> For the complete documentation index, see [llms.txt](https://learn.withessential.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://learn.withessential.com/administration/using-the-essential-assistant/editing-an-inspection-report-through-the-assistant.md). # Editing an inspection report through the Assistant You can ask the Essential Assistant to fill in or correct an inspection report — the same edits the person filling it in makes on the report itself. Open the report, open the Assistant, and describe the change in plain language: "mark extinguisher 3 as failed, the gauge is low", "pass everything on floor 2", "add two more extinguishers to the lobby list", "record the annual service as completed yesterday". The Assistant proposes the edit as an action card, you confirm it, and only then does the report change. Nothing is written while you are reading the card. ## Before you start * Your role needs permission to edit inspection reports. If it doesn't, the Assistant won't offer these edits. * Have the report open, or name it in the chat ("the extinguisher report from Tuesday at the Elm Street building"). The Assistant already knows the report you're viewing. * Report **forms** — which questions exist, their wording and order — are set up by Essential and can't be changed this way. Use the **Talk to a human** tab for those. ## What it can change * **Answers**, both on the report as a whole and on each asset: pass or fail, readings, dates, choices, comment boxes. * **Assets**: add one to a list, duplicate one several times, skip an asset, remove it, or reorder the list. * **A whole selection at once**: pass, skip, remove, or answer the same question across many assets — the chat equivalent of the report's bulk actions. * **Repeatable list sections**: add and edit rows. * **Deficiencies**: raise one against a failed question, edit its details, or remove it. (Resolving, reopening, retesting and clearing a deficiency are separate actions — see [Asking the Assistant to do things](/administration/using-the-essential-assistant/asking-the-assistant-to-do-things-actions-you-confirm.md).) * **Service completion dates**: record or correct when a service was completed. * **Photos** you have attached to the chat — see [Photos](#photos) below. ## Adding assets Describe the assets the way you would to a technician: "add 4 smoke alarms in the lobby", "add 400 smoke alarms", or "add smoke alarms for a 40 floor building, 4 on each floor — they all pass except 2 on floor 3 and 1 on floor 6". * **Which list they go in.** The Assistant matches what you name to the right asset list on the form and, when the list has a device-type question, sets it (a smoke alarm lands in the device list with its type set to the smoke-alarm option). If two lists — or two types — could fit, it asks which one you mean. * **Details per asset.** Anything you describe in a pattern is filled in per asset: "4 per floor across 40 floors" gives each one its floor in the location, and "all pass except 2 on floor 3" marks exactly those as failed. Questions you don't mention start with the same defaults a technician gets when they tap **Add**. * **When something required is missing.** If the form requires a detail you didn't give — usually the location — the Assistant asks once before proposing, and offers to leave it blank for the technician to fill in on site or to number the assets ("Smoke Alarm 1" to "Smoke Alarm 400"). Say "leave the locations blank" up front and it won't ask. If an asset has several checks and you only say it "failed", it asks which check. * **Failures become deficiencies.** Every failed check on a new asset gets a deficiency record, exactly as tapping **Fail** on the report would; add notes or photos to them on the report afterwards. * **Large adds go to a review page.** Adding up to 15 assets uses the ordinary action card. Adding more opens a [bulk review](/administration/using-the-essential-assistant/reviewing-a-bulk-change-from-the-assistant.md) with one row per new asset — each row reads like "New Initiating Device — Floor 3 — Smoke Alarm", with what it will be filled in with — so you can check them, leave some out, and confirm the rest. New assets join the property's equipment list when the report is submitted, the same as assets a technician adds. This also works on a report that isn't attached to an asset group yet. ## What you see before you confirm A value on a report means very little read on its own, so the card doesn't just show the raw edit. Each change is spelled out as a **before and after**: which question, on which asset, what it says now, and what it will say once you confirm. Anything the edit takes away is called out in red as cleared or removed, and those lines are never folded away — when an edit touches more than a handful of things, the rest collapse behind **Show N more changes**, but every removal stays on the card. Read that list before confirming: it is the record of the inspection. Two things the Assistant works from rather than guesses: * **It reads the report first**, so it knows which questions the form actually has, which assets are on it, and what is already answered. * **It checks its own edit against the form** before offering it. A value the form doesn't allow — a reading where the question expects a choice, an option that isn't on the list, an answer on a question that doesn't exist — never reaches the card. If somebody saves the same report while your card is open — a technician's phone coming back online, say — the action stops rather than running against a report you never previewed, and tells you so. Ask again and the Assistant re-reads the report and proposes afresh. ## The report on screen stays in step When you confirm an edit, the report you are looking at updates by itself — the new answers appear in the form without reloading the page, and whatever step or asset you had open stays open. The one time it won't is when **you** have unsaved changes in the form. Your typing is yours: the page never replaces it with something else, and the Assistant can't see it either — it works from the report as last saved. So if you ask for an edit while the form has unsaved changes, the Assistant says so and asks you to **save or discard** them first rather than proposing over the top of them. Save (or undo your changes), ask again, and it proposes against the report as it now stands. ## Photos Attach an image to the chat (see [Attaching files to the Assistant](/administration/using-the-essential-assistant/attaching-files-to-the-assistant.md)) and ask for it to go on the report. It can go onto a photo question or onto a deficiency, and the card names the file and where it will land. A photo question that takes several pictures keeps what's already there and adds yours. A question that holds just one photo works differently: your new photo **replaces** whatever was already on it. If you attach several pictures for the same single-photo question in one message, the Assistant asks which one you mean rather than quietly keeping only the last. Signatures are the one exception. A signature is somebody's attestation that they did the work or accepted it, so the Assistant will never sign on anyone's behalf, and it refuses any edit that would write into a signature box. ## Correcting a report you have already submitted Submitted reports can be corrected too, with one thing to understand. Submitting is what turns the answers into the customer's record: it saves the deficiencies, records the service completions, and renders the PDF. So a correction on its own would change the answers while those stayed as they were. When you ask for a change to a submitted report, the Assistant therefore proposes two actions: the edit, and a **re-submit** straight after it. The confirmation card also marks the edit itself with a note that the report is already submitted, so you can see at a glance that the customer's copy stays behind until it is submitted again. Confirm both and the whole record catches up — the deficiencies and service completions are recalculated and the PDF is rendered again. The re-submit doesn't email anybody; if the customer needs the corrected report, ask for it to be sent and you'll get a separate action for that. If you confirm only the edit, the report's answers change but its deficiencies, service completions and PDF stay as they were until it is submitted again. ## What it doesn't do * It doesn't sign anything for anyone. * It doesn't change the report form: adding, renaming, reordering or removing questions is report setup, handled by Essential through **Talk to a human**. * It doesn't decide an inspection for you. It records what you tell it, against the questions the form actually has. > **Where to find it:** Open the Essential Assistant panel from an inspection report — or name the report in any conversation — and describe the change. The edit appears as an action card in the chat with Confirm and Reject buttons. ## Related guides * [Asking the Assistant to do things: actions you confirm](/administration/using-the-essential-assistant/asking-the-assistant-to-do-things-actions-you-confirm.md) * [Attaching files to the Assistant](/administration/using-the-essential-assistant/attaching-files-to-the-assistant.md) * [Working with the assets on an inspection report (bulk pass, skip, delete)](/field-operations/inspection-reports/working-with-the-assets-on-an-inspection-report-bulk-pass-skip-delete.md) * [Save a draft and publish an inspection report](/field-operations/inspection-reports/save-a-draft-and-publish-an-inspection-report.md) * [Marking fail and recording a deficiency while filling a report](/field-operations/inspection-reports/marking-fail-and-recording-a-deficiency-while-filling-a-report.md) * [What the Essential Assistant can and can't do](/administration/using-the-essential-assistant/what-the-essential-assistant-can-and-cant-do.md) --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://learn.withessential.com/administration/using-the-essential-assistant/editing-an-inspection-report-through-the-assistant.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.