Nearly three years ago, I wrote a column based on a thought-provoking question I was asked by a founder at an entrepreneurship event. Are there metrics for deciding when to fund a Series A round?
Looking back, it seems like the answers I offered then are still valid. “Does the founder have a sense of mission?” Will this idea become a great product with great demand? Will the company be able to grow compelling metrics at scale?
But today, as AI takes center stage, I believe more pressing questions are emerging for both entrepreneurs and investors. What are the key considerations when funding an early-stage enterprise software company in the AI era?
This issue is becoming increasingly important as the venture capital community increasingly seeks promising ideas earlier, often before a single line of code is written and long before any revenue is generated. . Our recently launched bespoke company-building program, Greylock Edge, has seen an exceptional response from ambitious engineers and entrepreneurs at the seed, pre-seed and even pre-idea stages.
The potential of enterprise AI adds a welcome dimension to these conversations. In the past, many software founders were afraid to seek funding “prematurely.” They believed that without a beta product or initial revenue, they would not be ready to seek a partnership with an experienced venture investor. The AI gold rush changed that.
Now, when seed-level founders meet with us, they test ideas, brainstorm use cases, engage with customers with our help, and envision a product platform from the ground up. In discussions with them, we recommend focusing on his three key questions:
- What is the highest value customer use case you can solve?
- Is there proprietary data that can be accessed or developed to create a moat?
- Can I list your product alongside an existing vendor?
Develop the most valuable use cases
Unlike earlier eras of innovation, the largest software vendors are responding quickly and launching significant initiatives in AI. In an increasingly crowded field, the next generation of his AI founders must ask themselves: What fundamental use cases can you achieve that incumbents don't already cover?
It may be daunting to build a product that Microsoft or ServiceNow already has an installed base for, but large enterprises are finding high-value use cases that weren't possible before, or existing products that can be better solved. We still welcome new startup products that enable high-value use cases. Much faster or cheaper. Slack did this for his Outlook email. Palo Alto Networks entered the market at a time when Cisco and Check Point Software had the leading firewall software. MongoDB has entered a market historically dominated by Microsoft and Oracle.
A great example today is how startup Cresta saw an opportunity to improve sales performance without replacing existing sales infrastructure or salespeople. Cresta's management platform uses AI as a coach to assist sales teams and call centers. The company's software provides real-time behavioral coaching to improve soft skills, generates suggestions for answers to product questions, and provides insights into performance and customer trends. Performance metrics for sales professionals like CarMax, Cox, and Intuit clearly improve when working with Cresta AI coaches. Cresta's growing business provides a case study of how new AI startups can quickly gain enterprise acceptance by driving superior results in high-value use cases.
Build a data mote
There is a broad consensus emerging that access to unique data pools can give AI software businesses a competitive advantage. Snorkel founder Alex Ratner recently claimed: forbes, For some companies, there is little benefit to using an off-the-shelf LLM if competitors are using the same tools.
He's right. Most large enterprises leave data pools containing unique and valuable data sitting untapped in meaningful ways. This is true across industries such as healthcare, financial services, consumer goods, retail, and manufacturing. These companies are looking for ways to create value from their unique data beyond simple reporting and basic analysis. They can capture unstructured data and establish connections to automate business-related tasks, extract insights, build accurate models, and accurately predict future outcomes that lead to business benefits. We want to train an AI to: The specialized LLMs and other AI models that emerge from this effort form a moat against competitors.
Seed and pre-seed companies are unlikely to have proprietary data at the establishment stage. But pursuing unique data needs to be part of your strategy. Initially, large organizations may be building innovative tools that allow them to collect proprietary datasets that are not available to the public.
A seed or pre-seed company's long-term vision may include finding ways to retain rights to the data used to train models. You may also be planning to develop insights and metadata that will be useful to your first customers and applicable to your broader industry.
Unusual security is an example of how a data moat can be developed gradually by a startup. Although the company did not start with its own dataset, its software was designed to ingest a variety of existing signals and telemetry. This allowed us to develop our own database that we leveraged to build a baseline across users, cloud email, and collaboration applications. Now, the company combines this data with advanced AI to accurately detect anomalous behavior and automate remediation for more than 1,400 companies.
Start with an insertion strategy
If you talk to early AI product builders, they are realistic about the fact that very few large companies will immediately “rip and replace” their existing platform, especially for new AI vendors with incomplete and unproven products. is. With this in mind, we urge early-stage entrepreneurs to develop a clear implementation strategy that allows newly developed software to easily connect to or be used alongside existing software. It is recommended.
The importance of getting your insertion strategy right cannot be overstated. Early-stage AI products must be purposefully designed to deliver significant advances within business-relevant customer use cases. The key is that you need to be able to quickly demonstrate new capabilities in a low-friction way without introducing new risks to CTOs, technology teams, and enterprise forward-thinking champions.
Once a startup is able to establish a beachhead and provide growing value, it can gain legitimacy to expand its capabilities. If done correctly, this could lead to a growing platform that could eventually replace existing platforms. AI seed companies that build large franchises over time are ultimately companies that develop platform strategies (not just narrow niche feature modules).
Rubrik, which recently celebrated its 10th anniversary, started with an ambitious vision and the realization that it needed to be very careful about how it competed with established incumbents. The company started by offering a narrow range of backup and recovery products specifically for VMware. This was their insertion strategy. The reputation and trust they have built has enabled the company to scale to growing enterprise workloads across hybrid clouds. The company's growing platform, powered by applied AI, has evolved into a market-leading solution for ransomware and enterprise data resiliency. Currently, the company serves over 5,500 of his customers, with ARR in the low hundreds of millions of dollars.
The best mindset for AI startup entrepreneurs
As we head into 2024, we will see an explosion of new enterprise AI startups across infrastructure and foundational models that enable intelligent applications. We are still in the early stages of being able to imagine and predict what will happen. What is certain is that the demand for enterprise AI solutions is real and has the potential to disrupt and ultimately transform enterprise information technology as a whole. Startup entrepreneurs with the right mindset and a willingness to ask the right questions will be best positioned to win in a rapidly evolving market.
(Disclosure: Greylock is an investor in Abnormal Security, Cresta, Rubrik, and Snorkel, and was an investor in Palo Alto Networks.)
follow me twitter Or LinkedIn. check out my website.