Frequently Asked Questions
Answers to the most common questions about working with Liftric.
Frequently Asked Questions
We could scale — we choose not to. In regulated medical software, knowledge continuity matters more than headcount. At Liftric, you work with senior engineers who've been in regulated medical software for over a decade. The people you talk to in the first meeting are the people who build your product. That also means we're selective — if we say yes, we're fully committed.
We primarily work to IEC 62304 (software lifecycle) and ISO 14971 (risk management), embedded within the requirements of the MDR and IVDR.
Our experience covers the concrete regulatory demands of five markets: EU (MDR/IVDR), Switzerland (MepV/IvDV), UK (UKCA), India (CDSCO), and Canada (CMDR / SaMD). This means not just familiarity with each framework, but hands-on experience harmonizing and meeting the market-specific requirements for Software as a Medical Device.
Our core expertise is in diagnostic software — SaMD and IVD under MDR and IVDR. The underlying regulatory and engineering discipline (IEC 62304, ISO 14971, QMS integration) is shared across these domains. If your project sits at the intersection of diagnostics and digital therapeutics, we're happy to explore the fit in a Discovery session.
MDR covers software that is itself a medical device (SaMD) — for example, clinical decision support tools or monitoring apps. IVDR covers software used to evaluate in vitro diagnostic tests — for example, apps that read and quantify rapid test results. The regulatory pathway, classification rules, and documentation requirements differ between the two. We have hands-on experience with both and can help you determine which regulation applies to your product.
Pricing depends on scope, regulatory classification, and existing documentation maturity. Get in touch for a no-obligation introductory call — we're happy to put together a tailored proposal.
It depends on the scope. An Independent Review typically takes one to two weeks, a Regulatory Alignment four to eight weeks, and a full development engagement several months. We'll give you a realistic estimate in our first conversation.
Absolutely. We love working in smaller steps — PoC, demonstrator, MVP — always with regulatory requirements in mind. This lets you validate early, learn fast, and grow from a solid foundation.
One or two focused sessions, typically half a day each. Your team and ours sit down together and work through the fundamentals: stakeholders, use cases, regulatory classification, system architecture, and risk landscape. You walk away with a shared understanding and a concrete path forward — before any production code is written.
Every change to a certified medical device — whether a bug fix, a new feature, or an updated dependency — must be assessed for its regulatory impact. Under IEC 62304, changes go through a defined change management process: impact analysis, risk reassessment, verification, and documentation update. Not every change triggers a new submission, but every change must be traceable. This is exactly what our After Go-Live service covers.
Yes — that's exactly what our Regulatory Alignment package is for. We assess your current state and bring both documentation and processes up to the required standard.
Both. If you have an established QMS (e.g., based on ISO 13485), we integrate into your existing processes, tools, and templates. If you don't have a QMS yet — or if it's not set up for software — we can provide the structure and tooling needed to get your documentation audit-ready. Either way, the goal is the same: one coherent system, not parallel documentation tracks.
IEC 62304 calls these SOUP — Software of Unknown Provenance. Every third-party component must be identified, risk-assessed, and verified in the context of your device. In modern software development, where a single project can depend on hundreds of packages, this is one of the most time-consuming regulatory activities. We maintain tooling and processes to manage SOUP systematically — from initial assessment through ongoing vulnerability monitoring.
No. We're engineers who focus on the product and the desired outcome — not a particular technology. Whether it's Kotlin Multiplatform, Swift, React, Python, or cloud-native: we work with different stacks and pick the one that fits your project best.
Yes. We use AI tools to accelerate development — from code generation to documentation drafting. Every AI-assisted output goes through the same verification and review process as manually written work. Full traceability is maintained. We make the decision about AI tooling together with you — aligned with the confidentiality requirements of your project. Depending on your needs, we can omit AI tooling entirely or work exclusively with local LLMs.
AI/ML algorithms may affect the risk classification of your device under MDR/IVDR, and the EU AI Act introduces additional requirements for AI systems in medical contexts. We have hands-on experience integrating on-device AI into certified IVD products — from model training through classification implications to regulatory documentation. We treat the AI Act the same way we treat IEC 62304: as a framework to build into the process from the architecture phase — so there are no surprises at submission.
Yes. We prepare the technical documentation, train your team on what to expect, and can participate directly in audit sessions to answer technical questions about the software lifecycle, risk management, and architecture decisions. Whether it's a first-time submission or a surveillance audit, we've been through the process with multiple notified bodies including TÜV Rheinland.
Flowify AI
Flowify AI outputs measurable signal values with configurable thresholds — auditable and traceable. We deliver IEC 62304 software documentation, risk management documentation (ISO 14971), ML documentation (EU AI Act), and cybersecurity documentation. Published validation studies with transparent methodology provide the evidence base. Optional debug data collection gives you per-evaluation traceability for field validation and post-market surveillance.
The AI is trained on 1.2M+ images captured across diverse devices, lighting conditions, and backgrounds. It is validated across 32+ smartphones (Android & iOS). Rather than requiring manual calibration per device — as traditional image processing approaches do — the AI learns to generalize across camera sensors, white balance differences, and colour temperature from data.
Yes. All AI computation runs on the smartphone. No internet connection is needed to capture, evaluate, or display results. The cloud portal is optional and used for result management, analytics, and data export.
No. Flowify AI processes the live video stream, collecting a configurable series of validated frames. Frames with reflections, overexposure, or poor framing are automatically filtered out. A result is only produced when signal consistency across frames falls below a defined threshold. The median signal is taken as the final value.
Traditional machine vision requires manual calibration for each device, lighting condition, and environment. AI trained on sufficient real-world data generalizes across devices and conditions inherently. Flowify AI outputs quantified signal values — not opaque classifications — making results explainable and defensible. The model is lightweight, efficient, and runs entirely on-device.
By default, none — all processing happens on-device. Optionally, results and debug data (captured frames and inference details) can be transmitted to the cloud portal for traceability, field monitoring, and post-market surveillance. This is configurable per deployment.
Professional IVD, home and self-testing, veterinary diagnostics, food safety, environmental monitoring, and research. Flowify AI works with singleplex and multiplex tests, various colloid types (gold, latex, carbon, and others), and standard or custom cassette housings.
Yes. Flowify AI is designed to generalize across test formats — different housings, colloid types, and line configurations. For standard cassette designs, onboarding typically requires no retraining. For non-standard formats (unusual geometries, uncommon colloids, or multiplex configurations), we run an adaptation phase where we extend the model with your specific test data. This is included in the deployment pipeline.
Every Flowify AI deployment includes a full documentation package: IEC 62304 software lifecycle documentation, ISO 14971 risk management files, EU AI Act machine learning documentation, and cybersecurity documentation. The package is designed to slot directly into your technical file for MDR or IVDR submission — no additional documentation effort required on your side.
Dedicated readers offer controlled optics but come with hardware lock-in, per-device costs, and supply chain dependencies. Flowify AI turns any modern smartphone into a clinical-grade reader — no additional hardware, no per-device licensing, no inventory management. Our inter-device coefficient of variation is below 10% across 32 tested smartphones, and quantitative accuracy (R² 98.7%) is comparable to dedicated readers at a fraction of the cost.
Book a free introductory call — we'll discuss your project and outline actionable next steps.
Book a meeting
