Forward deployed engineers vs consultants vs systems integrators: who should implement your AI?
Three ways to get AI into an enterprise, and what each one actually delivers. A buyer's comparison of forward deployed engineers, management consultants and systems integrators for AI implementation, with the questions to ask each before you sign.
When an organisation decides to put AI into its operations, the first real decision is not which model to use. It is who does the work. There are three common answers: a management consultancy, a systems integrator, or the AI vendor's own forward deployed engineers. They are sold as interchangeable. They are not, and the difference shows up about three months after go-live.
The three models, plainly
Management consultants
A consultancy studies the organisation, identifies opportunities, and produces a strategy and a roadmap. Good ones are excellent at the first part: they get access to leadership, they frame the business case, and they make a decision possible. What they usually do not do is build. The deliverable is a document, and the implementation is handed to someone else, which is where much of what they learned is lost.
Systems integrators
An integrator builds to a specification. They are strong at large, well-defined programmes: ERP rollouts, infrastructure, integrations where the requirements can be written down in advance. The model is fixed scope, delivered and accepted. That works when the requirements are knowable up front. With AI they mostly are not: the important rules are discovered by watching the work, not by interviewing about it.
Forward deployed engineers
A forward deployed engineer is an engineer from the AI vendor who works inside the customer's operation, finds the work worth automating, builds it against the real systems and stays until it runs. The same person discovers the requirement and ships the fix, often on the same day. What they give up is scale: an FDE team can only take on as many customers as it has engineers.
| Consultant | Systems integrator | Forward deployed engineer | |
|---|---|---|---|
| Main deliverable | Strategy and roadmap | System built to specification | Automation running on real work |
| Who builds | Someone else, later | Integrator's delivery team | The same engineer who did discovery |
| Where the work happens | Workshops and interviews | Mostly remote | Inside the operation |
| How requirements are found | Interviews | Written specification | Watching the work, then building |
| Handles the unwritten rules | Documents some | Only if specified | Discovers and builds them in |
| Ends at | Report delivered | Acceptance | Measured results on live work |
| Best for | Deciding whether and where | Large, well-specified programmes | Workflows where the rules live in people's heads |
Why AI changes the answer
For most enterprise software, the specification is the product. For AI, the specification is the smallest part. An AI agent that answers service-desk calls needs to know which requests matter, how staff actually describe them, which system wins when two disagree, and exactly when to stop and hand over. Those are discovered by listening to calls and reading tickets, and they change the build daily for the first few weeks.
A model where one team discovers and another builds loses that information at the handover. A model where the builder sits inside the operation keeps it. That is the practical case for forward deployment, and it is why Voho delivers that way. The longer case is in how Voho puts AI to work across the whole enterprise.
They are not mutually exclusive
The combinations often work best. A consultancy can frame the programme and set priorities. An integrator can own the large platform work around it: identity, networks, the ERP itself. Forward deployed engineers then take individual workflows from problem to running automation inside that frame. In the Kingdom, it is also common for a Saudi prime contractor to hold the contract while the AI vendor's engineers do the build; that is how our own delivery at Aramco is structured.
Questions to ask whoever you hire
- Who exactly writes the code, and will they be on our site?
- Will the person who does discovery be the person who builds?
- What does the engagement end with: a document, an acceptance certificate, or measured results on our real work?
- How will you test it against our own history before we rely on it?
- What happens when the requirement changes in week three, as it will?
- Where does our data go, and can the whole thing run inside our network?
- Which of our systems will you need access to, and in which week?
If you want to see how Voho's forward deployed engineers would approach a specific workflow, book a working session. If the honest answer is that you need a consultant or an integrator instead, we will say so.
Frequently asked
- What is the difference between a forward deployed engineer and a consultant?
- A consultant identifies opportunities and delivers a strategy or roadmap; implementation is usually done by someone else. A forward deployed engineer works inside the operation and builds and ships the automation, owning it until it runs on real work.
- What is the difference between a forward deployed engineer and a systems integrator?
- A systems integrator typically builds to a written specification, often remotely, and hands over at acceptance. A forward deployed engineer discovers the unwritten requirements on site and builds them in, staying until results are measured on live work.
- Which is best for AI implementation?
- For workflows where the important rules live in people's heads rather than in a specification — service desks, contact centres, document-heavy back offices, operations — forward deployed engineers are usually the better fit. Consultants and integrators remain strong for strategy and for large, well-specified platform programmes, and the models often work together.
Keep reading