Integrator, engineering office or reinforcement

When a pharma production manager calls us, the first sentence is often the same. "We need help on automation, but we don't quite know what type of provider to look for." Behind this hesitation lies a Swiss market reality: three outsourcing models coexist, each with its own economic logic, deliverables and risks. And the wrong choice is expensive, sometimes six figures, without anyone realizing it before qualification.

This article describes the three models, the criteria for choosing, and the classic mistakes we see on the ground after ten years in French-speaking Swiss pharma.

The three automation outsourcing models

The turnkey integrator

The integrator takes full responsibility for a defined scope. They sign a contract for that scope, deliver a working solution, and remain accountable until commissioning. For a new sterile filling line, for example, the integrator delivers the functional analysis, PLC code, SCADA, HMI, sometimes the electrical cabinet, FAT and SAT tests, and GAMP 5 documentation.

The integrator defines scope and responsibilities before work starts. Any change to the initial scope requires an amendment.

For more on the qualification phase, see our practical FAT/SAT pharma guide.

The automation engineering office

The engineering office sells design, not commissioning. They deliver specifications, detailed functional analyses, network architectures, tender specifications. The implementation itself is then handed off to an integrator or done in-house.

This model is particularly suited to the upstream phases of large capex projects. When a pharma site plans a new 30-million workshop, the engineering office writes the User Requirement Specification, models data flows to the MES, sizes the OT infrastructure, and prepares the tender package for candidate integrators. Decoupling design from execution allows integrators to be put into competition on a clear scope, often saving 10 to 20 percent on the final ticket.

Our approach in this mode is detailed under our architecture and design service.

The automation reinforcement

A reinforcement is an experienced automation engineer who comes to work inside your teams, with your methods, on your project, without taking responsibility for the overall outcome. They integrate into your organization. Also called body shopping or technical staffing.

The model is useful when the need is short-term, the scope is fuzzy, or when you already have the in-house skills but not the bandwidth. A site needing to absorb a three-month commissioning peak on SIMATIC S7 doesn't need a turnkey integrator. It needs two experienced automation engineers who show up, understand quickly, code cleanly, and leave.

Our automation reinforcement offering covers this need, including short missions of 4 to 12 weeks.

When to choose which

Choose the turnkey integrator

Choose an integrator when the scope is clear, the deadline is known, you want to transfer the risk. It's the default model for new greenfield installations, complete lines purchased from an OEM, or major modifications with strong GMP impact.

The decisive criterion is specification maturity. If you know precisely what you want, the integrator will deliver it within the agreed scope. If you don't, the integrator will deliver something too, but according to their interpretation, and change orders will rain down.

Choose the engineering office

Choose the engineering office when you're in the upstream phase of a significant capex project, or when you want to prepare a competitive tender. It's also the right choice for pure consulting missions: OT architecture audit, site automation master plan, Industry 4.0 migration strategy.

The engineering office is also useful for arbitrating between integrators after offers come in. Having an independent third party dissect technical proposals avoids being sold an oversized solution or proprietary lock-in.

Choose the automation reinforcement

Choose reinforcement when your need is tactical, urgent, or hard to specify in advance. Three typical pharma cases:

First case, commissioning to absorb. Your integrator delivers, your in-house teams are saturated, you need competent hands for 8 to 12 weeks for testing, debugging, training operators. This need calls for someone working alongside your team. You need a reinforcement.

Second case, production support. A recipe stops running properly, a sensor drifts, the SCADA crashes once a week without anyone knowing why. The reinforcement comes in for investigation, identifies the cause, fixes it. A few days to a few weeks.

Third case, project with fuzzy or evolving scope. You know something needs modernizing, but the scope will firm up over time. Reinforcement lets the work progress as the need becomes clearer.

Concrete decision criteria

Here's the grid we use when a client hesitates.

Project scope

Clear, closed scope: integrator. Evolving scope: reinforcement. For complex projects, separating design from execution keeps technical oversight independent.

Urgency

Reinforcement mobilizes in 2 to 4 weeks. Turnkey integrator: count 2 to 4 months between signature and effective start (commercial cycle, kick-off, analysis). If you're in firefighting mode, the integrator is not the answer.

In-house skills

If your teams know how to do it but lack time, reinforcement. If your teams don't know how and don't want to learn, integrator. If your teams don't know how but want to upskill, mix reinforcement plus training, or integrator with a contractual knowledge transfer clause.

A classic pharma trap: the in-house automation team empties through successive retirements, no one trains the next generation, and everything goes to the integrator by default. Five years later, the site has no autonomous modification capacity. Maintaining in-house skills preserves site autonomy.

GMP sensitivity and qualification

For anything touching qualified systems, GxP, or with major CSV impact, favor providers who genuinely master GAMP 5 and associated documentation. It's less a question of model (integrator or reinforcement) than a question of profile. Our CQV pharma service specifies expectations.

The classic mistakes

Mistake #1, hiring an integrator on a fuzzy scope

Symptom: six months after signing, you've signed three change orders and the scope keeps changing. Root cause: scope wasn't mature at contracting. Solution: take an engineering office first, or a reinforcement to scope, then launch the integrator on rock-solid scope.

Mistake #2, hiring a reinforcement on a project that needs end-to-end responsibility

The reinforcement works for you, under your direction. If you ask them to "deliver the line" without anyone in-house actually steering, you'll have holes in the racket. No one runs the tests, no one validates the docs, no one challenges technical choices. Solution: if you don't have an in-house automation project manager capable of carrying responsibility, take an integrator. Or also take an outsourced project manager.

Mistake #3, hiring an engineering office that doesn't know how to deliver

Some engineering offices write magnificent specs that are disconnected from the field. The integrator gets the package, raises 200 questions, everything has to be redone. Solution: choose an engineering office whose engineers have also coded and commissioned recently. Five years without touching a PLC shows in the deliverables.

Mistake #4, mixing models without clear contracting

You start in reinforcement, it works well, you decide to switch the provider to integrator on the rest of the project. Without formal scoping of scope and responsibility transfer, you create a hybrid mode combining the worst of both: the provider keeps billing time-and-materials while taking on moral commitments they won't keep. Solution: at each model transition, redo a clean contractual checkpoint.

Mistake #5, neglecting the legal checklist in Switzerland

Long-duration reinforcement raises questions about staffing leasing (LSE) and possible reclassification into employment. Work with providers who have the proper authorizations and who structure their work correctly (mission, deliverables, operational independence). Otherwise, in case of audit, you, the client, bear the tax and social consequences.

Combining models

In practice, good pharma projects often combine all three models over time.

Phase 1, design: engineering office for specs and architecture. Phase 2, execution: turnkey integrator on the main scope. Phase 3, commissioning: reinforcement to absorb the activity peak and support qualification. Phase 4, operations: occasional reinforcement for evolutions, until the next major project.

This sequence optimizes total cost, keeps technical control, and maintains in-house skills instead of diluting them in a single provider.

See our pharma sector references for concrete mission examples in each of these modes.

Frequently asked questions

How do I know if my current provider is the right model for my need?

Ask yourself three questions. Is my scope stable and clearly specified (yes = integrator, no = reinforcement)? Do I want to transfer the risk or carry it in-house (transfer = integrator, carry = reinforcement or engineering office)? Do I have an in-house project manager capable of technically steering (yes = reinforcement possible, no = integrator recommended)? If your answers don't match your current provider's model, you probably have a contractual mismatch.

Can we change models mid-project?

Yes, but with proper contracting at each transition. The trap is sliding informally from one mode to the other without clarifying who carries what. The right reflex is to mark a stopping point between phases, do a review, and sign a new contract (or explicit amendment) for the next phase. The administrative cost is low compared to the risk of contractual gray zones.

How do I properly brief an automation provider at startup?

Prepare four things before kick-off. First, a clear functional description of what the installation must do (even imperfect, better than nothing). Second, the target technical environment (PLC range, SCADA, network, MES integration if applicable). Third, GMP constraints and expected CSV impact level. Fourth, the project governance mode (who decides what, which steering committee, at what frequency). With this foundation, any competent provider can start efficiently. For an in-depth scoping, let's talk directly.