A Mobile Permit Request Is Not Permission to Work
Maximo Mobile brings permit requests to the job site, but capturing a request is not the same as authorizing hazardous work. A defensible rollout tests authority, server-side controls and handback evidence rather than celebrating a successful form submission.
A Mobile Permit Request Is Not Permission to Work
A technician can stand beside an asset, identify a job that needs a permit, and start the request without returning to an office. That is the useful promise behind the field-based permit capability documented for Maximo Mobile and Maximo Health, Safety and Environment. It is also where an implementation can make a dangerous conceptual mistake: treating an easier request as an easier authorization.
IBM's HSE 9.2 documentation describes a Permit to work mobile application that lets technicians initiate requests at the job site while integrating with existing safety processes, approval, validation and controls. The Mobile 9.2 page describes creating and updating work permit records, using both Permit request and Permit to work wording in its feature descriptions. Neither description is a reason to collapse request, review, issue and acceptance into one business event. [1][2]
The practical question is not whether a phone can create a permit record. It is whether the complete operating process still distinguishes what somebody observed, what somebody requested, what an authorized person approved, and what the crew is permitted to do. Those distinctions need to survive device behavior, identity changes, incomplete preparation and the end-of-job handback.
A recent community walkthrough makes the server-side part unusually concrete. Mohamed Chihaoui follows an invented hot-work scenario on a demonstration installation of Manage 9.2.4 with HSE, documenting blocked issuance attempts as well as successful transitions. It is useful evidence of observed application behavior, not proof that every installation or mobile workflow behaves identically. [3]
The implementation guidance below turns those sources into an acceptance approach. Recommendations are deliberately separated from documented capabilities. It does not prescribe a safe isolation procedure, replace a site's approved control-of-work process, or claim that an offline request constitutes authority to begin work.
1. Define the transaction before designing the mobile journey
Start with the nouns. A work order describes maintenance work. A permit request initiates a control-of-work process. An issued permit records a particular authorization under particular conditions. A crew's acceptance records its side of the handover. Calling all of them a job ticket may make a presentation simpler, but it makes acceptance testing ambiguous.
IBM documents mobile-created permits integrating with existing HSE and Oil and Gas processes and workflows. That is the product boundary to preserve: the mobile entry point participates in the established process rather than becoming a competing permission system. The exact enabled application, supported deployment and configured workflow must still be confirmed for the estate being implemented. A release feature list is not an estate-specific compatibility assessment. [2]
As a design recommendation, write a transaction map in ordinary operational language before configuring navigation. Identify who requests the work, who reviews hazards, who controls the plant area, who issues, who accepts, and who closes the permit. For each action, identify the authoritative record and the visible evidence that the action succeeded. Where roles are combined under an approved procedure, record that explicitly rather than assuming that every organization separates them identically.
The community example links its permit to a location, an asset and an originating work order. These links are useful acceptance targets because they anchor the request to an identifiable job. The example's pump, work order and people belong to its demonstration scenario. Production teams should substitute reviewed local records, not copy the example's operational assumptions or treat its isolation plan as reusable engineering guidance. [3]
A meaningful mobile test therefore starts before the Submit button. Can the requester identify the correct asset rather than a similarly named neighbor? Does the selected work order describe the actual scope? Can a reviewer understand the location without interpreting an informal nickname? If the scope changes, does the approved procedure explain how the request and authorization are reassessed? These are proposed test questions, not claims that the product automatically resolves every ambiguity.
Make the status language equally explicit in training. A request successfully saved is evidence that information was captured. It is not, by itself, evidence that the required reviews happened or that the issuing authority authorized work. Avoid a local success message that says ready to start when it merely means request submitted. Review custom notifications with the same care as the underlying workflow.
The deliverable from this stage should be a short, approved state-and-responsibility description. Mobile designers, HSE owners and supervisors should be able to read it and give the same answer to one question: what exact evidence tells the crew that this particular work may begin?
2. Test authority with real roles, not an all-powerful demo login
The community walkthrough describes authority flags on People (HSE), including issuing, performing, isolation and gas-testing authorities, with review and approval rights also configured on people. It also points to qualifications and training applications as supporting records. These are concrete configuration areas to examine, but the article does not establish that every training expiry automatically revokes every authority in every implementation. Do not infer that integration without testing it. [3]
The distinction between access and authority matters. A person may need access to request work without being authorized to issue a permit. Conversely, a supervisor's authority in an approved operating procedure does not prove that the person's current account, person record and mobile access have been configured correctly. Treat the correspondence between organizational responsibility and application permissions as an explicit acceptance item.
Use separate test identities representing the actual roles. An administrator can help diagnose configuration, but an administrator completing the entire demonstration proves little about segregation of responsibilities. A suggested test pack includes a legitimate requester, an authorized issuer, an authorized performing authority and a person deliberately lacking the authority for a sensitive action. The negative identity is as important as the positive one.
For each role, capture the intended action, the expected permitted or denied result, and the authoritative evidence afterward. Test both a person who should be able to act and a person who should not. If a custom workflow routes approval through another application, include that path. Do not assume that a screen hidden on a phone is equivalent to a server-side prohibition; verify the actual configured boundary through supported application behavior.
Identity lifecycle changes deserve their own rehearsal. A proposed test is to change a test user's responsibilities using the normal administration process and then observe when the new state takes effect across the deployed application and device. Record the actual result. This article does not assert token lifetimes, cache behavior or revocation timing because the opened sources do not establish those details for the reader's environment.
The demonstration's insistence that people sign as themselves is a useful operating principle. Shared credentials undermine attribution even if the workflow completes. A signature trail should be understandable to a reviewer who was not present: who acted, in which role, on which record, and at which stage? Whether additional electronic-signature requirements apply is a matter for the organization's approved controls and the supported configuration.
This is not an argument for more permission groups everywhere. It is an argument for fewer untested assumptions. The outcome should be a role-to-action acceptance matrix maintained by the process owner, with mobile access and HSE authorities assessed together rather than in separate project workstreams.
3. Prove that incomplete preparation cannot become an issued permit
The strongest evidence in the community walkthrough is not a polished final screen. It is a pair of failed attempts. The author reports BMXAO0415E when attempting issuance without the required issue row, and BMXAO0419E when isolation operations had not been marked complete. Those observations show specific controls operating on that demonstration installation. They do not establish a universal guarantee that a completed digital row reflects a correctly isolated physical asset. [3]
That boundary is fundamental. Software can require recorded preparation and authorized actions; it cannot substitute a database checkbox for competent physical verification. A rollout should therefore test digital enforcement while keeping the approved field procedure authoritative for actual hazard controls. Do not rewrite an isolation method from a blog example, and do not use production hazardous work as the setting for exploratory software tests.
In a controlled test environment, build a negative-path pack around the organization's required gates. The example suggests two useful starting points: missing issuance evidence and incomplete isolation records. Add local conditions only where the approved procedure and supported configuration require them. State the expected refusal before running the test, so a surprising success cannot be rationalized afterward as acceptable behavior.
Capture more than an error screenshot. Record the starting state, the identity used, the attempted transition, the message received and the saved record state after the attempt. Then correct the test condition using the proper role and repeat the permitted action. This paired approach demonstrates both that the gate rejects incomplete preparation and that it allows the intended progression once the required evidence exists.
A mobile entry point adds another acceptance question: does the request arrive in the correct process with the correct related records? The sources establish integration with existing controls, but the community walkthrough is not an end-to-end mobile certification. A project needs its own evidence that the deployed mobile route reaches the configured HSE process and that no customization mistakenly treats successful request creation as successful issue.
Test clear communication after refusal. The requester should know whether the request remains pending, requires correction, or needs an authorized reviewer. A message translated into local terminology should preserve the distinction between incomplete data and lack of authority. Avoid instructing users to retry indefinitely when a deliberate control has blocked the action.
Finally, retain the failed test cases as regression assets. Changes to automation scripts, security configuration, workflow routing or mobile presentation can alter the user journey even when the underlying business requirement has not changed. Repeating the negative paths after relevant changes is an engineering recommendation, not a claim that the opened documentation prescribes a particular regression schedule. The useful outcome is evidence that convenience did not remove the gate.
4. Treat connection state as an evidence boundary, not a permission rule
IBM's Mobile 9.2 page documents a persistent indication when the app is disconnected, local saving of work during disconnection and synchronization when connectivity returns. It also documents individual record synchronization and updated refresh options. Those general mobile capabilities are useful context, but they do not independently prove every permit action's offline availability, conflict resolution or authorization semantics. Those details must be verified for the deployed application and workflow. [2]
The safe implementation conclusion is narrow: distinguish local capture from authoritative server state. If a device shows a newly entered request while disconnected, the project must establish what that display means for the supported permit journey. Do not turn a generic offline capability into a promise that issuance, approval or revocation operates offline in a particular way. The sources opened for this article do not support that promise.
A proposed test uses an approved non-production scenario and controlled connectivity changes. Observe what remains visible before a request is transmitted, what happens after connectivity returns, and what the server-side record contains afterward. Where an action is unavailable, document the supported fallback procedure rather than attempting to bypass the restriction. Where the UI retains data locally, verify how the user can distinguish pending capture from confirmed server processing.
Also rehearse uncertainty after an interrupted action. A user may not know whether a request completed if the connection fails around submission. The correct response should come from the supported application workflow and local procedure, not an improvised assumption that another tap is harmless. Test how operators confirm the existing record before creating another request, and document the observed behavior without promising universal duplicate prevention.
Inspection assignment is a related but separate concern. HSE 9.2 documents assigning inspections to individuals or groups to reduce unnecessary downloads of unassigned inspections, with ownership information used to route inspections appropriately. That is useful for a field process that involves related inspections. It is not evidence that an inspection assignment grants permit authority or that every permit-related inspection is automatically included in a technician's device data. [1]
Device evidence should therefore be paired with server evidence in acceptance records. Identify the record, the observed connection state, the action attempted and the authoritative result. Keep any sensitive operational information out of broadly shared screenshots. For production operating procedures, let the HSE owner determine what users must do when the authoritative status cannot be established.
This approach avoids both extremes. It does not dismiss mobile operation because connectivity is imperfect, and it does not declare every locally visible status safe to act on. It requires the deployment to explain exactly what the user knows, what the server knows, and what action is permitted while those views are uncertain.
5. Include handback and closure in the release acceptance gate
A mobile rollout can overinvest in the beginning of a job: selecting an asset, attaching evidence, submitting a request and notifying a reviewer. The end deserves equal attention. In the community example, handback includes distinct evidence for completed work, safe condition, isolation removal, return to production and closure review. Most importantly, the author notes that enforcement of particular handback signatures before cancellation and closure depends on organization-level HSE settings. [3]
That observation changes the implementation question. A field team seeing a Closed record should not have to guess which checks were mandatory in that organization. The project must compare the approved handback procedure with the actual configuration. A completed demonstration is not enough if the demonstration permitted closure without a signature the production procedure requires.
Use the community observation as a prompt for configuration review, not as a ready-made configuration prescription. Ask the HSE owner which evidence is required before cancellation and closure, who provides it, and whether each requirement is enforced digitally, procedurally or through another controlled mechanism. Record any gap explicitly. A gap is not resolved merely because the team agrees to remember the missing step during training.
Build negative closure tests in the same controlled environment used for issuance tests. Omit a required item and verify the expected result. Restore the item through the correct role and verify the valid path. If the application deliberately allows a step under the site's configuration, make that explicit in the acceptance record and obtain process-owner review. Do not describe every permissive result as a product defect when the configuration may be intentional.
Language matters here too. The community walkthrough explains that cancellation is used as formal end-of-job sign-off in its permit process, rather than simply abandonment. That makes local terminology and training important. Users accustomed to cancellation meaning nothing happened may misread the record unless the implementation explains the approved lifecycle. Validate the terminology with operations rather than translating labels mechanically. [3]
Finally, review the evidence as a connected chain. The request should identify the intended work; reviews should identify decisions; issue and acceptance should identify the handover; closure should explain how responsibility returned to operations. An auditor should not need to reconstruct the story from unrelated device screenshots. This is a recommended evidence standard, not a claim that every required artifact is automatically produced in one report.
The release decision should cover that whole chain. A request created successfully at the asset is an important milestone. A permit lifecycle that preserves authority, rejects incomplete preparation and records an intelligible handback is the actual operational deliverable.
Practical Implications
For a mobile deployment lead, the immediate task is to replace a form-centric demonstration with a process-centric acceptance pack. Keep the supported application and release scope explicit. Use actual role configurations in a test environment, include both permitted and prohibited actions, and retain the authoritative record state after each important transition. Do not infer permit-specific offline behavior from general mobile documentation.
For an HSE configuration owner, compare the approved site procedure with People authorities, issuance requirements and organization-level closure settings. The community walkthrough is valuable because it exposes both enforced gates and configurable closure checks. It should provoke verification, not be copied as proof of local compliance. Review the connection between training governance and configured authorities rather than assuming it is automatic.
For maintenance supervisors, define what a crew must verify before starting work and what to do when the authoritative status is uncertain. A submitted request, a locally saved change and an issued permit are different evidence states. Training should use the same distinctions as the application and the approved procedure.
For the release approver, require evidence from the complete lifecycle before authorizing rollout. Include a successful request, a blocked unauthorized action, a blocked incomplete-preparation path, a permitted issue and acceptance, and the configured handback path. These are recommended acceptance artifacts. They are not a substitute for engineering review, regulatory obligations or the organization's established control-of-work responsibilities.
Bottom Line
Maximo Mobile's field-based permit capability can put the request where the work is. IBM documents integration with existing HSE processes, and the community example provides concrete observations of server-side issuance controls and configurable handback requirements. The useful implementation story is not fewer clicks. It is better capture without weakening the distinction between asking for work and authorizing it. [1][2][3]
A defensible rollout keeps that distinction visible through role configuration, negative testing, authoritative record verification and closure review. It treats uncertain connectivity as something to investigate and document, not as a reason to invent unsupported approval behavior. It treats a demonstration's recorded isolation steps as evidence of software workflow, not proof of physical safety.
The acceptance question is simple enough for a supervisor and demanding enough for an architect: can the organization show who requested, who authorized, what gates were satisfied, and how the plant was handed back? If the project can answer only that the mobile request saved successfully, it has tested the opening screen rather than the control-of-work process.
Sources: [1] IBM HSE 9.2 documentation. [2] IBM Maximo Mobile 9.2 documentation. [3] Mohamed Chihaoui's permit-to-work community walkthrough. The walkthrough uses an invented plant and standard demonstration users; its observations are not production performance or safety certification.