Question 1

What is rmForge?

Accepted Answer

rmForge is a SaaS platform that generates a review ready medical-device Hazard Analysis (HA) documentation from your inputs. Answer a short device-classification wizard, or upload your IFU and product requirements, , and rmForge produces a Hazard Analysis workbook plus a customer acknowledgment form to support your design history file.

Question 2

Who is rmForge for?

Accepted Answer

Hazard Analysis consumes engineer hours, drifts between projects and review sessions, and is one of the artifacts FDA and Notified Body reviewers most often return for inconsistency, missing rows, or weak traceability. It is a project pain-point. rmForge is for medical-device manufacturers, contract design houses, and consultants across all device classes who want a structured Hazard Analysis built against a controlled and consistent ruleset, with traceability to Product Requirements.

Question 3

How do I get started?

Accepted Answer

During closed beta, request access by emailing support@rmforge.io or by clicking "Get started" from the home page. Once on the allow-list, you sign up at /signup, confirm your email, and start your first analysis from the dashboard.

Question 4

What inputs do I need to upload?

Accepted Answer

Nothing. Answer the wizard's questions and get started today with a preliminary hazard analysis. For an enhanced Hazard analysis, upload Instructions for Use (IFU) PDF, Product Requirements document, User Needs and Intended Use document, and any existing Hazard List. The richer the inputs, the more specific the HA rows we generate.

Question 5

How long does an analysis take?

Accepted Answer

A typical, single device, Class II completes in 10 to 15 minutes from upload to download. You can close the browser and we'll email you when it's ready. Medical Electrical equipment and multiple subsystems can take an hour or more depending on the number of subsystems uploaded through documentation or the wizard.

Question 6

Can I try rmForge with a sample device?

Accepted Answer

No. The home page offers a sample Hazard Analysis for your review.

Question 7

Do I need to install anything?

Accepted Answer

No. rmForge runs entirely in the browser on rmforge.io. There's nothing to install and nothing to configure beyond a verified email address.

Question 8

Can multiple people on my team use rmForge?

Accepted Answer

During closed beta, each team member needs to be on the allow-list and signs up with their own email. Team-shared analysis workspaces are on the roadmap for general availability.

Question 9

What browsers does rmForge support?

Accepted Answer

Current versions of Chrome, Safari, Edge, and Firefox. We do not test against Internet Explorer.

Question 10

Is there a free tier?

Accepted Answer

Closed beta is free.

Question 11

What is a Hazard Analysis?

Accepted Answer

A Hazard Analysis (HA) is a systematic identification of foreseeable hazards, hazardous situations, and harms associated with a medical device, with an estimate of risk for each. It is the foundational artifact of an ISO 14971 risk management file and is required by FDA and Notified Bodies.

Question 12

How does a Hazard Analysis differ from an FMEA?

Accepted Answer

A Hazard Analysis enumerates hazards starting from the patient or user harm and traces back to the device. A Failure Mode and Effects Analysis (FMEA) starts from component or process failure modes and traces forward to effects. Both are valuable; ISO 14971 requires the HA, not the FMEA. HA is a top-down analysis while FMEA is bottom-up analysis.

Question 13

What is a hazard versus a hazardous situation versus a harm?

Accepted Answer

A hazard is a potential source of harm (for example, electrical energy). A hazardous situation is a circumstance in which the user or patient is exposed to the hazard (for example, an exposed conductor during defibrillation). A harm is the physical injury or damage to property (for example, electrical burn or an equipment fire).

Question 14

What is the difference between a Normal Condition and a Fault Condition?

Accepted Answer

A Normal Condition is the device operating as designed without component failures. A Fault Condition is a single-component failure or single-fault state that the safety analysis must consider. ISO 14971 requires both to be evaluated. rmForge enforces a strict N versus F separation: every row is one or the other, never both.

Question 15

What does the rmForge HA Agent actually do?

Accepted Answer

It reads your inputs, builds a list of enumerated hazards using a controlled vocabulary (energy, biological and chemical, performance, information,), foreseeable sequence of events, hazardous situations, and assigns severity and probability per established categorical scales, looks up Risk Level from a four-band matrix, and traces every row to risk control measures..

Question 16

How many hazard rows should I expect?

Accepted Answer

Typical Class II devices produce 30 to 120 hazard rows. Class I devices may produce fewer; Class III implantables can exceed 200. rmForge does not pad rows to hit a quota; the count is driven by what your IFU and design inputs imply.

Question 17

What is "Po" in rmForge?

Accepted Answer

Po is the combined probability of occurrence, computed by looking up p1 (probability the hazardous situation arises) and p2 (probability the situation leads to harm) on a combined-probability table. Po is a categorical band, not a numeric score.

Question 18

What is Risk Level in rmForge?

Accepted Answer

Risk Level is the categorical output of the Severity by Po lookup. It has four bands: Negligible, Acceptable, Tolerable, and Intolerable. ISO 14971 requires this banding to drive risk-control decisions.

Question 19

Why doesn't rmForge multiply Severity times Probability to get RPN?

Accepted Answer

Ordinal multiplication of severity by probability is widely critiqued in the risk literature. A row with high severity and low probability presents a different real-world risk profile than a row with low severity and high probability, even when the numeric product is identical. rmForge follows a table-based categorical lookup approach rather than producing a synthetic numeric RPN.

Question 20

What is an "intended use" versus "non-intended use" hazard?

Accepted Answer

Intended use hazards arise during the use described in the IFU. Non-intended use hazards arise from reasonably foreseeable misuse. ISO 14971 requires both. rmForge categorizes every row's use classification accordingly.

Question 21

What is the four-band Risk Level matrix?

Accepted Answer

Negligible, Acceptable, Tolerable, Intolerable. Defined as a function of Severity (S1 through S5) and Po (P1 through P5). Intolerable rows require a risk control that demonstrably reduces the residual risk.

Question 22

Does rmForge consider Essential Performance?

Accepted Answer

Yes, for medical electrical (ME) equipment per IEC 60601-1. The Essential Performance (EP) agent activates when the wizard identifies the device as ME and performs a structured preparation and assessment sequence aligned with IEC 60601-1.

Question 23

How does rmForge handle the IFU when it is incomplete?

Accepted Answer

When the analysis encounters a gap (missing clinical function declaration, missing performance limit, ambiguous Basic Safety boundary), it emits a clarification query, checkpoints the pipeline, you respond in the in-app chat, and the analysis resumes.

Question 24

Can rmForge create a Hazard List from scratch with no inputs?

Accepted Answer

No. rmForge needs at least the Wizard questions answered, which are adapted from ISO TR 24971.

Question 25

What is the sequence of events that leads to a harm?

Accepted Answer

rmForge models the sequence as up to four steps from initiating cause to a hazardous situation and then a harm. The cause is one of Design, Hardware, Software, Process, Packaging, Labeling, User. The sequence is captured in the HA row.

Question 26

What hazard categories does rmForge use?

Accepted Answer

Three top-level categories: energy, biological and chemical, and performance including information and security. Aligned with ISO 14971 Annex C examples.

Question 27

What is a Risk Control Measure (RCM)?

Accepted Answer

A requirement (design input), implementation (design output), and verification (design verification and validation including process validation as appropriate) of a design or its processes that reduces the probability of occurrence, the severity, or both. rmForge proposes specific RCMs per row; ISO 14971 requires that the RCM rationale be traceable.

Question 28

What is a "revised harm" column?

Accepted Answer

After applying a risk control, the residual risk may correspond to a different harm than the original (for example, a fail-safe reverts the device to a lower-energy mode). The revised harm column captures the harm after the risk control is applied.

Question 29

Does rmForge generate the entire risk management file?

Accepted Answer

No. rmForge generates the Hazard Analysis workbook and the customer acknowledgment form.

Question 30

Where do the rmForge HA rules come from?

Accepted Answer

rmForge rules are documented internally and traceable to ISO 14971 and the related IEC, ISO consensus standards, and FDA guidance and regulations. The rule set is version-controlled, and the version in effect is recorded with each analysis.

Question 31

What language model does rmForge use?

Accepted Answer

rmForge uses state-of-the-art commercial large language models from leading providers, with a failover chain so that an outage at any one provider does not interrupt your analysis. Specific model selection is managed by the platform and tuned for the risk-analysis task.

Question 32

How long does rmForge keep my files?

Accepted Answer

Uploaded source documents and generated outputs are retained according to your choice as account initiation. They can be deleted immediately or they are retained for the life of your account. You can request deletion at any time by emailing support@rmforge.io but it will require an authentication from you.

Question 33

Can I download my data?

Accepted Answer

Yes. Every completed analysis produces a downloadable Hazard Analysis workbook plus a signed customer acknowledgment. Both download from your dashboard. There is no lock-in.

Question 34

Does rmForge train its models on my data?

Accepted Answer

No. Customer data is never used for model training. Our AI model providers contractually exclude customer data from training under their API terms. rmForge's invocation pattern is API-only. No model traffic touches the chat product. Provider API terms contractually exclude API traffic from training.

Question 35

How is rmForge different from a generic LLM?

Accepted Answer

Three things: (a) hard-coded task-specific sections and a controlled prompt designed around ISO 14971 and our internal specifications; (b) literature based risk techniques, purpose built code, deterministic grammar, and logic rules plus risk math (Po and Risk Level via table lookup, not AI judgment); (c) validation rules that check every row against the specification and flag any row that cannot be validated.

Question 36

What if the LLM produces a row that doesn't match the spec?

Accepted Answer

rmForge runs an extensive validation rule set on every row. Rows that fail validation are automatically rechecked, and any row that cannot be resolved is clearly flagged in the output for human review just like a normal risk analysis process, but with significant time savings.

Question 37

How does rmForge handle hallucinations?

Accepted Answer

Risk math is deterministic, not AI-generated. Po and Risk Level are looked up from controlled tables. Source quotations are verified against the uploaded document text. Sequence of Events, Hazardous Situation, and Harms follow programmed rules and logic significantly narrowing LLM decision making to a finite rule set in the similar manner that a qualified human would evaluate risk.

Question 38

Can I customize the HA prompt?

Accepted Answer

No. The prompt is controlled under our internal change-management process. Customization can be contracted through support@rmforge.io.

Question 39

Does rmForge support FMEA?

Accepted Answer

Not at this time. FMEA modules (Use FMEA, Design FMEA, Process FMEA) are planned for a later phase.

Question 40

Can rmForge handle non-English IFUs?

Accepted Answer

Today, English only.

Question 41

Is rmForge a medical device?

Accepted Answer

No. rmForge is not a medical device and not Software as a Medical Device (SaMD). It is a user review-required documentation-generation tool. The user remains the legal manufacturer with all responsibility for risk decisions.

Question 42

Who is responsible for the final risk analysis?

Accepted Answer

You are. rmForge provides a structured first draft, but the manufacturer is responsible for the analysis under ISO 14971, ISO 13485, and FDA QMSR. The customer acknowledgment form captures this.

Question 43

How often does rmForge update its rules?

Accepted Answer

Major rule changes (severity scale, probability scale, validation rules) are controlled under an internal change-management process with revision history. Cosmetic updates ship continuously. Each completed analysis records the rule-set version in effect at the time.

Question 44

Where is rmForge's infrastructure hosted?

Accepted Answer

Hosted on industry-standard cloud infrastructure. The application, database, and storage run on tier-1 US-region cloud providers. All traffic is HTTPS end-to-end.

Question 45

Is rmForge output suitable for a submission?

Accepted Answer

Final acceptance depends on the user-manufacturer's risk management process, the design history file, and Quality Management System (QMS). rmForge does not guarantee submission acceptance. rmForge explicitly states that qualified cross functional review and approval per a QMS-based risk management process is required by the user-manufacturer. rmForge does not accept responsibility for the content of a file or submission in any form.

Question 46

Does rmForge align with ISO 14971?

Accepted Answer

Yes, we endeavor to align with ISO 14971. Our internal specifications are written in harmony with ISO 14971 and ISO TR 24971.

Question 47

Does rmForge handle IEC 60601-1 Essential Performance?

Accepted Answer

Yes for ME devices. The Essential Performance agent implements a structured preparation and assessment sequence aligned with IEC 60601-1.

Question 48

Does rmForge handle IEC 62366-1 usability?

Accepted Answer

rmForge tags Intended, Misuse, and Abuse category hazards and surfaces use-related risk rows.

Question 49

Does rmForge handle IEC 62304 software lifecycle?

Accepted Answer

No. IEC 62304 covers software lifecycle processes (software safety classification, architecture, unit verification, integration, system testing). rmForge's HA does not produce a 62304 file. However, rmForge includes IEC 62304 requirement for Class C software to be held the maximum P1 probability of occurrence.

Question 50

Does rmForge handle ISO 10993 biocompatibility?

Accepted Answer

rmForge surfaces biological and chemical hazards in the HA. A full ISO 10993-1 biological evaluation plan and testing report is out of scope.

Question 51

Does rmForge produce a Risk Management Report?

Accepted Answer

No.

Question 52

Does rmForge handle post-market surveillance?

Accepted Answer

No.

Question 53

Can I use rmForge for veterinary or non-human devices?

Accepted Answer

rmForge's rules are anchored to human medical devices under ISO 14971. Veterinary use is not validated. Industrial or research-only devices are out of scope for now.

Question 54

What is the severity scale rmForge uses?

Accepted Answer

Five bands (S1 through S5): Negligible, Minor, Serious, Critical, Catastrophic. S5 Catastrophic covers death and permanent severe disability.

Question 55

What is the probability scale rmForge uses?

Accepted Answer

Five bands (P1 through P5): Very Low, Low, Medium, High, Very High. Each band has quantitative anchors that the agent applies when assigning probabilities.

Question 56

What are p1 and p2?

Accepted Answer

p1 is the probability that a hazardous situation arises. p2 is the conditional probability that, once the hazardous situation exists, harm actually occurs. Both are categorical (P1 through P5).

Question 57

How is Po computed?

Accepted Answer

Po is looked up from a deterministic combined-probability table as a function of p1 and p2. The table is identical across analyses. Language-model judgment does not override the table value.

Question 58

How is Risk Level computed?

Accepted Answer

Risk Level is looked up from a deterministic table as a function of Severity and Po. Four bands: Negligible, Acceptable, Tolerable, Intolerable.

Question 59

What if I disagree with a severity assignment?

Accepted Answer

You can edit the row in the downloaded workbook. The workbook will recompute it according to your choices. With all risk analysis, the user must perform a cross functional review and approval process for all changes.

Question 60

Why are there only four Risk Levels and not five?

Accepted Answer

Four bands map cleanly to risk-control decisions: Negligible = no action, Acceptable = monitor, Tolerable = consider control, Intolerable = mandatory control with verification. rmForge has the opinion that a five-band scheme adds ambiguity without evidence of improving safety outcomes. We make no claims of fact regardless.

Question 61

What is residual risk?

Accepted Answer

Risk remaining after the application of risk control measures and the potential for those risk control measures to create new risks. ISO 14971 requires that residual risk post risk control measures be evaluated and, if not acceptable, that further controls or a benefit-risk justification be documented.

Question 62

What is an acceptable level of residual risk?

Accepted Answer

The manufacturer defines acceptability in the manufacturer's Risk Management Plan per ISO 14971. rmForge's default policy treats Negligible and Acceptable as acceptable, Tolerable as conditional (requires justification), and Intolerable as not acceptable. You can override during analysis. This is simply a starting point for the user-manufacturer.

Question 63

Why doesn't rmForge use ALARP?

Accepted Answer

ALARP (As Low As Reasonably Practicable) is driven by the manufacturer. rmForge presents proposed risk levels and the manufacturer ultimately decides risk acceptability. Refer to ISO 14971 for detail.

Question 64

Can I change the severity or probability scale?

Accepted Answer

Not at this time. Custom scales are on the roadmap for enterprise customers. Custom scales must be controlled documents on the customer side; we do not let you change scales mid-analysis.

Frequently asked questions

Getting Started

Hazard Analysis Basics

The rmForge Platform

Regulatory Context

Risk Math (Severity, Probability, Risk Level)

Outputs and Deliverables

Account, Beta, and Billing

Security and Privacy

Troubleshooting

Product Roadmap