Why Field Service AI Needs Forward Deployment Engineers

Prateek Chakravarty, CEO at Zinier, explains why the most successful AI deployments in field service require embedded domain experts working directly with customers.

This week, I sat down with Prateek to discuss why companies are investing millions in AI implementation services and what field service organizations can learn from this trend. His perspective on "Service as a Software" challenges how we think about enterprise AI deployment.

Why Companies Pay Millions for AI Implementation

Oli: Hey so OpenAI just announced consulting engagements to deploy GPT-5. Why are companies paying millions for implementation?

Prateek: I fully get it. I think it's absolutely necessary.

If you think about it from a customer perspective, the basic concept of adopting a new technology which is game-changing - you cannot force fit that technology into the standard way of doing business. There has to be a new paradigm. There has to be an understanding of what problems need to be solved and how they have to be solved differently.

That's where companies like OpenAI and what we're doing come in - not treating it as SaaS where you just sell the product and run away, but treating it as a problem to be solved, the job to be done. We're looking beyond the product to understand what it would take in this phase of evolution to get the job done.

I love this. It makes total sense that they're looking into helping companies adopt AI and thinking beyond the standard definition of what a subscription is and what a service is. The job has to be done, and let's put the best engineers forward to get the job done.

AI deployment isn't SaaS - you can't sell the product and walk away. It's about solving problems and getting the job done with embedded expertise.

Domain Expertise vs. Horizontal AI: The Verticalization Advantage

Oli: There's also benefit from having domain experts, right? OpenAI are the experts in AI, Zinier are the experts in field service. Rather than expecting customers who aren't experts in AI to implement AI, or people who aren't field service management experts to deploy FSM solutions, there's room for experts there?

‍Prateek: To solve a customer's problems you have to understand the problem, empathize with the problem, and solve the problem. Companies like OpenAI will take a very horizontal approach - finding forward deployment engineers and putting them across multiple industries. We take a vertical approach. We're saying we want to be experts at solving field service and asset-related problems.

The tech expertise comes from our AI product managers, but the LLM expertise comes from OpenAI Cloud and all the other models we're using. We don't need to reinvent the wheel there. We have product managers and forward deployment engineers who understand the customer, understand the pain point, and talk in their language.

‍It's no point knowing the tech if you cannot talk in the language of the customers. That becomes very important. This is effectively verticalization of AI where we're going deep into field service and asset domain. That's the gap we're plugging in.

Horizontal AI deployment spreads thin across industries. Vertical expertise means understanding customer pain points and speaking their language, not just knowing the technology.

Learning from Palantir: Beyond the Standard SaaS Mindset

Oli: Palantir has been doing forward deployment for 20 years. What can field service companies learn from their model?

‍Prateek: I think Palantir has been around for a long time, and they've been beyond the standard definition of SaaS. 15-20 years back, they were pooh-poohed because they were not SaaS. They were very hands-on in delivering value, which doesn't fit into the SaaS mindset.

Right now, hopefully we're realizing that it's not about selling SaaS, it's about solving a customer's problems. If a simple point solution can solve the problem, great - low touch. But what AI has shown is there are better ways of solving the problem - quicker ways, more cost-effective ways. But to get there, companies like us have to be more implementation-heavy or more customer-oriented to help them understand what needs to be done.

There's a lot to be learned. Our job doesn't stop after selling the product. It only starts. Then what steps do we need to take to help our customers solve the problem in the way that we're advocating? We believe, just like the industry believes, that AI is a game-changer, and it truly is. There are so many different use cases, so many jobs to be done.

‍It's our job to help the customer, and that mindset shift becomes important. We have to help them understand that this is beyond the standard IT processes. This is beyond the standard platform SaaS play, and that's where forward deployment comes in.

Palantir was criticized 20 years ago for being "too hands-on" and not pure SaaS. Now we realize it's not about selling SaaS - it's about solving customer problems.

Service as a Software: Redefining the Customer Relationship

Oli: You mentioned "Service as a Software" on a recent podcast. How does that differ from traditional SaaS implementation?

‍Prateek: How did SaaS come about? SaaS came about because the software industry figured out you narrow down the problem to solve, solved a small problem, made it repeatable and low-touch. That is what SaaS is.

It's very product-centric, and you're relying on the fact that your customers will break the problem into smaller chunks, find a solution, and stitch things together at their end. When you think of the industries we serve - the core physical industries - they don't have huge IT Teams. More often than not, you don't expect them to have hundreds of SaaS applications and do the stitching at their end. Some will do, but most won't.

So then you have to look at the problem differently. My job is not to sell a product. My job is to solve your problem as a customer, and then I have to find the best way to solve that problem. I will use software, but then I'm offering you a service. I'm solving your problem. My job is not to sell the product. My job is to solve your problem.

If that problem is solved by selling a simple point SaaS solution, great. But if it involves leveraging LLMs and AI to help you evolve and do certain things faster, quicker, sooner, then I have to do that job for you. That's where the entire concept of service comes in.

Physical industries don't want hundreds of SaaS apps. My job isn't to sell a product - it's to solve your problem. If that requires AI and LLMs, I'll do that job for you.

Breaking Down Tech Stack Silos: Managing Workforce and Assets

Oli: This approach seems like it would help buyers too, no? Instead of asking "what software can I find," should they focus on their actual problems?

‍Prateek: It is us techies who are forcing customers and buyers to think in terms of tech stack. A good example in our world: we have field service management tech stack, we have computerized maintenance system tech stack, we have enterprise asset management tech stack.

‍But if you think from a customer perspective, the goal of a customer is they need help to manage their workforce and assets - these are people and equipment that allow them to deliver mission-critical service, whether it's electricity or gas or internet or medical service or construction, whatever it is. We're forcing them to think along these silos.

If you redefine the problem and say, "How can I help you better manage your workforce and assets?" the problem changes. It's not that you need an FSM and a CMMS and an EAM - you need a platform that can give you that visibility. Can I incorporate AI to solve those problems in a better way? Can I work with you to make sure that it is deployed in your processes the right way so that the results come?

‍Now it's no longer about finding an FSM product or a CMMS product or an EAM product. It's about solving the problems and ensuring that whoever you're working with is helping you solve that problem and getting to that outcome.

You might still end up with the same products - I'm okay with that. Or you might come to us and say, "We have one platform to do everything." But now you're focused on solving the problem and not buying a product.

Customers need help managing workforce and assets, not FSM+CMMS+EAM silos. Focus on solving problems, not buying products. The platform should give you visibility and results.

Bottom Line

The field service AI revolution isn't going slowly because of bad technology - it's failing because organizations are treating transformative AI like traditional software purchases. Success requires embedded domain experts who understand both the technology and the unique challenges of managing physical operations. The winners won't be those with the best algorithms, but those with the right deployment methodology and the commitment to make AI actually work in the real world.

‍Want to learn more about Zinier's forward deployment approach? Schedule a demo to see how our value release methodology can accelerate your AI journey.‍