintaining content that is consumable and beneficial for most enterprise businesses requires some tradeoffs. Consider this post as a starting guide rather than the most exhaustive set of options to consider.
The 2024 GenAI Stack Landscape for Enterprises
Without further ado, let me present the result of my research over the past few weeks:
Top 5 Lessons Learned Building The Landscape
This exercise has not only affirmed most of my initial assumptions, but has done so at a level I had underestimated.
Everyone is Building an AI Company: A quick look at the surge in ".ai" domain registrations is the easiest way to confirm this. "CompanyName (space) AI" entities are created every day, indicating the widespread creation of AI businesses.
Companies are Mostly Young (Baby?) Startups: OpenAI was founded in 2015, HuggingFace in 2016, and Anthropic in 2021. Even in this era of modern technology, these organizations are remarkably new. Many companies are still in the preparatory stages for their public launch, merely suggesting users to "Join the Waitlist." Contextual AI and Adept AI are just a couple of such examples.
The Market and Technology are Incredibly Complex: While every tech market possesses a certain level of complexity, AI is bringing this complexity to a whole new level. Enterprise sales will need to start with education. Some companies excel in understanding this, with Cohere setting an example through their LLM University.
Partnership Ecosystems are Forming: Despite the existing fierce competition, numerous partnerships are already occurring across the AI space. These partnerships will play a pivotal role for various reasons, including reducing friction to integrate different tech components.
The Space is Moving Extremely Fast: Beyond the hype and despite being in the early stages of the journey, technology is progressing at an impressive pace. For instance, consider MidJourney’s rendering of the same prompt from their v1 to v6. The substantial number of papers published by researchers and collaborations contributes significantly to this swift progress.
The Top 5 Recommendations To Get Started
Here are some considerations to kickstart your GenAI investments:
Pick One Use Case: Begin with a single use case to avoid feeling overwhelmed. Analyze your existing company workflows to identify potential areas of application. While supporting customer service teams (chatbots, voice) is popular, a lower-risk option might be a use case for marketing teams.
Explore Possibility In Your Current Tech Stack: Many organizations invest in cloud data warehouse solutions. These solutions are already offering diverse services and products to support GenAI efforts, across every layer. For this reason, I’ve kept them in a special category, “AI Services Across The Stack”. While I won’t compare them here, exploring the options where you already have investments and store data can be a good first step.
Interoperability Remains Key: As the space has yet to mature, maintaining flexibility and control is crucial. I recommend exploring existing tech investments, but you should also ensure you avoid being locked-in for the future. For instance, consider the ability to use different models for various use cases or different priorities (cost, scale, etc.).
Stay Tuned: This recommendation is closely tied to the fifth learning in the previous section. With the rapid evolution of the space, defining a strategy and executing it in isolation is not enough. Stay vigilant and keep an eye (and ear) out for developments in the AI space. Although it’s exciting to see the wealth of research and recommendations being published, finding the right balance amidst the information overflow is crucial.
Governance, Security, Privacy, Safety, Explainability, Transparency: Surprisingly, these terms haven’t been mentioned until now in the post. Undoubtedly, they are critical topics to address from the very beginning. Define early on how to embark on your GenAI journey, recognizing that you won’t have all the definitive answers. Establish clear boundaries—such as refraining from feeding public LLMs with personal identifiable information (PII). However, many blurry lines will persist, and each country, industry, company, and team will adopt a different stance on these evolving topics. Be prepared for continued evolution in these areas.
This Is Just The Beginning
Envision a world where a market landscape captured might be missing key options within 12 months of its publication! Maybe I should timestamp the work presented today! The AI/GenAI space is undoubtedly one that will continually challenge everyone's ability to keep up.
Stay tuned for more!