Summary
- R+D’s value is not simply that it can combine robots, filling, inspection and packaging. It is that one engineering organisation can make the mechanical, electrical, software and validation decisions meet at a usable operating window.
- That concentration can reduce hand-off risk during a difficult project, but it can also leave recipes, change parts, source files, alarm logic, qualification evidence and troubleshooting knowledge unusually dependent on one supplier.
- In regulated production, nominal speed is a weak measure of capacity. Accepted output after changeovers, rejects, interventions, cleaning, calibration, deviation investigation and controlled software change is the measure that matters.
- A buyer should therefore procure a recoverable operating system around the machine: documented acceptance criteria, tested backups, editable source, spare-part strategy, cybersecurity controls, change-impact rules and a credible route to independent support.
Fifteen parts per minute, and the real capacity question
Start with a compact filling cell. In a public application note, R+D describes a semi-automatic aseptic syringe filler designed for three-, five- and ten-millilitre syringes at up to 15 parts per minute. Its disclosed features include selectable recipes and fill levels, quick-change tooling, dial indexing, six-axis cleanroom robots, a peristaltic pump, vision confirmation, environmental monitoring, and data collection and reporting. Each element sounds like a discrete capability. In operation, they form a chain in which a single weak link can turn apparent speed into unusable output.
A recipe must call the correct motion and fill parameters. Tooling must locate each syringe repeatedly after a size change. Tubing and pump behaviour must deliver the intended volume. A robot must present the component without compromising the clean operation. Vision must distinguish a satisfactory fill from a fault without rejecting good units or accepting bad ones. Environmental and production records must remain attributable to the relevant run. An operator must understand an alarm well enough to intervene correctly. The line is productive only when those actions produce conforming units and defensible records together.
That is why custom life-sciences automation is best understood as a capacity system rather than a machine purchase. Nameplate rate describes a favourable instant. Useful capacity subtracts time lost to product change, cleaning, calibration, planned maintenance, material variation, false rejects, jams, investigations, software changes, replacement-part delays and requalification. It also discounts units that cannot be released because their electronic or paper evidence is incomplete. The buyer is purchasing the repeatability of a physical process and the repeatability of the explanation for that process.
The distinction is regulatory as well as economic. The US Food and Drug Administration’s process-validation guidance treats validation as a lifecycle extending from process design through qualification and continued verification. It does not transfer responsibility for the process to the equipment builder. R+D can supply engineering, controls and qualification support; the regulated manufacturer still has to show that its own process remains in control.
This creates the central tension in R+D’s proposition. A builder that keeps many disciplines together can resolve interactions faster than a buyer coordinating separate mechanical, controls, vision and service firms. Yet every interaction it resolves can become knowledge that is expensive to reproduce elsewhere. The same integration that lowers execution risk before acceptance may raise dependency after acceptance. The qualification question is therefore not whether R+D can make a line move.
It is whether the engineering, software, validation, spares and service package leaves the customer able to keep that line productive through product changes and regulated scrutiny.
One operating company, three names, and a parent boundary
Identity matters here because procurement rights, warranty duties and technical custody attach to legal parties, not merely to a familiar logo. The assigned directory entity is R+D Custom Automation, Inc. The current operator’s contact page identifies R+D Custom Automation, LLC at 11052 254th Court in Trevor, Wisconsin. A 2026 Wisconsin manufacturing roll lists “R+D CUSTOM AUTOMATION INC” at that same address. That roll is useful evidence of continuity around the name and site, but it is not a corporate-status certificate and should not be treated as one.
The commercial name changed again in 2023. A company announcement distributed through PR Newswire said the business would operate as R+D Automation rather than R+D Custom Automation. The announcement also described a history beginning in 1977 and positioned the new name as part of a closer association with Krones. The evidence supports continuity between the assigned Inc. name, the current LLC wording, and the R+D Automation brand at one Wisconsin operation. It does not make those labels legally interchangeable in a contract.
The ownership boundary is clearer. Krones disclosed that it acquired 80.5 per cent of R+D Custom Automation, LLC, effective 8 November 2022. It said management retained 19.5 per cent and that Krones held an option over the remaining interest. At the transaction date, Krones described R+D as having more than 60 employees and about US$43 million of 2021 revenue; those are point-in-time transaction figures, not current staffing or revenue. Krones’s 2025 annual-report page still includes R+D Custom Automation in the group’s bottling and packaging portfolio.
For a buyer, the practical conclusion is precise. R+D is the specialist engineering and service counterparty to examine; Krones is the majority parent with a much larger international organisation. Parent scale may improve access to resources, but it does not automatically place a Krones balance sheet, service promise, cybersecurity control or spare-part obligation behind every R+D contract. A serious tender should state which entity signs, which entity owns or licenses deliverables, which service organisation responds in each territory, and whether any parent commitment is contractual.
Brand continuity is evidence of an operating bridge, not a substitute for those terms.
The integration surface R+D is selling
R+D’s history page says the business began as a tool-and-die shop in 1977, moved toward automation, and has focused on life sciences for more than three decades. It describes a 65,000-square-foot Wisconsin facility and claims that 98 per cent of a custom machine build remains in-house. The same page presents a wide scope: consulting, process development, mechanical and electrical design, software, manufacturing, installation, commissioning, qualification, preventive maintenance and continuing support. The percentage is a company assertion without a published measurement method, but the claimed span of work is the heart of the offer.
The site’s application catalogue shows why that span can matter. It names assembly and test processes for IV bags, filters, pump cassettes, syringes, inhalers and auto-injectors; filling operations; radiopharmaceutical traceability; diagnostic and lateral-flow products; vision inspection; laser processing; adhesive dispensing; radio-frequency and ultrasonic welding; feeding; servo motion; and robotics. Its pharmaceutical page adds prototypes, robotic cells, inspection and filling, and says the company works across both low-mix, high-volume and high-mix, low-volume settings. These are supplier descriptions of capability, not a disclosed installed-base census or verified performance record.
In a custom line, breadth is valuable because the interfaces are where projects fail. A nest may satisfy dimensional tolerances yet present the part poorly to a camera. A vision threshold may work until material reflectivity shifts. A servo move may meet cycle time but disturb liquid. A reject path may be mechanically reliable while its electronic count fails to reconcile with production records. A guarding change may affect reach, cleaning access or sensor placement. When the same organisation owns more of those decisions, the buyer has fewer contractual seams across which responsibility can be passed.
R+D’s recruitment material offers a rare view into how it organised that work, although the pages are historical job descriptions rather than evidence of present headcount. Its preserved careers page describes mechanical engineers developing concepts, bills of material and test plans tied to user requirements; electrical engineers handling panels, programmable logic controllers, industrial networks, robots, vision, operator interfaces, input/output checks and as-built changes; application engineers building concepts, budgets and supplier quotations; and after-market staff working on parts, service, retrofits and upgrades. That is consistent with a design-build-service organisation in which requirements are translated through several engineering disciplines.
The benefit is compression of the feedback loop. A controls engineer can challenge a sensor choice before wiring is final. A manufacturing technician can expose an inaccessible fastener before validation. A service engineer can argue for a replaceable component before the enclosure is closed. But in-house breadth does not guarantee that these conversations occur, nor that their results reach the customer in maintainable form. The procurement test is not the organisational chart.
It is whether interface decisions appear in controlled drawings, code, risk records, acceptance tests, spare lists and training materials that survive the departure of the original project team.
There is also a scaling question. A custom builder must balance bespoke engineering against repeatable practice. Too little reuse makes every line an experiment; too much reuse can force an ill-fitting platform onto a process. R+D’s own “why R+D” page says it uses proven technologies, can supplement capacity through trusted partners, supports equipment built by others, and works from prototypes to capital projects. Those are useful claims to test in diligence. A buyer should ask what is genuinely standard, what will be newly engineered, which third parties will touch the project, and which deviations from prior designs create new qualification or service risk.
From an imperfect component to an accepted unit
The most revealing R+D publication is not a polished machine description but its discussion of variation. In “All Products Must Be the Same?”, the company describes how out-of-tolerance parts, excess flash and variation can slow feeding, reduce overall equipment effectiveness after a production-lot change, disrupt pick-and-place operations and increase rejects. It argues for understanding the operating window rather than assuming every incoming component is identical. That observation goes to the core of life-sciences automation: the line has to control a distribution of real parts, not a perfect drawing.
Imagine a device assembly with a moulded housing, spring, glass container and label. The feeding system has to orient components whose surfaces and dimensions vary. Grippers must hold without damage. Motion must generate enough force to assemble without creating a hidden defect. Vision must observe features that lighting, reflectivity and camera angle can alter. Functional testing must separate process noise from genuine failure. The reject mechanism must physically segregate a failed unit and preserve count reconciliation. Each station’s tolerance consumes part of the product’s overall tolerance budget.
R+D’s published examples show that chain at different points. One auto-injector project description presents an 80-part-per-minute system that loads and threads a glass cartridge into a plastic syringe housing, using a chain conveyor, servo motion, encoded torque and vision before downstream labelling and packaging. A 2018 project announcement describes an awarded cleanroom bag-labelling cell in which vision coordinates are sent to one robot and a second robot removes accepted bags. A 2017 laser-marking announcement describes a target above 220 parts per minute, inspection and a reject station.
Those examples establish the kinds of functions R+D has publicly said it was selected to provide. They do not establish final site acceptance, sustained yield, regulatory outcome or current availability of the designs. The distinction is important. An award-stage speed target is not field performance, and a machine description is not a release record.
Buyers should request evidence at the level at which they expect to operate: comparable component tolerances, tested failure modes, achieved cycle-time distribution, false-reject behaviour, intervention rate, changeover time, cleaning access, and acceptance results under realistic material conditions.
R+D’s own automation rules reinforce several sensible principles: address risk early, retain control of parts, make operator-interface messages useful, choose an appropriate machine size, and avoid immature technology where reliability matters. The commercial question is how those principles become enforceable project decisions. “Useful alarm” should become an alarm philosophy with unique conditions, probable causes, safe recovery steps and tested messages. “Control the part” should become agreed input specifications and challenge material. “Address risk” should become traceable mitigations and test evidence.
The resulting operating window is jointly owned. The builder controls machine design and integration; the manufacturer controls product design, incoming materials, process knowledge, utilities, people and procedures. If a line struggles, attributing failure can become expensive because the causes cross that boundary. A strong contract defines test materials, permitted variation, environmental conditions, utility quality, staffing assumptions and acceptance methods before design freezes. Without that discipline, every exception can become an argument over whether the machine, the component or the process is at fault.
The buyer’s project begins before steel is cut
Custom automation starts as a translation problem. The manufacturer knows the product, critical quality attributes, demand pattern and regulatory commitments. The builder needs those facts expressed as requirements that can drive mechanisms, controls and tests. R+D’s historical job descriptions say its mechanical engineers prepare test plans linked to user requirements and its application engineers develop concepts, budgets and supplier quotations. That points to the right sequence, but a buyer should insist on seeing the actual trace from requirement to design choice to verification.
The first procurement package should distinguish needs from proposed solutions. “Maintain container integrity” is a need; “use a six-axis robot” is one possible solution. “Prevent an unapproved recipe from running” is a need; a particular operator-interface arrangement is a solution. Preserving that distinction lets bidders challenge unnecessary complexity and prevents an early concept from becoming an unexamined requirement. It also makes later change control more intelligible: a different component can be evaluated against the purpose of a requirement rather than merely against the old hardware.
An effective front-end study should close the largest process risks with physical evidence. Can flexible material be separated reliably? Does adhesive cure within the available cycle? Can vision see a defect across supplier lots? Does a weld window remain stable at environmental extremes? Can a filled container tolerate acceleration? What happens when the line stops mid-cycle? R+D advertises consulting and process-development work, which gives a buyer a place to locate those experiments. The deliverable should be more than a successful demonstration: raw results, failed trials, parameter ranges, sample provenance and a clear account of what remains uncertain.
Design reviews should then examine the complete use cycle. Operators need safe, comprehensible recovery. Maintenance teams need access without disturbing calibrated assemblies. Quality staff need traceable results and controlled overrides. Information-technology and security teams need an inventory of connected assets and supported software. Cleaning teams need compatible materials and reachable surfaces. Production planners need credible changeover assumptions. Validation staff need stable requirements and testable functions.
A machine can be technically elegant and still fail as an operating system if any one of these users arrives too late.
Acceptance should be split without leaving a gap. Factory acceptance can expose design and assembly faults where the builder has tools and staff. Site acceptance can confirm utilities, upstream and downstream interfaces, environmental conditions, materials, data flows and operator execution in the real facility. Installation and operational qualification can reuse evidence only when its provenance, configuration and applicability are clear. Performance qualification belongs to the manufacturer’s process and cannot be reduced to a builder’s dry cycle.
R+D’s after-market page says it provides installation and site-acceptance support, training, preventive maintenance and emergency service. Those services can preserve continuity from build to operation. They can also obscure whether the customer has acquired enough knowledge to stand alone. A well-run handover should deliberately test independence: have the customer restore backups, execute a changeover, diagnose representative faults, replace selected wear parts, reconcile reject counts, and retrieve the evidence needed for a deviation. Training attendance is not the same as demonstrated competence.
Validation is a production system, not a document hand-off
R+D’s home page says its equipment is ISO 9001 certified, UL 508 compliant, and supplied with documents based on GAMP 5 that customers can use for FDA validation. The company’s exact wording is available on its main site. These claims require careful parsing. They can indicate useful quality and engineering practices, but none relieves a regulated manufacturer of demonstrating that its particular process, configuration and use are suitable.
ISO’s explanation of ISO 9001 says the standard addresses an organisation’s quality-management system, that certification is voluntary, and that independent certification bodies—not ISO—perform certification. A buyer should therefore request the certificate, issuing body, scope, covered site and expiry rather than infer product approval from a website phrase. Similarly, UL’s industrial-control-panel programme describes training, qualified staff and marks for covered panels under the relevant programme. Panel conformity should not be casually expanded into a claim that an entire automated process is certified for its intended pharmaceutical use.
The regulatory burden reaches deeper. 21 CFR 211.68 requires appropriate checks under a written programme for automatic, mechanical and electronic equipment, limits changes to authorised personnel, calls for input/output accuracy and requires suitable backup arrangements. 21 CFR 211.100 requires written production and process-control procedures, review and approval of changes, and recording and justification of deviations. These duties turn engineering details into recurring operational controls.
Electronic records create another layer. The FDA’s Part 11 guidance explains the agency’s narrower enforcement approach while preserving applicable predicate-rule obligations. It points to authorised access, operational and authority checks, training, documentation controls and signatures where relevant. The FDA’s data-integrity guidance emphasises complete, consistent and accurate records, attribution, contemporaneous recording, original or true-copy evidence, accuracy, audit trails and associated context. A machine that measures correctly but cannot preserve intelligible evidence can still create release risk.
For medical-device manufacturers, the timing is especially relevant. The FDA’s Quality Management System Regulation page says the QMSR took effect on 2 February 2026 and incorporates ISO 13485:2016 by reference, with an increased emphasis on risk management across the product lifecycle. The agency’s QMSR frequently asked questions say inspectors may review management-review, quality-audit and supplier-audit reports under the new framework. Equipment evidence therefore sits within a broader quality system that must withstand inspection, not in a sealed validation binder.
For aseptic operations, the FDA’s aseptic-processing guidance addresses the suitability of facilities and equipment, process validation and quality control. European operations also work within EudraLex Volume 4, whose page records full applicability of revised Annex 1 from 25 August 2024. The exact validation plan depends on the process, jurisdiction and intended use, but the direction is common: contamination control and process evidence must be designed into operation.
The best supplier package therefore behaves like a maintained map. User requirements trace to functional and design specifications. Risks trace to controls. Controls trace to tests. Software and hardware versions trace to evidence. Deviations trace to investigation and resolution. Drawings, component lists and code reflect the accepted configuration. Calibration and maintenance tasks trace to critical functions. Changes identify the evidence that must be repeated. If the map is handed over as static paperwork while the machine continues to change, its value decays immediately.
This is where an integrated builder can reduce risk. The same organisation that designed a station can write a more informed challenge test and explain the failure response. But the customer must own the approved requirements, risk acceptance and validation conclusion. Supplier authorship can accelerate evidence creation; supplier custody of the only intelligible evidence is a dependency. Procurement should specify editable formats, review rights, defect correction, version identification and delivery at defined milestones, not just a documentation quantity at the end.
Software becomes the memory of the machine
In a modern custom line, software contains much of the process memory. Programmable logic defines sequences and interlocks. Robot programmes encode paths and recovery positions. Vision files preserve lighting, regions of interest and decision thresholds. Motion settings determine force, speed and timing. Recipes group product-specific parameters. The operator interface mediates access, alarms and manual actions. A supervisory layer may collect results, while networked devices carry their own firmware and configuration.
R+D’s public material establishes that this layer is material to its work. Its syringe-filling description names recipe selection, data collection and reporting, environmental monitoring, vision and robots. Its historical electrical-engineering description names programmable controllers, industrial networks, robots, vision, operator interfaces, input/output checkout, sensor challenges and as-built changes. Its after-market offering advertises controller and operator-interface updates or replacement. Together, these sources show that software is not an accessory to R+D’s machines; it is part of their operating and service surface.
That surface has at least four control problems. First is configuration: which exact versions and parameters created the accepted state? Second is access: who can view, edit, approve and deploy changes? Third is recoverability: can the site restore a known-good state after device failure or corruption? Fourth is obsolescence: what happens when an operating system, engineering tool, controller, camera or communications component is no longer supported?
The public R+D pages examined here do not disclose standard terms for source-file ownership, engineering-tool licences, password custody, code escrow, version-control practice, software bills of material, vulnerability handling or backup restoration. That absence is not evidence of poor practice. It means those protections cannot be assumed from marketing material and must be made explicit in diligence and contract schedules.
A buyer should request a software-deliverables register before purchase. It should cover editable controller, robot, motion, safety and vision source; compiled runtime files; device firmware; operator-interface projects; recipe definitions; configuration exports; licence requirements; engineering-tool versions; user and service accounts; network settings; checksums; and restoration instructions. The accepted release should be copied into customer-controlled storage and restored on representative spare or test hardware. A backup that has never been restored is only a hope.
Change control should separate parameter adjustment from programme modification without pretending that parameters are harmless. A fill volume, torque limit, vision threshold or reject delay can affect product quality even if no code changes. Each adjustable value needs a permitted range, unit, rationale, access level, audit behaviour and impact rule. Critical values should not live only in screenshots or an engineer’s memory. If recipes can be copied or imported, the procedure should verify identity, completeness, compatibility and approval before execution.
The commercial lock-in does not arise simply because a particular controller brand is used. Widely available hardware can still be difficult to support when comments are sparse, custom libraries are undocumented, safety signatures are unavailable, licences sit with the supplier, passwords are unknown, or as-built files lag behind the machine. Conversely, a proprietary element can be manageable if interfaces, recovery, spares and long-term rights are strong. The right question is not “Is it standard?” but “Can another competent party safely understand, restore and change it under controlled conditions?”
Product change is where flexibility becomes dependency
Life-sciences manufacturers often buy flexibility because the product will not remain fixed. Volumes change, container suppliers revise designs, a device gains a presentation, a label acquires new content, or a process moves from clinical to commercial scale. R+D advertises both high-mix and high-volume capability on its pharmaceutical page and quick-change features in its syringe-filling example. The value of that flexibility depends on what a change actually requires after acceptance.
Some changes are anticipated: exchange a defined nest, select an approved recipe, confirm sensors and run a documented challenge. Others alter the operating window: a new plastic resin changes friction, a container dimension shifts, a label reflects light differently, or a higher output target shortens process time. Still others modify critical architecture: a new vision camera, robot, controller or record interface. Treating all three as “changeover” conceals radically different cost and validation exposure.
The first procurement defence is a change matrix. For every product family, it should identify common tooling, dedicated tooling, recipes, expected duration, required skills, cleaning steps, calibration checks, line-clearance controls, first-piece verification and evidence required before release. Tooling needs unique identity and controlled storage. Mechanical interfaces need mistake-proofing where practical. Software should prevent an incompatible recipe and tooling combination from running. Changeover time should be demonstrated with normal operators and realistic line clearance, not an expert team working on a clean machine.
The second defence is modularity with evidence. A station designed around clear mechanical, electrical, software and data boundaries is easier to replace or extend. Yet “modular” has little value if every module shares hidden timing assumptions or a single undocumented programme structure. Buyers should ask R+D to identify interfaces, capacity margins, reserved inputs and outputs, physical space, network addresses, data ownership and failure propagation. They should also ask which changes the original risk assessment anticipated and which force a new concept study.
The third defence is commercial. Rates and response times for later engineering should be agreed while the original project is competitive. The customer should have rights to use delivered engineering material for maintenance, modification and requalification, subject to clearly defined supplier intellectual property. Critical custom parts should have drawings, materials, tolerances and approved manufacturing routes. If R+D alone can reproduce them, the stocking strategy should reflect lead time and failure consequence.
R+D’s claim that it can retrofit its own and other manufacturers’ equipment suggests an organisation comfortable entering existing systems. That can be useful for lifecycle extension, but it also supplies a useful reciprocity test: would the original R+D package enable another qualified integrator to enter one of its lines with comparable clarity? A buyer does not need to plan an immediate supplier exit. It needs a credible alternative so that a future product change is a controlled engineering decision rather than a hostage negotiation.
Service, spares, and the economics of a stopped line
Once a line enters production, the economic unit changes from project milestone to minute of recoverable capacity. A failed sensor may cost little while the investigation, controlled replacement, testing and lost output cost far more. A custom machined nest may be simple but unavailable for weeks. A camera may be in stock yet require an obsolete project file. An emergency intervention may restart motion but create a documentation gap that delays release.
R+D’s after-market page advertises a two-year warranty, on-site and remote support, installation and site-acceptance help, customer training, replacement and spare parts, preventive maintenance, emergency service, upgrades and retrofits. It says the company can support equipment built by others, offers operator training at its Wisconsin facility, and proposes service contracts beginning at twelve-month intervals and then moving to six-month intervals. These are advertised service features; the public page does not provide warranty terms, exclusions, response commitments, geographic coverage, labour rates or uptime guarantees.
A buyer should convert those headings into an operating plan. The criticality analysis starts with functions, not component price. What single failure stops production? What failure can silently affect quality? What item is unique, repairable only off-site, approaching end of support or subject to a long supplier lead time? What can be bypassed safely, and under whose approval? The answer should drive on-site spares, supplier-held inventory, repair arrangements and preventive replacement.
The spare-parts package needs more than manufacturer part numbers. It should identify installed firmware and configuration, approved substitutes, calibration needs, shelf-life or storage conditions, special tools, replacement instructions and post-replacement tests. Custom fabricated parts need controlled drawings and inspection requirements. Commercial components need an obsolescence-monitoring owner. If an approved substitute changes form, fit, function or software behaviour, the quality system needs an impact route before the old part becomes unavailable.
Preventive maintenance should likewise be connected to risk. A calendar interval may be convenient, but usage, environment, wear and failure history may support different frequencies. Maintenance tasks should state the accepted condition, measurement method, tools, parts, data to record and escalation threshold. The line should distinguish routine restoration from a change that affects the validated state. Service technicians need a controlled way to record what they found and altered, including temporary measures.
Training is another capacity control. Operator knowledge tends to narrow toward common faults; rare recovery steps fade. Maintenance skill can disappear with staff turnover. The most useful programme is role-specific and practical: safe stop and restart, alarm interpretation, jam recovery, changeover, inspection challenge, backup retrieval, part replacement, calibration and escalation. Refresher training can use recurring faults and near misses. The customer should retain materials that match the current configuration, rather than depend on the availability of the original R+D engineer.
There is a strong case for keeping the builder involved. The builder understands design intent and can see patterns across similar mechanisms. The risk begins when involvement is the only route to recovery. Service quality should therefore be measured by response, first-time resolution, repeat failure, documentation closure, remote-access discipline and the customer’s growing capability—not by the number of visits alone.
Remote support opens a security contract
Remote support can shorten a stoppage dramatically. It can also create a path into the controls that govern physical motion, electronic records and product decisions. R+D publicly offers remote and on-site support, but the examined public pages do not state how remote sessions are authenticated, approved, segmented, logged, recorded, time-limited or revoked. Nor do they disclose a vulnerability-reporting channel, standard patch process or supported cybersecurity architecture. Those are open diligence questions, not findings of an exposed system.
The correct reference point is operational technology, not ordinary office computing. NIST Special Publication 800-82 Revision 3 addresses operational-technology security while recognising performance, reliability and safety constraints. The line may not tolerate an untested update or an abrupt security scan. At the same time, uninterrupted legacy operation is not a security strategy. Controls have to fit the process and be tested against it.
A 2025 joint procurement guide led by CISA tells asset owners to make security part of buying operational technology. It highlights recurring weaknesses such as weak authentication, known vulnerabilities, inadequate logging, insecure defaults and passwords, and legacy protocols. The guide is not an assessment of R+D equipment. It is a useful checklist precisely because a custom line combines components from multiple suppliers, each with its own lifecycle and default behaviour.
The ISA/IEC 62443 series overview divides responsibilities across asset owners, service providers, systems and components and addresses security across the lifecycle, including patch management. That division maps well to an R+D project. The customer controls the site and risk acceptance. R+D integrates the system and may service it. Controller, robot, vision and computing vendors control components and updates. A contract should assign who monitors each layer and who acts when one vendor discloses a weakness.
Remote access should be customer initiated or explicitly approved, unique to an identified person, strongly authenticated, restricted to the required assets and functions, and disabled when the work ends. Sessions should pass through a managed access point rather than an uncontrolled direct connection. Logs need sufficient time, identity and activity context for investigation. File transfer should be controlled and scanned in a manner compatible with the environment. Emergency access needs its own tested procedure; an emergency should not legitimise a permanent shared credential.
The software inventory discussed earlier becomes the foundation for vulnerability response. The customer cannot assess a notice if it does not know which controller firmware, industrial computer, operating system, remote-access utility, camera package and library is installed. It also needs a safe test path. Patching production first can damage availability; never patching preserves known exposure. A representative test environment, supplier assessment, backup, rollback plan and defined approval route turn that tension into managed change.
The FDA’s 2026 guidance on computer software assurance for production and quality-system software supports a risk-based approach tied to intended use and foreseeable failure, while addressing security, data integrity, storage, transfer and operating error. It does not require every software function to receive identical effort. That is commercially important: a risk-based programme can focus evidence on functions that affect product, records, safety and recovery rather than generating undifferentiated paperwork.
Cybersecurity therefore belongs in the machine acceptance package. The buyer should test account roles, failed access, session termination, time synchronisation, logging, backup restoration, secure transfer and recovery from representative component failure. It should receive a supported-lifecycle statement and notification route. Remote support is valuable when these controls make it auditable and revocable. Without them, faster troubleshooting is purchased with an undefined access path.
Pricing is hidden; total dependency is not
R+D does not publish a price list for custom lines or service on the public pages examined here. Its historical application-engineering description refers to concepts, budgets, supplier quotations and requests for quotation, which is consistent with project-specific pricing. Its after-market page lists service categories without public rates or response commitments. Any precise claim about R+D’s margin, hourly rate or comparative price would therefore outrun the available evidence.
The relevant commercial calculation is total cost of controlled capacity. The purchase price is only the first term. The customer also funds internal process development, samples, facilities, utilities, project management, validation, training, spare inventory, planned maintenance, licences, security controls, change engineering, requalification and eventual obsolescence. Delay has a cost; so does a rejected production lot, a slow changeover, an extended deviation or a product launch that waits for equipment.
An integrated builder can lower some of those costs by reducing coordination. Fewer engineering seams may mean faster fault resolution, a more coherent qualification package and less duplicate management. R+D’s broad in-house claim is economically meaningful if it leads to fewer redesign cycles and clearer accountability. It is not meaningful merely because payroll boundaries are different. Buyers should compare deliverable completeness, residual risk and lifecycle obligations, not just quoted capital.
Commercial structure also shapes engineering behaviour. A fixed-price contract can discipline scope but encourage disputes over uncertain process work. Time-and-materials work can accommodate discovery but transfer efficiency risk to the buyer. Milestone payments can support progress, provided milestones measure evidence rather than elapsed time. Holdbacks tied to final documentation, software delivery, unresolved defects and site performance can preserve leverage when the machine has already shipped.
Lifecycle pricing should be exposed early. Ask for agreed categories for travel, emergency response, remote support, future engineering, replacement fabrication, software upgrades and obsolescence work. Ask which licences recur, who may renew them and what happens if a product is discontinued. Ask whether the two-year warranty advertised by R+D begins at shipment, factory acceptance, site acceptance or production use, and what pauses or voids it. The public wording alone does not answer.
Switching cost is not inherently bad. A high-performing specialist can deserve a durable supplier tie. The danger is unpriced, unmeasured switching cost that appears only during a failure or product change. A buyer should know which capabilities it is deliberately retaining from R+D and which it is acquiring for itself. That decision turns dependency from an accident into a governed sourcing choice.
The procurement test that matters
A strong request for proposal should force the line’s future operating life into the present negotiation. The following tests are more revealing than a generic capability presentation.
Identity and responsibility. Require the precise contracting entity, manufacturing site, parent role and service territories. Clarify whether references to R+D Automation, R+D Custom Automation, LLC and the historical Inc. name point to the same proposed counterparty for each obligation. Identify every material subcontractor and the work it will perform. If Krones resources are part of the offer, state which resources and make access contractual.
Comparable process evidence. Ask for examples that match the product’s handling difficulty, speed, cleanliness, inspection and record requirements—not merely the same market label. R+D’s published examples range from robotic syringe filling to auto-injector assembly and vision-guided cleanroom handling. The buyer should request acceptance evidence or customer references under confidentiality where public material stops at description or award.
Operating window. Define the component and environmental variation that acceptance will challenge. Use multiple production lots, boundary samples and known defects. Agree how false accepts, false rejects, jams, micro-stops, interventions, scrap and rework will be counted. Measure sustained accepted output, not a short peak. Include planned stops and restart behaviour. State whether upstream starvation or downstream blockage counts and how it is simulated.
Failure and recovery. Test loss of power, communications, air, vacuum, a sensor, a camera and representative actuators where safe. Confirm state preservation, safe motion, product disposition, restart logic and records. Challenge every critical alarm and interlock. Require messages to identify the condition and safe next action. Demonstrate recovery with customer staff rather than only builder experts.
Quality evidence. Trace every critical requirement to a risk control and an objective test. Define calibration and measurement-system expectations. Confirm reject segregation and count reconciliation. Verify that manual interventions, overrides and reprocessing cannot bypass required evidence. Tie each delivered software and hardware version to the test package. Close deviations through assessed corrections, not unexplained retesting.
Software custody. Inventory editable source, runtime files, firmware, recipes, vision configurations, licences, accounts, network settings and backups. Demonstrate a customer-controlled restoration. Identify supplier-owned libraries and the customer’s rights if R+D becomes unavailable or declines future work. Define comments, naming and change-history expectations. Require final as-built delivery after site changes.
Data integrity. Map each record from creation to review, transfer, storage, backup, retrieval and retention. Establish time synchronisation and unique identity. Test authority limits and audit behaviour. Determine what happens during network loss, storage exhaustion or interrupted transfer. Confirm which system is authoritative when machine counts and enterprise records disagree. Align the design with applicable FDA electronic-record guidance rather than treating a generic “Part 11 ready” phrase as proof.
Security. Require an asset and software inventory, architecture diagram, port and protocol list, account design, hardening record, remote-access method, logging, backup, patch and vulnerability processes, support lifetime and incident contacts. Test revocation and recovery. Use NIST’s operational-technology guidance and the CISA procurement considerations as question sets, not as decorative compliance labels.
Spares and obsolescence. Rank components by production and quality consequence. Require recommended quantities, lead times, shelf conditions, approved substitutes and custom-part drawings. Identify items already late in their lifecycle. Define notification periods and last-buy support. Demonstrate replacement and post-maintenance testing for selected critical parts.
Change readiness. Run a representative product change during acceptance. Measure line clearance, physical tooling, recipe selection, verification and first accepted output. Ask for an impact assessment of a plausible future component or format. The response reveals whether flexibility is designed into interfaces or depends on another engineering project.
Service outcomes. Turn “support” into severity definitions, response and escalation paths, availability by time zone, secure remote method, site-arrival expectations, documentation closure and spare dispatch. Confirm what is included in warranty and service contracts. Ask how support transfers if the original project staff leave. Measure recurring faults and time to permanent correction, not only initial response.
Exit and continuity. Request the information and rights needed for a qualified third party to maintain and modify the line. That need not disclose unrelated R+D know-how. It should prevent loss of customer-specific configurations, drawings, test evidence and recovery ability. Define what is delivered at each payment milestone so continuity does not depend on a final handover after leverage has disappeared.
This diligence is not an argument against custom automation. It is how a buyer captures its value. If R+D can answer these tests with coherent artefacts and demonstrated behaviour, its integrated approach becomes a measurable reduction in integration risk. If answers rely on future understanding, unnamed standard practice or the continued presence of one engineer, the same approach becomes concentrated dependency.
A crowded field with different answers to standardisation
R+D competes not against one uniform category but against several ways of buying automation. ATS Life Sciences presents broad life-sciences automation, configurable platforms and lifecycle services including remote and on-site support, parts, retooling and relocation. Mikron Automation presents systems spanning laboratory, pre-production and high-volume assembly, with validation documentation and international support. These firms can be compared with R+D where a programme centres on assembly, test and integrated handling, but their public breadth does not establish better fit for a particular process.
Filling specialists represent another route. groninger describes equipment across research, clinical and commercial production, including ready-to-use vials, cartridges and syringes. Bausch+Ströbel presents filling for ready-to-use and bulk containers, flexible systems, containment, secondary packaging and lifecycle services. Where aseptic filling is the dominant risk, a buyer may value a supplier’s deeper standard platform and process history over a broader custom-integration offer.
The choice should not be reduced to company size. It is a choice among risk allocations. A standard platform can bring repeatable design and a wider service base but constrain unusual products. A custom specialist can fit the exact workflow but create more unique parts and code. A large integrator can carry a complex programme across sites but impose heavier governance and cost. A filling specialist can own the critical process core while leaving peripheral integration to others.
R+D’s position appears strongest where the differentiating problem lies at the interfaces: unusual component handling, assembly, inspection, filling and packaging combined into one workflow; a need for process development before final design; or an existing line that needs modification. Its application breadth and retrofit claim support that interpretation. It is still an inference from supplier material, not a public win-rate or performance comparison.
A disciplined shortlist should give each bidder the same boundary cases and lifecycle questions. Compare how much is standard, how much is new, who owns each interface, what is demonstrated, which evidence is delivered, how failures are supported and how changes are priced. Reference visits should focus on machines several years into operation, after staff turnover and at least one product or component change. A beautiful factory-acceptance demonstration cannot answer the lifecycle question.
Competition also gives the buyer leverage to unbundle choices. A specialist filling core might be integrated by R+D; an R+D-built assembly line might use widely supported component families; service may combine supplier escalation with trained site capability. The best architecture can deliberately place specialism where it creates value and portability where it limits risk. The procurement goal is not the fewest suppliers at any cost. It is the fewest unmanaged interfaces.
The public incident record is thin, so diligence must be thick
The public material examined for this assessment does not contain an independently documented R+D machine failure, cybersecurity incident, regulatory enforcement action tied to an R+D line, or customer acceptance dispute. It also does not contain a public installed-base list, field-availability series, repeat-order rate, false-reject record, response-time performance, vulnerability disclosure record or customer-satisfaction series. These are evidence gaps, not proof that the underlying outcomes are good or bad.
One public customer-related case concerns pandemic capacity. In July 2020, R+D announced an agreement with SiO2 Materials Science to build automation for coated plastic vial production. A separate US Department of Defense release confirms a US$143 million federal award to SiO2 and a plan to add capacity for 120 million vials annually. The government release names SiO2, not R+D; the bridge to R+D comes from the company announcement. The pair supports R+D’s participation claim and the scale of the customer programme, but not the equipment’s eventual acceptance, output or quality.
The absence of public operational detail is understandable in confidential pharmaceutical and medical-device work. It still changes how a buyer should investigate. Evidence must move into a controlled diligence room: anonymised acceptance results, reference calls, quality metrics, repeat failure patterns, service records, qualification examples, security procedures and proof of insurance. References should be selected for technical comparability and machine age, not just customer prestige.
Negative evidence also needs discipline. No claim about social-network staffing, employee reviews, complaint sites, domain routing, hosting geography or upstream network ownership is included here because the examined evidence set did not establish a relevant, sufficiently bounded fact. Such signals can occasionally guide inquiry, but they should not be transformed into operating conclusions without a verifiable bridge. For this company, direct project, regulatory, ownership and procurement evidence is more probative.
What Krones changes—and what it does not
Krones’s acquisition changed R+D’s resource context. The 2022 transaction release says Krones bought 80.5 per cent to expand in pharmaceuticals and biopharmaceuticals. R+D’s 2023 name-change announcement says association with Krones creates possibilities for international service and support. Those statements support strategic intent. They do not disclose an R+D-specific global response map, parts network, common engineering process or service-level commitment.
Krones’s scale is nevertheless material. Its 2025 annual-report page reports group revenue of €5.6638 billion and year-end orders on hand of €4.1904 billion, while continuing to list R+D in the group portfolio. Those are Krones group figures and must not be attributed to R+D. They show the parent is operating at a scale far beyond the specialist it acquired.
For R+D, a larger parent can potentially widen purchasing power, recruitment reach, geographic support and access to adjacent packaging expertise. It can also introduce portfolio priorities, common component preferences or organisational hand-offs that differ from the specialist pattern a customer expected. Neither outcome should be presumed. Buyers should ask for concrete evidence: which Krones locations can service an R+D line, what training they receive, which spare inventories they access, how escalation crosses the organisations, and whether R+D engineering remains accountable after deployment.
The ownership structure also makes continuity questions more important, not less. Contracts should address assignment, changes in control, intellectual-property licences, access to customer files, warranty responsibility and service continuity. If a parent support promise matters to the award, it belongs in the signed package. If the buyer values R+D’s specialist autonomy, it should identify the named functions and governance that preserve it.
Krones ownership therefore strengthens the option set without answering the operating question. The decisive evidence remains line-specific: acceptance, documentation, recoverability, service and change performance. Corporate scale can support those outcomes; it cannot replace them.
The watchlist for the next operating cycle
The first watchpoint is whether R+D turns the R+D Automation brand into observable service reach. The company has claimed that Krones association creates international possibilities, but the public material examined here does not set out coverage by country, response time or trained location. Evidence of cross-trained technicians, regional spare holdings and completed international interventions would make the parent relationship operational rather than rhetorical.
The second is standardisation without loss of fit. R+D’s attraction is custom integration, while customers need predictable delivery and support. Useful disclosures would include named repeatable modules, supported component families, software and security baselines, interface standards and obsolescence plans. Standardisation should reduce unique lifecycle burden while leaving process-critical elements genuinely tailored.
The third is software custody. Buyers should look for clearer standard deliverables around editable source, version history, restoration, accounts, licence continuity, inventory and vulnerability notification. These can be handled confidentially; they do not require public release of customer code. Consistency across projects would make a major source of lifecycle risk easier to price.
The fourth is validation efficiency under the FDA’s 2026 framework. The computer-software-assurance guidance encourages risk-based assurance for relevant production and quality software, and the QMSR embeds risk management more deeply in the device quality system. R+D can create value if its requirements, risk and test structure helps customers focus effort on high-consequence functions while keeping configuration evidence exact.
The fifth is cybersecurity as a service lifecycle. A one-time hardening exercise will age as component vulnerabilities and remote-support tools change. Watch for contractual notification, supported-lifetime statements, tested patch and rollback practice, unique service identity, customer-controlled access and an accurate installed-software inventory. The relevant signal is not a badge; it is the speed and discipline with which an affected site can determine exposure and act safely.
The sixth is the after-market feedback loop. R+D advertises preventive maintenance, replacement parts, retrofits and support. The strategic test is whether service findings alter new design, spare recommendations, alarm quality and obsolescence planning. A builder with decades of field exposure should be able to convert recurring faults into engineering standards. Buyers can probe this through change histories and examples of design lessons, even where customer details remain confidential.
The seventh is demonstrated change performance. Case material often celebrates initial speed, but regulated capacity is revealed by the second product, a new supplier lot, an obsolete component and the first major software change. References that document those transitions would be more valuable than another launch announcement. They would show whether R+D’s integrated structure creates durable flexibility or merely relocates future engineering work to the original supplier.
Finally, watch the boundary between R+D and Krones. Continued inclusion in the parent portfolio is evidence of ownership continuity, not of how engineering authority is evolving. Customers should monitor legal counterparty wording, leadership, service routes, product positioning and any transfer of functions. None is necessarily adverse. Each can change where knowledge and accountability sit.
Verdict: buy controlled change, not just motion
R+D’s public record supports a credible identity as a Wisconsin builder of custom life-sciences automation with mechanical, electrical, software, manufacturing, qualification and after-market ambitions under one roof. Its disclosed applications span assembly, filling, inspection, robotics, testing and packaging. Krones acquired a majority interest in 2022 and continues to include the business in its portfolio. The record is much thinner on independently verified operating outcomes, current service performance, software custody, cybersecurity practice, pricing and customer acceptance.
That combination makes R+D neither an obvious low-risk choice nor an opaque one. Its integrated structure addresses a real source of project failure: interfaces among parts, process, motion, vision, records, people and validation. It can be especially valuable when the process itself needs development and when a clean division among equipment packages would be artificial. The same structure concentrates design intent and troubleshooting knowledge, so the customer has to procure transferability deliberately.
The governing metric should be accepted, explainable output over the line’s life. A machine that reaches a high instantaneous rate but requires frequent expert intervention, fragile changeovers or undocumented recovery has low effective capacity. A slower line with a stable operating window, rapid controlled recovery, strong evidence and predictable product change may create more saleable output. Validation, change control and exception handling are therefore not overhead around production. In regulated manufacturing, they are production capacity.
A buyer should award R+D only after the future line has been rehearsed on paper and in tests: component variation, a representative changeover, fault recovery, record reconciliation, backup restoration, remote service, a critical spare replacement and a plausible obsolescence case. It should receive controlled source, drawings, configuration, qualification evidence and rights sufficient for continuity. It should define where R+D remains the preferred expert and where the site can act independently.
If those terms are strong, R+D’s design-build-service concentration can be an advantage. One team can solve the interactions that determine whether a sophisticated life-sciences process actually runs. If the terms are weak, that same concentration becomes a toll gate on every change and exception. The decisive purchase is not a robot, filler, inspection station or packaging cell. It is the ability to change a regulated process without losing control of either the product or the evidence.