Lessons from building an AI-powered ‘star system’ ERP with Everest Systems

If you’re going to be taking on the company that developed the first true enterprise resource planning (ERP) system, SAP, it helps to have two co-founders who between them have 50 years of experience at that company.

Everest Systems is among the most ambitious startups we’ve met throughout our Systems of Action roadmapping process, where we’ve explored the companies taking on one of the stickiest legacy platforms in tech: systems of record (SoR). Several upstarts are looking to break through the moats of the platforms serving as a business’ source of truth by wedging into the workflow with one or a few specific solutions and building up to full feature parity from there, but Everest came out of four years of stealth last November with a full end-to-end, AI-native ERP for software businesses.

The idea that became Everest started in 2020 when Sandeep Chopra, a seasoned product leader who was entrepreneur-in-residence at Sutter Hill Ventures in Palo Alto, joined two senior technology leaders from Germany. Franz Faerber was EVP of Technology at SAP and was the original architect and head of development for SAP HANA and Joachim Fitzer was Chief Development Architect at SAP Business By Design, and they shared the belief that the ERP as we know it is ready for disruption, and that that disruption was unlikely to come from within the four walls of SAP or Oracle.

Today, Everest is using its $140 million in funding toward building a modern, full-featured ERP, growing its early set of customers, and building a brand within software business planning with a dual headquarters in the Bay Area and Germany (a stone’s throw from SAP HQ).

We invited Sandeep to our San Francisco office for another episode of Systems of Action, and used the time to learn what he, Franz, Joachim, and their team of over 100 employees have gathered from developing what Sandeep likens to a “star system” ERP.

Watch the full interview

Insights for founders when building Systems of Action

• Be clear from day one what your ideal customer looks like, and if you will target them with a land-and-expand strategy or end-to-end solution.
• Leverage your platform’s AI-native capabilities to dissolve silos between businesspeople and developers, and even bring in the C-suite when appropriate.
• Put guardrails in place to prevent any AI-powered tools from creating outputs that destabilize your SoR or violate legal requirements.
• Build momentum among your early customers by deliberately acquiring lighthouse logos folks can reference.

Find the right combo of horizontal and vertical

The Everest team knew that AI was going to disrupt ERPs and that they were confident in their abilities to take advantage of the opportunity, but the question then became how they were going to bring their AI-native ERP to market. “You could build it for everything, but that’s a 20-plus year journey. Or build it generic, but that’s not magical enough,” Sandeep said. In other words, the question became finding the right combination of horizontal and vertical.

They spoke to hundreds of potential customers across industries like manufacturing, healthcare, and software to determine what their ideal customer profile (ICP) would look like. Ultimately, they decided to build Everest v1 specifically for software businesses and with their early customers already live and successful, are now expanding to other industries. Sandeep and his team could not only leverage their own software expertise when working with customers, but also given that their initial hurdle would be fighting inertia among ERP customers, the fast-moving, experimental nature of leaders and operators in the software industry made it their best first target.

They also determined that they were going to go for an end-to-end solution that they could bring to other industries as Everest scaled, rather than taking a land and expand approach. As Sandeep put it, they wanted to build a “star system” rather than just a planetary system. He clarified that both have their place and that there isn’t a “right” approach, but that his biggest advice to anyone building a challenger in the wider SoR space is to know very clearly from the outset who your customer is and what strategy you will use to serve them better than any other software vendor over the long term.

Use AI to bridge the gap between business and dev teams

The “magic moment” that most impresses Everest’s customers (and us) is its Live Sandbox feature, which allows users to leverage generative AI to develop and tailor features within the ERP best suited to their business processes. “In a very ERP way, what we’ve given you the ability is to be able to review, simulate, and publish inside that one thing all of the changes that you’ve ever wanted,” Sandeep said, noting that it prevents the punting back and forth of files with “vF” ironically compounding in the title as a team collaborates on a project.

And with a value proposition that separates it from the established ERP players, it bridges the gap between business and developer teams by presenting material in a way that non-technical team members can understand and work with. Of course, you’re not going to get your finance team to start writing code, but as Sandeep put it, “I never thought I’d hear an accountant say, ‘I’m going to merge a change inside the system!’” The result is a smoother, more integrated, and version-controlled ERP system that doesn’t remain esoteric and isolated.

Leverage guardrails to keep unpredictable processes on track

It’s crucial that such a powerful feature has some safety built in, both to avoid creating confusion within the platform or accidentally violating legal requirements. As we put it in our roadmap, it’s important to “wrap deterministic guardrails around probabilistic processes.”

Everest’s Live Sandbox allows users to experiment and collaborate with such guardrails in place, as they try adding attributes, simulating contracts converted, adjusting revenue rules, and closing workflows, including rolling out new AI agents to see the impact in real time before anything goes live.

Get referenceable customers early

To break through the historically deep ERP moat, Everest said referenceability is key. “A lot of people, rightfully so, are going to make decisions based on who else in their peer group has made a similar decision,” A trustworthy peer group varies not just by industry segment, but also company size and complexity, for the true ERP needs change significantly from startup finance operations to a public ready and internationalized enterprise, Sandeep said.

Everest started selling in the mid-market, with software companies that are either just about ready for an ERP or already have had one, and they’re selling directly to CFOs; they are now also selling to companies in inventory and project based industries.

For this approach, he’s getting customers that endure the pain of a full-blown replacement, whether off their pre-ERP suite or their existing ERP, and that’s why the Everest team is focused on getting logos they can then pitch to other CFOs who have scaled teams and taken companies public.

Sandeep has learned a lesson in Everest’s early days that’s applicable to anyone taking on SoRs. “You need the right kind of believer for taking on a big project like this at this stage,” he said. The SoA upstarts, then, won’t be able to easily sway potential customers with the value prop of being AI-native alone—they need to find companies that are willing to make a significant change to their business operations with the belief that the platform will grow alongside their company, and it’s up to the team to convince them.

This series’ name comes from how AI is moving systems of record, the sources of truth for an enterprise’s data, into “Systems of Action.” As we build out our roadmap for emerging opportunities in enterprise resource planning (ERP), customer relationship management (CRM), and human capital management (HCM), we’ll be sharing our conversations with founders daring to challenge deeply-entrenched incumbents.

Subscribe to Atlas for more AI insights and follow us on YouTube to catch up on the series.