Mobile2b Compliance Platform#
The Compliance Platform runs the audit and CAPA lifecycle for multi-site operators. You define what your standards require once. The platform plans the audits, grades the findings, creates the corrective work, and keeps the record.
The lifecycle#
| Stage | What happens | Where |
|---|---|---|
| Plan | Checklists carry planning rules. The calendar is compiled from them. | Calendar |
| Execute | The auditor opens the audit and answers the questions. | Running an audit |
| Grade | A failing answer becomes a finding. The severity scale sets the grade. | Findings |
| Correct | The grade selects a CAPA route. The route creates and assigns the tasks. | CAPA workflows |
| Prove | Analytics, reports and the activity log show what was owed and what was done. | Analytics |
Objects#
Organization#
The plant hierarchy, five levels deep: Organization, Site, Area, Subarea, Equipment. Audits happen at a unit, findings inherit the unit, planning rules target units, and CAPA owners are resolved from the unit.
See Organizations.
Management system#
One compliance domain, for example Food Safety, HSE or Quality. A checklist belongs to one management system. A question belongs to at most one. A finding is graded against one.
A management system either follows a standard from the catalog, in which case the standard supplies its severity scale, or it defines its own scale. It cannot do both.
See Management systems.
Standard#
An entry in the Mobile2b standards catalog, for example IFS Food, ISO 22000, IATF 16949 or VDA 6.3. A standard supplies a severity scale: the grades it uses, what each grade requires, and the days it allows.
Mobile2b maintains the catalog and ships updates when a standards body publishes a revision.
See Standards.
Checklist item#
One reusable question in a shared question bank. It carries the answer type and its non-conformance rules, which decide which answers count as a deviation and at which grade.
See Checklist items.
Checklist#
An ordered set of sections that draw questions from the bank. A checklist also carries the planning rules that place it on the calendar. It is published as a revision, and an audit binds to the revision that was live when it was created.
See Checklists.
Task#
Audits, corrective actions and plain to-dos are all tasks, in one list, each with an assignee, a due window and a status.
See Tasks.
Finding#
A recorded non-conformance. It carries a case reference such as C-2026-0184, a grade, the clause it breaches, and the CAPA route it runs. The finding page is the case cockpit.
See Findings and CAPA.
Key behaviors#
These differ from most audit tools, and knowing them in advance explains most of what the interface does.
- Grades are assigned automatically. The auditor describes the non-conformity; the question's non-conformance rules set the grade from the answer, and the severity scale sets the corrective workflow and the deadline. The auditor cannot change the grade, so the same answer produces the same grade at every site and with every auditor, which is what makes results comparable.
- The calendar is computed, not entered. No dates are authored. A checklist states an obligation, and the calendar derives the occurrences. Changing a rule changes the plan everywhere at once, and the platform can list what is not covered.
- A checklist takes effect on publish. A draft plans nothing and cannot be audited. An audit binds to the revision that was live when it was created, so a completed audit always shows the questions that were actually asked.
- A completed audit is read-only. It cannot be edited or reopened. Corrections run through the findings and their corrective actions, which leaves the original record intact.
- Work leaves the plan with a reason. A task is either completed or marked not done with a recorded reason. This keeps the audits-versus-plan figure meaningful, which is the number a certifier asks for.
- Grade consequences cite their source. Where the platform states what a standard requires, it names the clause it comes from.
- People are told once a day, not at every change. Each person gets one digest of their own overdue, due and awaiting-approval work, in their own morning and their own time zone. Only corrective-action work that has just become theirs is emailed immediately, because a CAPA step does not exist until it is workable. See Email notifications.
AI assistance#
The platform can help an auditor write, and can recall what the organization has already seen. It never decides.
Both assists are in the audit runner, on the finding panel. See Running an audit.
| Assist | What it does |
|---|---|
| Draft finding | Turns an auditor's keywords into the written finding, using the question, the requirement, the answer, the grade and the location. |
| Repeat deviation | Names the earlier findings that describe the same deviation, across every site. |
Four rules apply to both, and they are deliberate:
- The grade is never AI-decided. Grades come from the question's non-conformance rules, as they do everywhere else in the platform. A drafted finding cannot change, restate or influence a grade.
- Nothing is written without a review step. A draft is marked as a draft and can be undone. What is saved is what the auditor accepted or corrected.
- AI is off until you connect a provider. The platform has no provider of its own, and sends nothing anywhere until you configure one. See Getting started.
- Your provider, your choice. OpenAI, Anthropic, Google, Azure or a self-hosted Ollama. Self-hosted keeps every request inside your own network.
The platform records which drafts your auditors keep and how they correct the ones they change. It uses that record as the house style for the same question in later audits, so drafts read more like your own team over time. That record stays with your account.
Multi-language content#
Text you enter (unit names, question text, help text, section titles, grade labels, descriptions) is stored per language.
Each of these fields shows a badge with how many languages currently hold a value, out of the languages configured for your account. Select the badge to expand the field and edit each language separately.
An empty language can be filled with Auto-translate this language, which translates from English, or from German when English is empty.
The interface itself is currently English.
Limits#
| Value | Limit |
|---|---|
| Organization hierarchy | 5 levels: Organization, Site, Area, Subarea, Equipment |
| Photos per question during an audit | 8 |
| Photos on a task completion form | 8 |
| Reference photos per question | 8 |
| Score answer range | 0 to 10 |
| Risk matrix | 5 x 5 |
| CAPA routes available | 7 |
| Task auto-archive after completion | 14 days |
| Earlier findings shown on a question | 15, reaching back 24 months |
| Audits rendered on the Analytics audits tab | 30 |
| Activity log entries loaded per request | 50 |
Each page carries a Notes and limits section covering that area.
Next steps#
For a new account, follow Getting started, which covers setup in the order the platform requires.