We have redesigned the process from "Decide before creating" SOPs to "Confirm what is decided and verify against your organization's policies" just before publishing as a continuous workflow. At the same time, we have consolidated policy, how to write SOPs, and export format in one place in the dictionary's "Authoring rules" section, adding a mechanism where how to write SOPs and policy interpretation develop from what actually happens on site.

① We have redesigned "Consistency check" into "Pre-publish check"

The previous "Consistency check" was a screen for checking whether decisions and policies aligned. This time, in addition to verifying against policies, we can now output draft reasons for publishing, summaries, and tags, so we have renamed the screen to "Pre-publish check". You can open it from the top of the SOP editing screen.

You can confirm the following 7 items. When you select an axis on the left, you confirm only that axis. Axes that are not open do not run.

• What is decided: purpose, readers, type, and scope decided before creating
• Policy: whether it violates your registered internal policy text
• How to write SOPs: whether it follows the phrasing, granularity, and prohibitions your organization has decided
• Export format: whether there are any fields that remain empty when transcribed to your own company format
• Storage folder: whether the location matches what was decided. You can change it on the spot if there are suggestions
• Reason for publishing and summary: draft of reason and summary you enter in a publishing request
• Tags: suggestions for tags to add and tags to remove

The left side is a list, with result, date and time held, and completion time lined up for each item. Before you open it, you can see how many findings there are or "No issues", axes that are not applicable to this SOP, and axes available in higher-tier plans. You can also run them all together with "Check all".

● We have aligned some screen names

We have unified areas where the same thing was called by different words on different screens as follows. The functionality itself has not changed.

• Consistency check → Pre-publish check
• Internal policy → Policy
• Procedure Authoring Policy → How to write SOPs
• Storage location → Storage folder
• Approval route → Approval workflow
• Decide before creating → AI consultation before you build

② You can now verify SOPs against your internal policy text

Until now, whether "your SOP followed internal policy" relied on the author's memory. Policies are often multi-page PDFs, and re-reading them every time you create an SOP is not practical.

We have made it possible to verify the SOP you are editing while the AI reads the original policy text page by page among registered policies. Every finding is accompanied by the source (policy name and page number) and a quote from the original text that supports it. Findings without a source or findings where the quote cannot be found in the original text are removed before they appear on your screen.

You can open the policy mentioned in a finding right there and check the original text.

● Suggestions for corrections come with the findings

Findings include which field of the step (step name / work content / tips / caution / safety) to correct, plus a suggested correction. For each field, you can turn on "Apply to this field" and reflect the suggestions into the main text all at once. For fields where no suggestion could be provided, we indicate that fact and allow you to write and apply corrections yourself from "Fix it here".

There is no limit to the number of findings. Even for SOPs with many steps, we output all violations found.

● You can also register policies that define work itself

Documents that define the work itself, such as quality manuals, safety and health regulations, ISO documents, and legal standards, can be used directly for verification. Not just documents that define how to create SOPs, such as "Work Standard Creation and Management Policy".

When interpreting a policy that defines work, the type and format may not be read, but that is not a problem. It can still be used to verify SOPs.

● Policies are confirmed with "Save"

Adding or removing files is not finalized immediately. When you press "Save" along with the reading results, they are finalized as one version. Update history is displayed at the top of the screen, and you can revert to previous versions.

③ Address findings one by one and mark the item complete

For each finding that appears, select either "Fixed" or "No action needed". You can also select them all at once. When you have filled in all of them, you can click "Mark the check complete" for that item. When you mark something complete, you can leave a brief note about your reasoning (optional).

The list on the left shows result, date and time held, and completion time for each item. You can see at a glance how far you have gotten. Completion can be canceled at any time.

● Verification results are saved

The results do not disappear when you close the screen. You can read them as-is if you open it again, so you do not need to run the verification multiple times for the same SOP. Results are carried over when you revise, but not when you copy as a separate SOP.

If you edit the SOP after verifying it, the item will display "The SOP has been edited after this", and the completion status is removed. Please verify again with the updated content.

● Approvers can see the same list

Even if you cannot edit the SOP, you can read the check results left by the creator. You can open the details only for items that the creator marked as "Check complete". Until now, there was no way to verify from the approval screen whether "it was checked against policy".

④ How to write SOPs and policy interpretation develop from what happens on site

When findings are skipped, when they are deemed not needing action, or when they are sent back at approval — we record what actually happens on site and use it to rebuild how to write SOPs and policy interpretation (matching types and formats). Records are stored only within that team and do not affect other teams. This is not learning or retraining the AI model itself.

● Automatic updates and decisions made by people

The criterion for dividing them is "whether you can notice if something goes wrong".

• Automatic updates: add required fields / review granularity / add rows to how to write SOPs (if you make a mistake, more findings appear on screen)
• Must be decided by people: remove required fields / rewrite existing rows in how to write SOPs (if you make a mistake, findings silently disappear)

● Only one setting

In team settings, under "Grow it from the records on site", choose either Update automatically / Do not update automatically. Even if you choose "do not update automatically", suggestions will accumulate, so you can review the contents and adopt them manually.

⑤ Insights accumulate while you edit

We have added a panel to the SOP editing screen for accumulating insights. It stays silent while you write, and only shows a badge with the count when you finish writing a step. You can decide whether to open it.

The panel only shows what can be determined without calling AI.

• Steps with required fields left empty
• SOP type is not yet decided
• Registered policies have been revised (previous verification results are not reliable)
• There are unresolved points in the policy
• Items not yet verified in pre-publish check
• Items in the SOP edited after verification

⑥ We have renamed "Decide before creating" to "AI consultation before you build" and it is now available in Core plan and above

"Decide before creating", which you can open from the SOP creation screen, has been renamed to "AI consultation before you build". The name now makes it clear at a glance that this is a feature for deciding through conversation with AI. The name of the decided results, "What is decided", remains the same.

At the same time, it is now available in Core, Pro, and Enterprise plans (previously Pro and above). The purpose, readers, and scope decided in the consultation are automatically carried over to the "Description" field when subsequently analyzing materials with AI. This description is the input that most significantly affects the precision of the step divisions and questions that AI creates.

On the creation screen, we have placed the entrance at the bottom left of the card with a one-line explanation of why to click it. On smartphones, the conversation expands to fill the screen, and what is decided is summarized in a strip at the bottom.

⑦ We have created a new "Authoring rules" tab in the dictionary

We have consolidated what to follow when creating SOPs in one place under the "Authoring rules" tab in the dictionary. What was previously lined up in parallel within templates has been rearranged to show relationships.

• Policy — What to follow
• How to write SOPs — How to write it
• Export format — Which form to export in

When you identify "rules for how to write SOPs" from reading policies, you can add them to how to write SOPs with "Import into "How to write it"". We only add; we do not rewrite existing rows. We do not add duplicate rows with the same content. You select the import destination (company-wide / location), confirm what will be added, and then finalize. The history of what was imported is saved in the update history of how to write SOPs with its source.

⑧ Available plans

The tag feature itself is available on all plans as before. The table above shows the part where AI suggests tags from the pre-publish check.

Impact on existing customers

Screen names will change, but the policies, authoring policies, export formats, and what is decided that have been registered so far are carried over as-is. Re-configuring settings is not necessary.

Related information

Decide the purpose and format of an SOP before creating it (AI consultation before you build)

Verify SOPs with pre-publish check

Interpreting policies that define how to write SOPs

Decide how to write SOPs (reflect your organization's phrasing and standards to AI)

We thank you for your continued use of Dive ◎