Is Generative AI going to be transformative or incremental for your startup? Are you re-vampimg your entire product roadmap, or sprinkling AI in along the way? A panel of AI-driven organizations took the stage at SaaStr Annual to talk about the trends and opportunities of GenAI and how it’s shaped their companies.
In this session you’ll hear from:
- Ravi Rajamani, Managing Director and Global Head of GenAI at Google
- Lilian Breidenbach, Co-founder and Managing Director of Legal OS
- Harpinder Singh, Partner at Innovation Endeavors
- Sarah Al-Hussaini, Co-founder and COO at Ultimate
As a startup, one of the more complicated things you can do is integrate AI into your product roadmap because of the speed of evolution of the tech itself.
Keeping up with the Speed of Generative AI
At Legal OS, they’re building an AI-powered legal assistant for revenue-driving functions. It’s been a very radical process for them. Almost overnight, with the emergence of LLMs, a previously unsolvable problem became solvable. At the end of last year, they scrapped their entire roadmap and almost 90% of their code base. They were building a product almost from scratch thanks to the introduction of GenAI.
If that sounds like a painful experience, it was. It was also the right way to go. Should you scrap your entire roadmap and pivot? That depends on whether GenAI will be transformative or incremental for your product.
If it’s incremental, there’s low risk. You can test many open-source technologies, and you probably don’t care that GPT4 is expensive and slow because your usage will be low. If it’s transformational, like for Legal OS and Ultimate, it becomes an existential discussion. Ask yourself:
- Do you want control over the technology?
- Do you want to build your own LLM and build it in-house?
- Is time-to-market absolutely critical for you?
If so, you will be wrestling with something open-source while making a plan of action to make it long-term. It’s very easy to build a cool demo, but to get from that 80% to 100% that’s honest, in scope, and controllable requires a Herculean effort.
Key pieces of advice for founders:
- Assess the risk profile of your customer base. Will they be ok with a better product that can go off-piste while you work out the kinks, or does it need to be production-ready from day one?
- Think very ambitiously about how much resource you dedicate to this. If the technology allows you to solve a customer pain point much easier and better, you need to go all in if you have confidence in your team and are early enough in your organization to do it.
- You can iterate much quicker if you allow people within the entire organization to prototype, not just the engineering team. The beauty of startups is you’re agile, and with LLMs, it’s language-driven.
A VC Perspective — How Should Startups Approach GenAI
Generative AI is moving extremely fast, yet the basic things around having the right team, the right market, and a solution that’s 10x better are equally as relevant today as ever. So, as a startup, you need to think deeply about the customer experience and what GenAI can do for you.
Many startups build a thin layer on top of OpenAI or whatever stack. You need to think about the durability and defensibility of your value prop, especially as these GenAI models get better. As a founder, you can’t differentiate yourself if you only sprinkle AI into the product.
Deep customer empathy and real transformational use cases are where the magic lies. The beauty and the curse of this market is that everyone has access to the same toolkit. How much can you incorporate customer feedback and improve your models?
Speed and a nimble team that can out-execute everyone will win.
The Challenges and Opportunities of GenAI
At Ultimate, customer service is the number one industry disrupted by AI, especially GenAI, making the experience more human. Currently, the industry is rudimentary, but as it evolves, it will make it incredibly easy for support teams to deploy bots that work across channels, languages, and systems and solve 80% of customer requests in a human-like way.
Over the next 12-18 months, Ultimate will make chatbots launch faster, with less effort to manage, and they’ll go live without all the extra work. The experience will be much more human-like, so much so that people will struggle to know whether they’re communicating with a human or a machine.
For Legal OS, the people who buy and set up their product are legal compliance teams. If you give them an AI product, that’s a black box. They’re not going to use it unless it’s accurate and secure. From the beginning, you have to focus on making this AI steerable and auditable. You might also have to sacrifice the speed of response to ensure accuracy.
If you can build a steerable, auditable, and very secure AI, then these risk-averse teams can develop a tremendous amount of trust in the product and deploy it quickly and widely.
The only advantage startups had in the past was moving fast. They could take risks that Google couldn’t. Now, the toolkit is democratized, so anyone can build a prototype and put something up within 30-90 days. Companies of all sizes claim everything, so it isn’t easy to differentiate yourself.
The takeaway:
It’s easy to claim you have all this stuff, and as a startup, you have to be very savvy on GTM. It’s exciting to be working at the cutting edge of technology and to be the one defining patterns. No one knows what it’ll do to the industry, so speed is critical. In addition to speed, you want to play the long game. Looking at previous waves like Cloud and data, we’re in the 15th year of Cloud, and it’s still not done. Choose good long-term partners, speed, and a GTM for Acts II and III.
The post How to Solve Unsolvable Problems with Generative AI for Startups with Google Cloud appeared first on SaaStr.
via https://www.aiupnow.com
Amelia Ibarra, Khareem Sudlow
