Why the Midmarket Needs Forward Deployed Engineering

What Is Forward Deployed Engineering?

Forward deployed engineering is the practice of placing a senior engineer or a team of engineers from a technology provider inside a client’s environment. Working alongside the client’s team, these engineers are responsible for moving the technology into production and for keeping it running.

This model was pioneered by Palantir, which learned early that powerful software doesn’t create value on its own. Someone has to sit with internal teams, understand the messy reality of their data and systems, and keep building (and fine-tuning) until everything works. In the last few years, leading AI companies have adopted this approach for their largest customers: They are providing engineers to help ensure that AI projects keep moving forward.

Three qualities define forward deployed engineers:

• They work inside, not outside. Forward deployed engineers build in their client’s environment, with their client’s data, applications, and security controls. They don’t simply present a slide deck or work from a remote lab.
  
  
• They own the outcome. The job is not done until the system is running in production and people are using it.
  
  
• They close the loop. What engineers learn in the field shapes what an organization builds next, so every deployment makes the next one better.

Comparing with Consulting and Staff Augmentation

Forward deployed engineering might sound similar to traditional consulting or efforts to augment staff with contractors. The primary difference is accountability. With forward deployed engineering, the engineers are the ones accountable for delivering results.

Consider a few additional differences:

Consultants can help organizations decide what to build, and contractors can provide the added capacity that organizations need. But neither consultants nor contractors are fully accountable for the projects they contribute to. They aren’t tasked with making an AI system work—or keeping it working—in a real environment.

Why Have Only Enterprises Used This Model?

Enterprises have proven that forward deployed engineering works. Midmarket companies haven’t employed it primarily because no one has offered it to them.

The firms that offer forward deployed engineering have focused on the biggest accounts, which can meet engagement minimums built for enterprise budgets. Similarly, the AI platform vendors who use this approach reserve their deployment teams for the largest enterprises. Regional consultancies can offer the right pricing for midmarket companies, but these consultancies often choose to advise rather than build.

The 300-person midmarket company with an AI mandate has just three unappealing options: hire an AI team it can't easily afford or retain; pay enterprise prices; or keep launching projects and hope something sticks.

Meeting a Critical Need in the Midmarket

The irony is that midmarket companies need forward deployed engineering even more than enterprises. Enterprises can, if necessary, build in-house teams to move projects from pilot to production. But midmarket companies have much more limited internal resources.

Moving AI into production requires five best practices that midmarket companies are often unable to execute on their own:

1. Establish governance and security. Organizations must control who uses the AI system, what data goes into the system, and what the system costs.
2. Develop a roadmap tied to value. Decision-makers should rank use cases by return on investment rather than relying solely on recommendations from AI vendors.
3. Integrate engineering. Any new AI capabilities have to be integrated with real systems, use live data, and go through a thorough security review.
4. Draw from AI expertise. Midmarket companies launching their first few AI initiatives are unlikely to have team members with deep AI expertise. They need people who have shipped production AI before, because those are the ones who know how to spot and overcome potential roadblocks.
5. Assign ownership for after the launch. Getting a new AI tool or system up and running in production is a key milestone. But organizations must also be sure they have team members accountable for monitoring, tuning, and improving AI going forward.
6. Will your engineers work inside our environment, or advise from outside it?
7. Who owns the outcome: you or us?
8. What happens the day after launch? Who monitors, tunes, and fixes the AI system?
9. How do you handle governance and security before we scale?

Midmarket companies might be able to implement a few of these practices in-house. Adopting forward deployed engineers, along with a partner that can govern AI and run it afterward, can fill in the gaps without requiring companies to build full-fledged AI departments.

Envisioning Forward Deployed Engineering in Practice

Using forward deployed engineering can help a wide range of companies that have tried and failed with AI. Consider a family-owned produce distributor. In the past, team members might have collected field sales data from handwritten forms, manually typing information into spreadsheets. Maybe the company tried AI extraction but was unable to produce usable, repeatable results. And they couldn’t generate an audit trail to support FDA traceability.

Instead of running another pilot, the company could work with forward deployed engineers. Those engineers could build a production system using a cloud-based AI tool with human-in-the-loop validation. The engineers would be fully integrated into daily operations, ensuring that everything worked as it should.

The engineers could significantly enhance the accuracy of data input, accelerate task completion, and implement full traceability. With this approach, the company could achieve a return on investment in about six months.

Four Questions to Ask Any Potential Partner

If your organization is considering forward deployed engineering, use these questions to find the right partner:

If the answers are vague, you might end up with another pilot.

From Pilot to Production and Beyond

NexusTek brings forward deployed engineering to the midmarket through our Advise ->Build -> Run model. We start by putting control around the AI you already have and helping you decide where AI can deliver the biggest returns. Our engineers then build AI into your environment and own the result. Once your implementation is live, NexusTek managed services keeps it running and improving.

Ready to move forward with AI? Book an AI readiness conversation with a NexusTek expert.