There's a quote from a recent Andreessen Horowitz article that perfectly captures what we're seeing in the AI space:

”Enterprises buying AI are like your grandma getting an iPhone: they want to use it, but they need you to set it up." — Andreessen Horowitz

This hits the nail on the head. And it's exactly why at Barnacle Labs, we're adopting the Forward Deployed Engineer (FDE) model that companies like Palantir pioneered and OpenAI has now embraced. Let me explain why this approach has become essential for delivering real AI impact, not just AI demos.

The Problem: Everyone's Missing the Forest for the Trees

Here's what we keep seeing: companies get excited about AI, hire a consultant who shows them how to automate their expense reports, and then wonder why they're not seeing the transformative impact they'd hoped for. Don't get me wrong—automated expense reports are nice. But they're not going to revolutionise your business.

The issue is that most people who aren't engineers simply don't understand what's possible with AI. More importantly, they don't understand what's not possible. This knowledge gap leads to one of two problems: either they aim too low and stick to the obvious use cases, or they aim way too high and expect magic that current technology can't deliver.

Why Technical Understanding Matters More Than Ever

AI projects require a unique blend of creativity and deep technical understanding to break through those obvious-but-mundane automation use cases. The profound possibilities, the ones that can actually transform how a business operates, require someone who understands both the technology's capabilities and the business's workflows at a granular level.

In traditional technology projects, everyone knows the playing field. No consultant needs a primer on what a database can do, or how a CRM system works, or the limitations of a standard web application. These technologies have been around for decades and their capabilities and constraints are well-established and widely understood. In that world, technical expertise matters most during implementation, not during the initial scoping phase. You can hand a business analyst a requirements document and trust they'll understand what's realistic to ask for.

But AI is different. The technology landscape is vast, unfamiliar to most people, and evolving at high speed. Techniques that were cutting-edge research papers last year are now production-ready APIs. This rapid pace means that anyone who isn't deeply embedded in the AI world has a fundamentally limited view of what's actually possible today, let alone what might be possible tomorrow. That's exactly why we believe so strongly in the FDE model: figuring out what an organisation can or should do with AI needs to be led by people who live and breathe these technologies.

As vindication of our belief, OpenAI recently noted that "over the past year, we have experienced a significant increase in demand for OpenAI's hands-on technical expertise to translate abstract ideas into production applications." They've responded by building out their own Forward Deployed Engineering team, with job postings describing roles where engineers "embed deeply with strategic customers to understand their business challenges and technical requirements in detail."

This isn't just about implementation, it's about discovery. When engineers work directly alongside users, they gain contextual understanding that's impossible to get from a requirements document. They see the inefficiencies that business stakeholders have go so used to that they don't even mention them. They spot opportunities for AI to solve problems the client didn't even know they had.

Our Two-Week Discovery Process

Here's how we're structuring this at Barnacle Labs: we send an FDE into an organisation for two weeks with one mission—suss out the broad possibilities.

During those two weeks, our FDE talks to everyone from those working the warehouse floor to the C-suite. They're not just asking "what problems do you have?" They're asking "what are you trying to achieve?" and "what would you do differently if you could wave a magic wand?" Then they apply their technical understanding to figure out which parts of that magic wand are actually achievable with current AI technology. The value that the FDE brings is in their ability to reinterpret what they’re told and use their understanding of the technology to define something that’s both ambitious and achievable. This isn’t a job that someone who doesn’t understand the technology can do.

But here's the key part: they don't just create another presentation. They build something. A small proof of concept that brings the idea to life. We've seen this over and over—something working captures the imagination in a way that PowerPoint simply cannot. When a CEO sees an AI system actually booking their meetings or an operations manager watches an AI agent successfully negotiate with a supplier in real-time, that's when the lightbulb goes off.

Our lab-based researchers and engineers support the FDE during this PoC phase, so we're deploying real technical skill to work out the possibilities, not just theoretical knowledge.

More details on our model here.

Why FDE Oversight Matters for Bigger Projects

After those two weeks, when we sit down with the client to scope out a broader engagement, the FDE stays involved as the project overseer. This isn't just because we like the model—it's because they understand the client and their needs in a way that traditional project managers and consultants simply don't.

As Palantir describes their Forward Deployed Software Engineers:

“FDSEs responsibilities look similar to those of a startup CTO: you'll work in small teams and own end-to-end execution of high stakes projects."

That startup CTO comparison is spot-on. They're not just managing a project; they're thinking strategically about how to deliver the most value.

The FDE has context that's impossible to transfer completely. They know which stakeholders are most influential, which processes are most critical to get right, and which features will actually get used versus which ones just sound good in meetings. They've seen the client's "ah-ha" moments and understand what resonated most deeply.

The Broader Trend: From Product-Led to Services-Led Growth

What we're doing isn't unique to us. The whole industry is shifting this direction. [Andreessen Horowitz](Andreessen Horowitz recently wrote about how "for the better part of the last decade, it's been broadly assumed that product-led growth (PLG) is superior to implementation-heavy enterprise software." But in AI, that assumption is breaking down.

Why? Because AI applications that solve complex, end-to-end workflows require what A16Z calls "active management, guided learning, and rich context that comes from read/write access to internal systems." You can't just ship a SaaS product and expect customers to figure it out themselves.

“Salesforce, ServiceNow, and Workday were able to achieve success because they did a significant amount of implementation work. Enterprise AI products have an even more pronounced implementation requirement, as they require deep integrations and context. To navigate this complexity, services organizations handle the heavy lifting of securely connecting the AI application to internal databases, APIs, and workflows, ensuring models have the context they need — historical records, business logic, and more — to deliver value.” — Andreessen Horowitz

Even OpenAI, despite having some of the most advanced language models in the world, has found that their biggest enterprise customers need hands-on technical guidance. Their Forward Deployed Software Engineers "embed directly throughout the project lifecycle, working on-site and being hands-on with every aspect of the solution from design to production."

The Bottom Line

The companies that will win in AI aren't the ones with the best algorithms (though that helps). They're the ones that can most effectively translate AI's capabilities into real business value. And that requires people who can bridge the gap between cutting-edge technology and practical business needs.

Forward Deployed Engineers aren't just another type of consultant or customer success manager. They're technical professionals who can write code, understand machine learning systems, and explain complex concepts to non-technical stakeholders. Most importantly, they can spot opportunities that others miss because they understand both what the technology can do and what the business actually needs. They’re a rare breed, but one worth cultivating.

At Barnacle Labs, we're betting that this model will become the standard for AI consulting. Because at the end of the day, AI isn't just about the technology, it's about the implementation. And implementation, especially good implementation, requires people who can see possibilities that others can't.

The future of AI consulting isn't about selling software or selling hours. It's about delivering outcomes. And Forward Deployed Engineers are how we're going to get there.

I speak to a lot of executives about AI and the story is always the same — confusion about what’s possible, teamed with frustration with technology companies all pushing their product as the only answer. In this environment we’ve designed our FDE model at Barnacle Labs as a way to cut through the confusion and hype to discover what’s actually possible and achievable. Our initial scoping exercise takes only two weeks and is a great way to short-circuit those endless consulting engagements.

Right from the start, we’ve setup Barnacle Labs as a team that’s heavy on AI expertise. We deliver way more code than we do powerpoint slides and we’re doubling down on that. If you need a team of AI experts who understand the possibilities and have the creativity to imagine ideas that others will have missed, give us a call!