Beyond chatbots, businesses can use large language models (LLMs) as a natural-language interface to work that used to need specialist skills: condensing large volumes of information into clear summaries, and operating complex software, from media editing to professional tools, by describing what you want. Four early investments of the OpenAI Startup Fund, Descript, Harvey, Mem and Speak, show the pattern, and fields such as design, accounting and marketing are next.
Key takeaways
- Natural language is becoming the most natural way for people to communicate with technology, which widens the market and simplifies customer experiences.
- The real value of ChatGPT-style models lies less in chat and more in distilling vast information into concise, understandable summaries.
- LLMs remove the need for extensive training in complex skills and software, so tasks like image editing open up to far more people.
- The window before the market saturates is limited, so companies that move early, with IP behind them, stand to gain most.
Why is natural language becoming the interface to technology?
The move from mouse and keyboard to touchscreens changed how we interact with technology. Sam Altman, the CEO of OpenAI, believes the most natural way for humans to communicate with technology is through natural language. This approach not only bridges the gap between humans and machines but also widens the market and simplifies customer experiences.
Each interface shift removed a layer of learning. A command line required memorised syntax, a graphical interface required knowing where things were, and a touchscreen required only pointing. Natural language removes the next layer: you no longer need to know how a tool works, only what you want from it.
ChatGPT, built on OpenAI's GPT-3.5 series of models, may not have represented a radical leap in AI in terms of size or architecture, but its public release democratised access to advanced AI. It made techniques such as one-shot learning and prompting accessible to everyone, often with results that exceed expectations.
What did the OpenAI Startup Fund's investments signal?
OpenAI has broadened its focus beyond AI development and advocacy. Through the OpenAI Startup Fund it stepped into venture capital, investing in innovative companies and starting a new chapter. As of July 2023, the fund had made significant investments in four DeepTech startups: Descript, Harvey, Mem and Speak. These companies stand out for using LLMs to change how people and machines interact.
Each applies the same idea to a different field. Descript lets people edit audio and video by editing a text transcript. Harvey builds AI for legal work. Mem applies AI to notes and personal knowledge, and Speak uses it for language learning through spoken practice. None of them is a general chatbot; each puts a language model behind a specific professional task.
Choosing these four startups from thousands worldwide points to a deeper strategy. Beyond the popular view of ChatGPT as a tool for generating marketing content or acting as a chatbot, its real value lies in distilling vast information into concise, understandable summaries. That capability broadens its appeal to a much wider range of users.
What can LLMs do for a business beyond chatbots?
Most useful applications fall into a few patterns, each of which replaces a skill or a slow manual step:
| Pattern | What it does | Typical business use |
|---|---|---|
| Summarise and synthesise | Condenses long or numerous documents into the points that matter | Research digests, meeting notes, customer feedback reviews |
| Extract and classify | Turns unstructured text into structured fields and categories | Tagging support tickets, reading contracts, sorting social posts by topic |
| Operate software in plain language | Translates a request into the steps a tool needs | Editing media, querying data, building reports |
| Draft and transform | Writes first drafts and adapts text for new formats or languages | Proposals, product copy, localisation |
All four work best with a person reviewing the output, because language models can state wrong things confidently. Our guide to checking AI copy before it ships covers how to build that review into a workflow.
Why does the market opportunity matter to investors?
These advances have profound effects on the venture capital (VC) market. The expected growth of the total addressable market (TAM) suggests significant opportunities ahead. Technical barriers used to limit who could put these technologies to use. Now tasks once considered complex, like editing an image in Photoshop, can be simplified with AI and made accessible to a far wider audience.
This progress opens up large growth opportunities, as individuals and businesses alike explore what becomes possible. In a landscape changing this quickly, time is of the essence. There is a limited window before the market becomes saturated and the novelty of LLM-based technologies fades.
Why does timing matter for companies adopting LLMs?
To capitalise on the opportunity, companies need to act quickly and use LLMs to establish a strong presence in the market before it saturates. Recognising the importance of time is key to competing in this space and securing a leadership position.
What should business leaders do next?
Business leaders should look beyond chatbots and explore the potential of LLMs in industries such as design, accounting and marketing. These models remove the need for extensive training in complex skills and software, allowing technology to fit more naturally into professional workflows.
The market offers many opportunities for rapid expansion in a short time. Early adopters with intellectual property behind them, who can quickly reach new customer bases across industries, stand to gain significantly. These companies are likely to become prime targets for venture capital and private equity investment in the coming years.
A sensible first step is small: pick one workflow that involves a lot of reading or writing, put an LLM behind it with a person checking the output, and measure the time it saves. In marketing, that might mean connecting a model to your own data, as in our guide to connecting Claude to marketing data with MCP.
Related reading
- How can LLM-backed startups stay defensible enough to win VC investment?
- How to connect Claude to marketing data with MCP: a guide for agencies
- AI hallucinations in marketing: how to check AI copy before it ships
- SOMIN research: peer-reviewed papers and talks
- SOMIN's AI marketing agents
FAQ
What can large language models do for businesses beyond chatbots?
Their main value is distilling vast information into concise, understandable summaries and letting people use complex software through natural language. That removes the need for extensive training in specialist skills and tools.
Which startups had the OpenAI Startup Fund invested in?
As of July 2023, its significant investments included four DeepTech startups: Descript, Harvey, Mem and Speak. Each uses LLMs to change how people interact with machines.
Why was ChatGPT significant if it was not a radical leap in AI?
Its public release democratised access to advanced AI. Techniques such as one-shot learning and prompting became available to everyone, often with results that exceed expectations.
Which industries can benefit from LLMs?
Design, accounting and marketing are clear examples, along with any field where complex software or specialist skills have been a barrier. Early adopters with IP behind them stand to gain most.
