The Forward-Deployed Engineer Is Coming to Enterprise Support
TCS just announced it is building a forward-deployed AI engineering unit of up to 8,900 people. Engineers physically embedded at client sites, working alongside the customer’s teams to implement AI in the real world. Their reasoning: AI can’t operate in the physical environment without humans watching. Without local context. Without someone who can see what’s actually in front of them.
I want to sit with that framing for a moment, because it maps almost perfectly to a problem enterprise support leaders have been struggling to articulate. The forward-deployed engineer isn’t a new concept — Palantir perfected it, Anduril built a culture around it. However, now the largest IT services firm on the planet is betting that this model isn’t a boutique consulting luxury. It’s the operational infrastructure for AI at scale. That has implications far beyond IT services. In fact, it has direct implications for how enterprise support gets done.
What Forward-Deployed Actually Means
The forward-deployed model is simple in principle: you put a skilled expert physically close to the problem, in the environment where the work happens, with real-time context that a remote operator can never have. The insight is that some problems can’t be solved from a distance. The information needed to fix them doesn’t travel well through text descriptions or ticket forms. You have to be there, or you have to have eyes there.
Palantir built an entire services business on this insight. Their forward-deployed engineers weren’t just implementation specialists — they were context gatherers. They saw the customer’s actual environment, understood the edge cases that never made it into any requirements document, and built solutions that reflected the physical and operational reality of the deployment. Consequently, remote teams building to spec routinely missed those things. FDEs didn’t.
TCS is essentially saying: that’s what AI implementation requires at enterprise scale. You can’t drop an AI agent into a complex operational environment and expect it to work without someone on the ground. The physical world is messier than any training dataset.
Enterprise Support Has Had This Problem for Years
Here’s the thing: enterprise support teams have been running the forward-deployed model informally, under-resourced, for decades. Every field technician dispatched to a customer site is a forward-deployed support engineer. Every senior agent who knows they need to “see it to believe it” before they can diagnose the issue is applying FDE logic. The problem isn’t that enterprise support doesn’t understand this model. The problem is that the tools for scaling it have never existed.
What TCS is building with 8,900 people, enterprise support organizations need to replicate with the agents they already have — without sending every one of them to a physical site for every issue. That’s the operational constraint. You can’t hire your way to the forward-deployed model at ticket volume. Therefore, you need a way to project those eyes and that context remotely, at scale.
This is where the conversation stops being abstract and starts being urgent.
Concentrix Named This the CX Challenge of 2026
Earlier this year, Concentrix called it explicitly: the defining challenge in customer experience for 2026 is moving AI from pilot to production. Not building AI. Not proving that AI can deflect tickets in a controlled environment. Moving AI into the messy, physical, context-dependent reality of enterprise support operations.
Why is this hard? Because AI agents are exceptional at closing tickets where the information needed for resolution is already in text form. In contrast, they’re not good at the tickets where the information needed for resolution is visual. Physical. Spatial. “The installation doesn’t look right.” “The device is behaving oddly but I can’t describe how.” “There’s something here you need to see.”
These aren’t edge cases for companies supporting hardware, infrastructure, field equipment, or complex physical products. As AI sweeps the easy, text-solvable tickets off the queue, what’s left is disproportionately the hard, visual, physical kind. Moreover, the AI-to-production gap that Concentrix identified isn’t a training data problem or a model capability problem. It’s a context problem. You can’t route AI to what it can’t perceive.
The Support Agent at the Customer Site Is Already the FDE
Think about what a field support technician actually does. On arrival, they show up with context from the ticket and immediately begin reading the physical environment. In practice, they notice things the customer never mentioned — the cable routed wrong, the unit installed in a configuration not in any guide, the environment running hotter than spec. Without hesitation, they make decisions based on what they see, not what was described. Functionally, they are forward-deployed engineers.
The problem is that this capability doesn’t scale. One FDE handles one site at a time. One field dispatch handles one problem per truck roll. If the ticket volume requiring physical context is 20 percent of your queue, you need a lot of trucks. As a result, if AI deflects the easy 60 percent and what’s left is disproportionately the visual, physical cases, your field deployment burden doesn’t shrink proportionally — it may actually grow as a share of remaining work.
Enterprise support leaders are trapped in a model where the most expensive resolution method — field dispatch — is exactly what the AI transformation era is concentrating their remaining tickets toward. That is a cost structure that gets worse before it gets better, unless something changes about how remote agents can acquire physical context.
Remote Visual Guidance Is the Scaling Layer
The scaling layer for the forward-deployed model in enterprise support isn’t more people. It’s better eyes. A support agent who can see what the customer sees in real time — through their phone camera, annotating on a live video feed, directing attention to exactly the right component — is approximating the FDE model without the travel. In other words, they have context, visual confirmation, and the ability to diagnose before dispatching. They can guide remediation without rolling a truck.
This is precisely what live visual support technology enables. One expert, working remotely, can project their presence into the customer environment through the customer’s own device camera. The installation becomes immediately visible. Annotation on the live feed is instant. The agent can direct the customer’s attention to exactly the right spot and make the same judgment calls a field engineer would make — without the dispatch cost, the scheduling delay, or the geographic constraint.
A camera-equipped support interaction IS the forward-deployed model, scaled. The FDE insight — that some problems require physical presence and visual context — doesn’t require physical presence. It requires visual context. Ultimately, remote visual guidance delivers that context at a fraction of the cost of a truck roll and a fraction of the delay of scheduling a site visit.
Why This Is an Enterprise Bet, Not a Feature
The reason the TCS announcement matters isn’t the headcount number. It’s the strategic signal. A company with $30 billion in annual revenue is reorganizing around the insight that AI implementation requires physical presence and real-world context. They’re treating forward-deployment not as a premium service tier but as a core delivery model for enterprise AI.
Enterprise support is about to face the same strategic reckoning. AI deflection strategies that don’t account for the visual, physical residual are incomplete. For instance, support organizations that cut headcount, deploy AI deflection bots, and declare transformation done are building a system optimized for the tickets that were already commoditizing — and leaving the hard tickets, the expensive tickets, the renewal-deciding tickets with fewer people and the same old phone-and-email toolset.
The forward-deployed model for enterprise support isn’t about sending armies of consultants. At the same time, it’s about equipping every remote support interaction with the capability to become a forward-deployed interaction the moment the problem requires physical context. That capability has to be on by default: cameras always available, visual annotation one tap away, every agent trained to reach for visual context the way they currently reach for a knowledge base article.
The Window Is Now
Concentrix said it plainly: the move from AI pilot to AI production is the CX story of 2026. TCS is saying something compatible from a different angle: you can’t make that move without eyes in the physical environment. Both observations point to the same gap. Therefore, the enterprise support organizations that close this gap now — by building visual support into their standard resolution workflow — will have a structural advantage as AI reshapes the ticket composition of every support queue.
The ones that don’t will find themselves with AI handling the easy cases, a shrinking team handling the hard ones, and no way to scale the physical context that hard cases require. They’ll be paying for FDE-level field dispatch to solve problems that a remote agent with a live camera feed could have resolved in ten minutes.
TCS building 8,900 forward-deployed AI engineers is a statement about where the value is. The value is in the physical world. The value is in the real environment, not the documented one. For enterprise support, that value starts with seeing what the customer sees — before you dispatch anyone, before you escalate, before you schedule a site visit that could have been a five-minute video call.
The forward-deployed engineer model is coming to enterprise support. The question isn’t whether. The question is whether your support org is going to scale it intelligently or pay for it the expensive way.