5min read

The Missing Link for Enterprise AI Adoption

The Missing Link for Enterprise AI Adoption

Authors

Kaushik Subramanian

Tom Mendoza

Pierrick Cudonnec

Conrad Schoo

Nitya Hinduja

We all know the same statistics: every enterprise has understood that AI can drive business outcomes. Yet adoption does not mean ROI, and the reasons AI implementations fail vary. Some employees feel uncomfortable sharing how they work with a system that may replace them over time. Others do not know how the other functions of their enterprise run. Both paths often result in the same situation: expensive, shiny agents producing output that is incorrect or irrelevant. 


Source - Edra

It’s almost ironic that we have unlocked unprecedented intelligence, but are limited by the understanding of our own organizations. The uniform response to this problem has been a magic wand known as FDEs. They are the connecting tissue that enters enterprises, maps out processes, and ensures that new software, often custom-built for the specific use case, finally works.

What if your software could deploy itself?

Enterprise software companies are hiring forward-deployed team members at scale, and the supply of FDEs is thin. As if the compute load of agentic workflows were not enough to crush margins, deployment costs push gross margin down further.

To see whether you can avoid the FDE drag on your margins, start with what FDEs actually do. In essence, they learn the intricacies of how a business runs and build a software solution around it. 

Every business is different, but the process of mapping out an organization can be encoded and accelerated. By reading an organization’s logs, tickets and messages, you can get to an understanding of a business within a week, without interviews or workshops. These learnings can be stored in a context layer, which can then power all agentic and human workflows across the organization at a speed and cost that manual mapping can’t match. It’s not about understanding a given workflow and developing a bespoke solution for it, but about the meta-learning of understanding any workflow based on the same procedure. 

This insight led our investment team to search for the best team globally to tackle this problem. Together with EQT’s software procurement team, we evaluated all leading companies that turn organizational knowledge into reliable, scalable instructions for AI agents. One team was head and shoulders above the rest.


Edra & ContextOps as the answer

Eugen & Yannis have worked on this problem since long before it became en vogue. They co-led Forward Deployed AI Engineering at Palantir, where they spent years inside large enterprises moving AI solutions out of demos and into production.

These learnings shaped their approach in a space they coin ContextOps: turning operational evidence and expert judgment into tested, approved instructions, and maintaining those instructions as the business changes.


Source - Edra

They track executable coverage, the share of cases an Edra-powered agent handles correctly using approved instructions without any human steering. When coverage drops, Edra proposes updates to the instructions, and the people who own the process review and approve them until coverage recovers.

As a result, enterprises own their scattered knowledge and no longer depend on engineers to translate and maintain it. Edra offers one centralized interface to create, review and deploy the instructions behind every agent.

Soon, there will be 2 types of companies: those that run their AI agents on a self-healing and compounding context loop that captures the essence of each workflow; and those that will struggle with AI workflow efficiency, delays, and implementation costs. 

EQT as investor, customer, and partner

EQT Group deployed Edra on its internal IT service desk, where the team handles 3k+ tickets a month. Already in week 1, Edra identified 130+ previously undocumented resolution patterns. By week 2, it proposed resolutions on live tickets, and by week 4 it almost ran on autopilot. The strong results quickly excited our wider EQT portfolio of 330+ owned companies to evaluate Edra, and by now several EQT portfolio companies, such as Keywords Studios, are running pilots with Edra. 

Our conviction in Eugen & Yannis drove our investment, but the results they are achieving position them for much more. We believe Edra is the missing link in enterprise AI adoption, from app-layer AI companies struggling with deployment complexity, to established SaaS companies looking to strengthen their core offering.

This is a marketing communication. Please refer to the information document of EQT Ventures before making any final investment decisions.

More to read

20 articles