INSIGHTS

The Forward Deployed Fiduciary: AI Built to Belong to You

By: Michael Patton, Chief Operating Officer and Co-Founder at MUDRICK & ASSOCIATES

Why most “embedded AI” approaches still leave organizations exposed — and what a fiduciary alternative looks like

Most organizations exploring custom AI solutions today eventually encounter some version of the Forward Deployed model. Popularized by Palantir and since adopted by a wave of AI platforms, the approach places engineers inside the client’s organization to identify use cases, build solutions, and drive adoption in real time.

On the surface it sounds progressive, but in practice it often recreates old problems under a new name.

The embedded team is typically incentivized to expand usage of the vendor’s platform. Solutions are built on top of that platform. Data frequently moves into the vendor’s environment. Over time the client accumulates technical debt and switching costs. What began as “help us get started with AI” quietly becomes a long-term dependency and true vendor lock in.

From a consulting perspective, onsite embedding is not new. McKinsey has been doing it since the 1920s. Lean practitioners have long emphasized going to the gemba — the real place of work — to understand processes before prescribing solutions. For us, the real difference begins at the incentive structure level.

A Different Starting Point

At Mudrick & Associates we start from a different premise.

We treat the AI and machine learning solutions that we build as assets that belong to, and drive value for, the client. Our role is closer to that of your internal AI department and our incentive structure is closer to that of a fiduciary than a platform vendor: under our fiduciary model, we are compensated for helping the client grow the value of those assets over time, not for locking them into our infrastructure.

This produces several practical differences:

Solutions are designed and deployed inside the client’s own environment (or a client owned cloud environment we help establish for them).
There is no vendor lock in created by proprietary platforms or accumulated technical debt.
Data doesn’t leave the client’s environments or control.
Our incentives remain aligned with the client’s long-term outcomes rather than short-term platform adoption metrics.

We call this the Forward Deployed Fiduciary (FDF) model.

How Engagements Typically Begin

We begin with a structured discovery phase. This includes onsite time, job shadowing, stakeholder interviews, touring your facilities, and process mapping. The deliverable is a clear architectural view of the current state, how that architecture can evolve, and a practical Enterprise AI Transformation roadmap sequenced by value, speed, and risk.

Discovery is a paid engagement. We have found that treating it as an investment on both sides produces better focus and better outcomes than offering it as a free sales exercise.

From there, the preferred structure is a flexible engagement model that allows the client to start at a manageable level of commitment and expand as measurable value is demonstrated. We maintain a consistent embedded presence — regular working sessions and onsite collaboration as needed — while the scope and intensity of the work grow in proportion to the results being delivered.

This approach lowers the barrier to entry for the mid-market while preserving the ability to scale with our clients over time. It is intentionally designed for organizations that want durable AI/ML solutions that scale, rather than a series of disconnected projects.

What This Looks Like in Practice

One recent engagement with a mid-market company helps illustrate the pattern. The organization recognized the potential of AI to aid in their growth but lacked a clear path forward. Discovery produced a prioritized roadmap. Early work focused on a high value, relatively contained use case that could demonstrate impact quickly while building internal confidence.

As that foundation took hold, the work expanded into broader agentic capabilities and additional machine learning opportunities tied directly to revenue and operational efficiency. Throughout, the solutions remained client owned assets running in the client’s environment. Governance and human-in-the-loop processes were designed so the client’s own teams retained meaningful control and could continue improving the systems over time.

The relationship has continued to deepen because the value has continued to compound.

Why This Structure?

Two common alternatives are pure fixed fee projects and traditional retainer models. Both have their place — we engage in both — but neither is fully optimized for long term engagements focused on building durable, value generating AI/ML capability.

Fixed fee works well when the scope is fully known in advance. Building production grade, context aware AI systems rarely is. Early work surfaces new data realities, higher value opportunities, and adoption challenges that could not have been fully specified at the outset. A rigid fixed fee structure may constrain the work artificially or create ongoing friction around change orders. It also tends to optimize for delivering the original scope on the original date rather than maximizing long term value.

Traditional retainer models offer predictability and consistent capacity. In many cases that is exactly what a client needs. The limitation is that the relationship often remains largely transactional, and the work does not naturally compound into lasting internal capability or client owned assets.

The FDF model is designed differently. It combines embedded presence for real operational context, client ownership of the resulting assets, and commercial terms that expand only when measurable results justify it.

This structure sometimes raises a question about accountability: if a timeline slips, the monthly engagement continues, whereas a fixed fee model can withhold final payment until delivery. That observation is true on a narrow, schedule only definition of accountability. Fixed fee creates sharper short term pain for the vendor when the original date is missed.

We believe accountability in this domain is broader. It includes transparency of progress, shared visibility into intermediate results, the ability to course correct in real time, and commercial terms that only grow when value has already been demonstrated. Persistent under performance still carries clear consequences — slower or zero expansion of the engagement, eroded trust, and a higher likelihood the relationship ends. Because the model is built for multi year partnerships, reputation and reference value matter a great deal.

In short, the structure is less optimized for enforcing an arbitrary original timeline on imperfectly understood work, and more optimized for ongoing results and the long term health of the engagement. That is a deliberate trade off. It is better suited for organizations that want to build scalable AI solutions that compound in value over time.

Closing Thought

We believe that the power of comprehensive, context aware, secure, and scalable AI and ML solutions should not be reserved for Fortune 500 companies. The most common roadblock for the mid-market is a lack of resources to design and execute a multi-year AI/ML roadmap that drives real value without handing the keys to a third-party vendor and creating lock in.

The Forward Deployed Fiduciary model exists to remove that roadblock. Client owned assets. Data that stays under client control. Incentives that expand the engagement as results are delivered and continue to scale for the client.

We are still in the early innings of what will be one of the most significant growth opportunities most companies will face. Sitting still carries risk. Rushing into opaque technical debt carries its own. The right path is building real capability with clear ownership and aligned incentives.

That’s the work that we do.

Interested in exploring whether this model is a fit?

We work with a limited number of mid-market organizations that have the ambition to build and own their own custom AI and machine learning solutions to drive growth for their organization, rather than to rent it.

If that describes you, we’d be glad to have a conversation.

Let's Talk