Harnessing AI: Evolving Business Models and the New World of Work

Investing in AI-powered solutions now is critical as they drive innovation, boost efficiency, and provide a competitive edge across industries. 

‍

Why is this happening now?

1. Computational capabilities have increased, both with the advancement of GPUs and the massive quantities being produced. Hyperscalers like Amazon, Google, and Microsoft are making substantial investments in AI infrastructure, driven by the soaring demand for capacity to support advanced AI models. 
   
   ‍
2. Managing unstructured data has forced database technology to make exponential strides (more on this later). In the 2000s, as the volume and variety of data grew rapidly, “big data” emerged, leading to the development of new technologies for processing and analyzing large datasets. An array of new wearable devices featuring sensors, cameras, and speakers is set to launch soon, marking a significant advancement in our capability to capture real-time, contextual data.
   
   ‍
3. We have been witnessing continuous breakthroughs in AI technology, which seem to be occurring on a weekly basis. AI models now exhibit near-human levels of comprehension, understanding context, sarcasm, and nuanced human emotions in text. Recent developments in multimodal AI models highlight the ability of these models to perform a wide range of tasks across different domains, including visual and linguistic tasks. These advancements are paving the way for more flexible and adaptive AI systems capable of seamlessly integrating multiple types of data. OpenAI most recently introduced GPT-4o, an enhanced version of GPT-4, which is significantly faster and more capable across text, voice, and vision tasks. GPT-4o excels in multilingual support, reducing token usage dramatically in various languages.
   
   ‍
4. Businesses are increasingly recognizing the value of AI in driving productivity and fostering innovation, leading to initial adoption of AI technologies across various industries. Software coding assistants lead the way in initial productivity gains, with massive improvement in developer code writing velocity and quick ROI. Interactive chatbots are enhancing personalization and real-time customer interactions. In analytics, generative AI allows natural language queries for enterprise data. Industries such as healthcare, legal, and media, which generate large volumes of unstructured data, present significant opportunities for generative AI to streamline and automate the processing and summarization of information.

‍

What is the impact on existing business models?

Initially, capital flows favored chipmakers and major cloud service providers as conduits of innovation. As the ecosystem evolves, the application layer is set to become a key beneficiary. Following the release of ChatGPT and the surge in available LLM resources, many Enterprise SaaS vendors quickly integrated AI-augmented features like chatbots, copilots, and content generation into their solutions, leading to the rapid commoditization of these capabilities. 

‍

Today, incumbents showcase a business model advantage by positioning themselves in the workflow layer, capitalizing on their installed base, and subsidizing AI by bundling it with existing product lines. However, platform SaaS companies that have built unique data assets over time are best positioned to deliver differentiated outcomes. The ex-CEO of Snowflake, Frank Slootman, put it well: “There is no AI strategy without a data strategy.” Additionally, domain experts who understand their workflows have a unique advantage and understanding pain points of end users. The Casetext team (now within Thomson Reuters) intimately understands lawyers and built a data asset over time, not just on legal cases, but on how lawyers researched and drafted briefs. This positioned the company to become an early entrant in the generative AI space and resulted in an incredibly rapid market adoption of the company’s AI legal assistant CoCounsel in early 2023. 

‍

Proactively recognizing and integrating AI advancements can elevate a company, allowing it to optimize operations, improve customer experiences, and develop innovative applications. 

‍

What are some examples of AI applications designed for compelling enterprise use cases today?

‍

Agentic Workflows: AI agents are designed to observe their environment, process information, and take actions based on that information to achieve specific goals. These agents can communicate with each other, share data, and optimize workflows using real-time information and learned experiences. Such workflows involve AI agents continuously gathering data to understand the context and relevant information needed for their tasks, analyzing this data, and then evaluating different options to select the best course of action. After executing these actions, the AI agent evaluates the outcomes and learns from them. This feedback loop enables the agent to improve its performance over time by adjusting its algorithms and decision-making processes.

‍

Search & Information Retrieval: As AI models advance in generating human-like outputs across text, images, and audio, they heavily depend on vector data representations, or embeddings, to comprehend and produce contextual meaning. Certain generative models require databases specifically engineered to handle large vector data sets and enable the instant retrieval of vectors that are semantically similar. Vector databases use vectors to represent data points, enabling faster and more relevant data retrieval based on similarity. Additionally, developers can enhance the capabilities of LLMs by also integrating knowledge graphs, which allow developers to incorporate structured relationships and entities. Last year, Squirro developed a Retrieval Augmented Generation (RAG) offering, in which Squirro’s semantic search technology queries an Enterprise’s documents and emails for relevant information. Then, the appropriate information is fed into large language models that Squirro integrates with to generate more data and context-driven responses.

‍

AI-Enabled Analytics: AI can play a role in enhancing data analysis, including automated report generation, anomaly detection, interpreting and generating visualizations, and predictive analytics. Fiix was a company that delivered workflow software for scheduling, organizing and tracking of equipment maintenance, and over time created data assets focused on machine-level data. Fiix seamlessly connected the shop floor’s IoT solutions, parts suppliers, contractors, and corporate IT systems to improve the way physical assets and people interact to drive better business outcomes.

‍

Design & Content: AI is revolutionizing design processes by automating the creation of complex designs and content. VidMob uses AI to understand the intricate elements of a creative video ad (color, text, objects, emotions in ads), and then closes the loop by understanding ad performance (click-through rates, etc.). With this data set, it can score ads based on predicted performance, and the algorithm improves with more data. VidMob now helps brands better understand diversity shown within an ad as well as if messaging is emotional or functional.

‍

Logistics & Shipping: Logistics has historically relied on manual data entry and numerous Excel spreadsheets. AI will ultimately optimize supply chains and enable quick and intelligent decision-making. Time savings is the most valuable outcome for companies looking to optimize their supply chain operations. Flowspace’s AI-powered solution functions as a supply chain analyst, helping customers apply real-time automations, use natural language to monitor inventory and order deliveries, take actions on those orders, and streamline processes involved with freight management, from booking date confirmations to documentation.

‍

What key considerations should organizations keep top of mind as they prepare for the new world?

‍

Pricing: In a world where organizations are reducing staff due to optimizations delivered by AI solutions, seat-based pricing may come under significant pressure. AI is likely to compel organizations to adopt usage-based, transaction-based, value-based pricing models, or simply just increase prices. 

‍

Team Talent: When it comes to building and deploying AI solutions at scale, highly specialized expertise is of course required to implement, manage, and scale the necessary computing infrastructure. In addition to this, having experts in the specific use cases for which AI models are being trained is absolutely critical, whether in manufacturing, customer service, or law (as our co-founder, Jim Curry, mentions in our three-part series: AI as the Next Infrastructure Wave).

‍

User Interfaces: Advancements in natural language processing are set to revolutionize traditional user interfaces (UIs) by making them more intuitive and personalized. Designers will need to focus on creating UIs that seamlessly integrate with AI capabilities, ensuring a cohesive and intuitive user experience.

‍

Leveraging Multiple Models: Organizations have shifted from testing a single model to experimenting with multiple AI models. This approach gives them the flexibility to customize for specific use cases considering performance, size, and cost, avoid vendor lock-in, and rapidly leverage innovations occurring on a weekly basis. The industry is also witnessing a shift in open-source usage as model performance begins to converge. Organizations can also leverage the capabilities of large models to manage and coordinate fleets of smaller models, directing them to handle specific tasks effectively.

‍

Data Advantage: Organizations should consider the various data sources available as they start to scale (customer generated, publicly available, synthetic, data derived from feedback, etc.) and ultimately align their data strategy with their AI strategy. Leveraging multi-modal approaches that utilize the distinct strengths of various data sources is likely to provide more sustainable advantages. We are living in a world where multi-modal capabilities enable AI models to deliver highly personalized and contextually relevant solutions by combining various data types. 

As Michael Bloomberg said, “In God we trust. Everyone else bring data.”

‍

‍

*BuildGroup invested in and exited Casetext and Fiix and currently invests in Vidmob, Flowspace, and Squirro.

The information in this blog post is provided in good faith without any warranty. It does not constitute investment advice, recommendation, or an offer of any services or products of BuildGroup Management and it is not intended to provide a sufficient basis on which to make an investment decision. This document is provided for educational purposes only. Discussions of current or former BuildGroup portfolio companies are intended for educational and discussion purposes only. Any portfolio company so discussed has been selected based on objective, non-performance based criteria.

This content does not constitute or form part of an offer of any investment advisory services of BuildGroup Management, LLC, nor does it constitute or form part of an offer to issue or sell, or of a solicitation of an offer to subscribe or buy, any securities or other financial instruments, nor does it constitute a financial promotion, investment advice or an inducement or incitement to participate in any product, offering or investment.

‍