Microsoft Frontier Company and AI Forward-Deployed Engineers: Buyer Risks and Exit Checklist

What FDE actually is​

An FDE is a vendor or partner engineer placed inside a customer's organization. The engineer is given an outcome to deliver rather than a tightly scoped statement of work.

Ryan Sheehan, a senior vice president at solutions integrator SHI, told The Register that the industry has "been doing FDE for a long time." He described it as sending top consultants to a client without tying them to a single project. They get a mission to build something valuable with new technology, and wide latitude to move quickly.

AWS describes its current version in similar terms. Its engineers work with a customer's business, engineering and security teams to build and deploy production AI systems using the customer's data, governance and processes. AWS says engagements are structured around shared business results rather than billable hours.

AWS's Taimur Rashid told The Register this is the key difference from a conventional engagement. There is no scope of work and no hourly charge. In his words, the customer wants to build and "the path is not determined."

The French restaurant origin story​

The term comes from Palantir. CTO Shyam Sankar says he coined it in 2007. According to his account, CEO Alex Karp asked why French restaurants are so good, then answered his own question. The wait staff are part of the kitchen staff. They understand the food and the technique, and they are not just couriers between kitchen and table. Karp asked Sankar to build the engineering equivalent.

This is Sankar's own recollection, published on his personal Substack. It is a good story, but it has not been independently verified.

Sankar also says Palantir's investors disliked the model because the cost of embedded engineers hurt margins. He argues it paid off because customers actually understood the product and got value from it. His line was that Palantir didn't want to throw software over the wall and hope customers would work out what it meant.

It's not new, and the vendors say so​

The Register's reporting shows how often vendors concede the practice is old:

• DoIT, a cloud monitoring vendor, says it has used FDEs since 2011. It adopted the label only in the last quarter of 2025, when it renamed its Customer Reliability Engineering practice.
• AWS says it started in 2017 with an ML Solutions Lab that loaned customers a data scientist to build a proof of concept. It sometimes called this a secondment, a resident architect or a resident scientist.
• Cisco's Carlos Pereira says embedding experts on-site to help customers adopt technology is not new, and that Cisco's Customer Success organization has done it for years. He argues AI adds new variables: non-deterministic models, evaluations, faster-moving requirements and skills many customers lack in-house.

Directions on Microsoft, the independent Microsoft analyst firm, offers a similar read on Redmond's move. Citing sources, it says the Frontier Company looks like a make-over of Microsoft Consulting Services and the Cloud Solution Architects program, among other existing groups. It also says it isn't clear whether the $2.5 billion is new money or recommitted consulting budget.

The Microsoft and AWS announcements​

The money is large, and the timing was close.

AWS (June 30, 2026). Amazon announced a dedicated Forward Deployed Engineering organization backed by a $1 billion investment. AWS calls the approach agentic-first and says it aims to compress deployments from months to days. It says customers should be self-sufficient when a deployment ends.

AWS says customers should end up with:

• deployed systems
• knowledge graphs
• runbooks
• architectural documentation
• trained internal champions

AWS also says a semantic layer is deployed into the customer's own AWS account. It lists hardware-based isolation and end-to-end encryption among its security measures. It names the Allen Institute, Cox Automotive, the NBA, the NFL, Ricoh and Southwest Airlines as customers. The "months to days" figure is AWS's own target. It has not been independently measured.

Microsoft (July 2, 2026). Commercial Business CEO Judson Althoff introduced Microsoft Frontier Company, a new operating business for AI transformation. Microsoft is investing $2.5 billion and embedding 6,000 industry and engineering experts at customers. They are meant to co-design, deploy and continuously improve AI systems against measurable business outcomes.

Microsoft says this "goes beyond" what has been labeled FDE. It says the unit combines industry knowledge, change management and enterprise AI engineering. Rodrigo Kede Lima is its president.

Microsoft's blog names LSEG, Land O'Lakes, Unilever and Novo Nordisk as customers. It describes an LSEG project that embeds AI into LSEG Workspace so finance professionals can query structured and unstructured content. It also lists FDE partnerships with Accenture, Capgemini, EY, KPMG and PwC.

On trust, Microsoft says a customer's data and IP will not be used to train models in ways that commoditize what differentiates them. It also pitches a model-diverse platform that runs models from OpenAI, Anthropic, Microsoft AI, open source or specialized vendors.

One commentator noted that deployments built on Microsoft's tooling naturally deepen Azure dependence over time, even with those assurances. That is analysis rather than a Microsoft admission. It is the right question to ask of any vendor-run program. Note too that Microsoft's own scale and outcome claims are unaudited.

Reporting on the launch indicates that the unit pulls together existing Microsoft forward-deployed engineers, technical consultants, support staff and salespeople with vertical industry expertise. That fits the "make-over" reading above.

What the analysts warn about​

The Register says Gartner and Forrester both see value in FDE and both flag risks. Gartner's March guidance, "An AI Leader's Guide to Forward Deployed Engineering to Scale AI," notes that FDEs often have full-stack access to the customer's environment and the vendor's codebase. That lets them build and iterate in real time. Forrester's June report praises the model for helping turn complex platforms into production-ready outcomes.

Both firms also warn of dependency. As quoted by The Register:

• Forrester warns of a "bespoke trap," where highly tailored solutions deepen vendor dependence and make future change costly.
• Gartner cautions against ill-conceived engagements that lead to vendor dependency, security and data risk exposure, technical debt and "talent atrophy." Talent atrophy is when internal teams become operators of a black box rather than architects.
• Gartner also worries that hard-coded, manually designed domain logic will be superseded as large language models improve, forcing costly rework.

Only Gartner's abstract is public, so the detailed quotes rest on The Register's reporting.

A buyer's checklist​

This list is my own synthesis from the risks above. It is not a Gartner or Forrester checklist. If a vendor or partner offers you an FDE engagement, get answers to these questions in writing:

1. Who owns the code, architecture and artifacts? Can you take them elsewhere?
2. Where does the system run? Is it in your tenant or subscription, under your governance and access controls?
3. What is the security model? Who holds credentials, and what level of access do the engineers have to your data and environment?
4. How does skills transfer happen? Look for runbooks, documentation and named internal staff who shadow, co-build and then operate.
5. How is success measured? "Outcome-based" should still mean agreed metrics.
6. What is the exit plan? Define the handoff date, and what it costs to change course afterward.
7. What happens when the models change? Ask how much hand-built logic must be rewritten.

Will FDE last?​

Omdia chief analyst Jay McBain told The Register that the big vendors are "playing both ends." They invest in FDEs while also investing heavily in partner enablement. He cites Omdia data that 82 percent of partners say they aren't ready to quickly grow their AI services. He predicts FDE investment will fall back into the channel by 2031 as the market matures.

His reasoning is that vendor investors don't want a low-margin services business diluting overall revenue. That is a prediction, not a fact. It does fit the way Microsoft is leaning on global systems integrators for scale. A Microsoft-oriented outlet reported that Accenture has already launched a dedicated Microsoft FDE practice.

The bottom line​

FDE is a rebranded and well-funded version of an old idea, and the AI moment gives it a real argument. AI projects often stall between proof of concept and production, and having builders in the room can help. The cost is the same one the analysts name: the people who build it may know your system better than you do.

For Windows and Microsoft shops, the practical takeaway is to treat an FDE offer as a build-and-transfer arrangement. Judge it by what you can run, understand and change after the engineers leave.