Ten companies shaping the Generative Enterprise™ Ecosystem

The Generative Enterprise™ is the emerging articulation of the pursuit of AI technologies based on large language models (such as GPT-4) through which organizations will reap substantial business benefits by generating new ideas, redefining how work gets done, and disrupting business models mired in decades of outdated process and technology. Leaders can deliver their update through the HFS Generative Enterprise™ Ecosystem.

The 2023 HFS Research Generative Enterprise™ Services Horizon report identified and compared the service providers playing their part in delivering the Generative Enterprise™. The report also called out the critical need for those service providers to play a role in ecosystem orchestration as part of the emerging Tech Arbitrage era. They have a crucial role to play in discovering, validating, and connecting the new capabilities the enterprise will need.

Meet the leading players across applications and infrastructure

The HFS Generative Enterprise™ Ecosystem (exhibit 1) maps out leading players across applications (for Consumer use, the Enterprise stack, Industry verticals, and Enterprise apps) and across the new AI infrastructure (in categories we are defining as Deploy and monitor, Train and fine-tune models, Open-source models and frameworks, Full-stack large language models, Store and Compute, and Hardware.

Those names featured in the short summaries that follow and across our ecosystem diagram are just the start. This space continues to explode, so we will continue to highlight the firms the enterprise needs to know about.

Exhibit 1: The HFS Generative Enterprise™ Ecosystem. Success in the AI age relies on how effective your version of this ecosystem becomes.

Source: HFS Research, 2024

Five example stars of the new infrastructure – laying the foundations for change

Open-source models and frameworks: Hugging Face bills itself as more a community than a company – one where the machine learning community comes together to shape new models, work on datasets, and develop applications. It offers an ‘enterprise-ready’ version of its AI platform through subscription. Hugging Face is the place to find everything from transformers and diffusion models to Python libraries.

Full-stack large language models: Cohere’s large language models are built on transformer architecture and trained on supercomputers to provide natural language processing solutions firms can deliver without additional machine learning development. Enterprise focused, Cohere applies Retrieval Augmented Generation (a technique for enhancing the accuracy and reliability of generative AI models with facts fetched from external sources) to answer questions and complete tasks using enterprise data as the source of truth. Cohere has just announced a new scalable enterprise language model – Command R.

Train and fine-tune models: Weights & Biases is best known for its support in evaluating model performance and iterating data sets – toward greater accuracy. However, it also offers model lifecycle management, parameter optimization, and experiment tracking. Customers include the likes of Meta and OpenAI.

Store and compute (Label and process data): Snorkel solves the problem of manually labeling and curating training data, helping customers deliver quality data as much as 100 times faster. Snorkel was developed at the Stanford AI Lab. Google is a long-term partner, and IBM and Microsoft are both clients.

Hardware: NVIDIA is the hot hardware ticket right now, thanks to its recently announced Blackwell platform – a response to increasing compute demand from GenAI, alongside Groot – a foundation model for humanoid robots. The firm also offers Jetson, a platform with scalable software, a modern AI stack, and application-specific workflows for ‘embedded’ computing. Microservices allow developers to access GenAI capabilities at the edge through API calls.

Five red hot example applications helping firms get work done

Consumer users: ChatGPT ripped up the rule book when it landed out of the blue on November 3, 2022. Based on a large language model, it became many consumer users’ first contact with generative AI (GenAI). ChatGPT is one of OpenAI’s products. Given the right prompts, it can help compose emails, reports and code, and summarize uploaded information into more consumable edits.

Enterprise stack: Jasper is rapidly becoming the go-to AI co-pilot for enterprise marketing teams. It recently expanded its multimodal capabilities by acquiring the AI image platform Clipdrop. It supports AI-assisted content with analytics and insight.

Enterprise stack: Synthesia.io is a video generator that supports making video via AI without the need for mics, cameras, lights, or real-world action. The primary use case is for training videos – replacing death by PowerPoint and e-books with engaging avatars. Synthesia boasts 50,000 customers, including Microsoft, Johnson & Johnson, and Ocado.

Industry vertical (Health): Pathai aims to advance the phases of drug and diagnostic development for patients through its AI-powered pathology technology and collaborations with biopharma, labs, and clinicians. It serves oncology, inflammatory Bowel Disease, and a form of liver disease known as NASH.

Industry verticals (Law): Harvey comes with funding from OpenAI’s start-up fund. It is a proprietary AI service targeting increased accuracy, security, and privacy for law firms by focusing on legal knowledge rather than the general knowledge focus of open LLMs. PwC partnered with them a year ago.

The Bottom Line: Get familiar with tomorrow’s stars today

Many of the names on the HFS Generative Enterprise™ ecosystem will be unfamiliar to many. But all are making an impact and have significant customers and partnerships. Some will go, and others will come to take their places – as with all great technological revolutions. Infrastructure plays a huge role in enabling the changes that are coming. The applications will deliver the change. You can’t win without the complete ecosystem.