EdmundDocumentation
User manual · User

Welcome. Find your answer with Edmund.

Need to find a procedure, understand a program or return to an earlier solution? This manual for the User role guides you through finding information and working with chat.

First time here?

What do you need to do?

3 areas

Something isn’t working as expected? Open help →

User · 17 chapters

Everything you need.
In a clear order.

New to Edmund? Follow the first steps, then continue to working with chat. If you have a specific task, open its chapter directly.

Start from the beginning →
User · Getting started

What is Edmund

Learn how Edmund works with your equipment’s knowledge and where to begin.

Edmund is an assistant for the shop floor. Ask it about a machine and it searches that machine’s own knowledge: its manuals, drawings, PLC programs, maintenance history and connected data. Links in the answer take you to the sources it used.

Enlarge example ↗

Projects are your machines

Each machine or production line can be a project. A project holds that machine’s knowledge and chats, so you ask in the context of the equipment in front of you. Open the right project from the list or by scanning its QR code. Find a project and its QR code.

Where answers come from

How you work with it

Ask in plain language in the chat. Edmund’s search assistants explore the available sources, and you can watch the search progress. An answer may include a citation: a link to a particular document or record. Database results may appear as a table. You can rate an answer with a thumbs up or down.

How to ask your first question · How to read an answer and open a source

When a source is missing

If the project’s knowledge does not cover your question, add the equipment or fault identifier and ask for a specific source. Tell whoever maintains that machine’s knowledge about a missing manual, drawing or record.

New here? Start by signing in and opening a project. Then ask one specific question and read the answer together with its sources.

User · Getting started

Sign-in and invitation

Accept your invitation and sign in with your work email and password or your Microsoft work account.

Your first invitation

Your administrator sends an invitation to your work email. Open the message and follow its link. For password sign-in, set your own password as instructed during first access. If the link has expired or you cannot find the message even in your spam folder, ask the administrator for another invitation.

An invitation is valid for 24 hours only. After it expires, ask your administrator to send it again and use the new link.

Sign in with email and password

  1. In Email address, enter the work address that received the invitation. Enter your password in Password.

  2. Click Sign in. If two-factor authentication is enabled, complete the additional verification when prompted.

  3. Use Forgot your password? to reset a forgotten password, then follow the email instructions.

Enlarge example ↗

Sign in with a Microsoft work account (SSO)

  1. If your organization uses Microsoft SSO, choose Sign in with Microsoft.

  2. Choose the correct work account in your company sign-in and complete the requested verification. On returning to Edmund, you see the workspace you can access.

For a Microsoft invitation, you do not create a new Edmund password; your name comes from the Microsoft account.

Enlarge example ↗

Signing in does not grant new projects. Your administrator manages missing access through teams. After signing in, set your account language and security.

User · Getting started

Personal settings and security

Manage your interface language and sign-in settings in your account.

Open the profile menu

Click your profile at the bottom of the left menu. The menu contains appearance choices, account settings, the privacy policy and sign-out. Use this menu for the following steps.

Enlarge example ↗

Choose the application appearance

In the profile menu, choose System, Dark or Light. System follows your device settings. You can choose the appearance without opening account settings.

Enlarge example ↗

Edit your profile and language

Now choose Account preferences in the profile menu.

  1. In Profile, edit Full name or choose a Language. The options are Čeština, English, Polski and Slovenčina.

  2. Confirm with Save. The language changes application controls; document names and earlier messages keep their original language.

Enlarge example ↗

Return to the product tour

The Product tour card lets you enable or disable guidance through Edmund’s features. If you are still getting familiar with the application, use its switch to enable the guidance again.

Enlarge example ↗

Change your password

The sign-in email in Account information is read-only. Under Password, open Change password and follow the form. With Microsoft SSO, sign-in is managed through your company account.

Enlarge example ↗

Set up two-factor authentication

Keep recovery codes private

Further down the page, Two-factor authentication shows whether the additional sign-in check is enabled. If it is Inactive, choose Set up and follow the wizard. Store recovery codes securely; do not share them in chat or documentation.

Enlarge example ↗

Privacy policy

The profile menu also contains Privacy policy. Open it to read how personal data is processed and protected. You can return to this information from the same menu at any time.

Enlarge example ↗

Sign out when finished

Open your profile again at the bottom of the left menu and choose Logout. This ends the session under your account, especially when using a shared computer.

Enlarge example ↗

Sign out of all devices

Open Account preferences and find Signed-in devices. If you lose a device or need to end all browser sessions, choose Sign out of all devices. This ends sign-in sessions across devices. To finish work in this browser only, use Logout in the profile menu.

Enlarge example ↗
User · Getting started

Finding your way around

After opening a project, explore the main menu, the path at the top and the chat list. You will know where to start and how to return.

Two levels in the left menu

The left menu is your starting point. Dashboard and Projects at the top belong to the organization. Dashboard shows a summary; the project list takes you to a particular machine.

After opening a project, its name appears above Chat, Project Dashboard, Documents, Automation Logic and Project journals. These entries concern the selected project. An arrow expands a submenu; automation logic contains PLCs and robots.

Enlarge example ↗

See where you are

A path above the content shows your location, such as the project name and Chat. Clickable parts return to those pages. The project name here and in the left menu identifies the machine you are working with. Open Projects again to choose another machine.

Enlarge example ↗

Chat has an additional chat list

  • New chat opens space for a new topic. Search helps you find an older chat.

  • My chats contains your own chats and folders. Shared contains chats made available to colleagues.

  • Read messages and answers in the center. The next-message field is at the bottom. A source panel can open on the right beside the answer.

  • The sidebar toggle hides or restores a panel. It does not change the project or delete chats.

Enlarge example ↗

Menus show the functions available to your role and enabled modules. Now open your first project.

User · Getting started

Organization, project and knowledge

An organization is your company’s workspace. A project brings together the sources for a machine or production line.

Organization and project have different scopes

  • Organization is your company’s workspace: it brings together projects, people and shared settings.

  • A project represents a specific machine, line or process unit. It groups that equipment’s knowledge and chats. Open the relevant project for another machine.

Enlarge example ↗

What Edmund uses to answer

  • Documents: manuals, schematics, drawings and other files.

  • Automation Logic: imported PLC and robot programs, signals and relationships.

  • Project journals: failures, repairs, adjustments and operating experience.

  • Database sources: assigned data such as stock items or fault records. Read query results as a table in chat.

Edmund searches sources available to the current project. Attaching a file to a message does not permanently add it to project knowledge. How to open original sources.

User · Getting started

Find and open a project

Choose the project for the machine or line you are working with.

From the left menu to a project

  1. Click Projects in the left menu.

  2. Search for the machine or line name. Available tags help narrow the list. Edmund remembers selected tag filters. If nothing appears, clear the search and filters and check further list pages.

  3. Click the project card. It opens that project’s Chat; its name appears on the left and at the top.

In the Filters panel, Machine 1 is selected. Matching projects remain on the right; purple tags on their cards show the assigned values. Project, group and tag names are organization data and are not translated when the app language changes.

Enlarge example ↗

After opening the project

Selecting the project card opens its chat. The machine name appears in the left menu and the navigation path at the top. Start a new chat here, or use the project menu to open its documents and other sources.

Enlarge example ↗

Quick access with a QR code

Scan the QR code assigned to the machine. Sign in if requested. The code is a shortcut to the project; it does not grant new permissions. If the project cannot be opened, send its name to your administrator.

In the project list, open the project’s QR icon. The dialog lets you download a PDF. Place the printed code by the correct machine and protect it from dirt.

Enlarge example ↗

For a data summary, open Project Dashboard in the left menu. See Organization, project and knowledge for the difference from the entire organization.

User · Working with chat

Your first chat: from question to answer

Keep one topic in a chat. Ask a question, wait for the answer and add further observations when needed.

Help Edmund find the right source

Write as if to a colleague who cannot see the machine. Include the machine or section, the exact symptom and code, the source you want and the desired output. For example: “The film unwinder shows [exact HMI code] after the cycle starts. Find its meaning in the manual and related program signals. Cite sources and list steps I can verify.” Replace the bracketed text with the actual code.

Copy alarm identifiers with their letters, leading zeros and punctuation. A short, specific request such as “Find the wiring diagram for sensor [identifier]” can be more useful than a long general description.

Write and send a question

In your project, open Chat → New chat. Write what you need in the message field. Include the alarm identifier, symptoms and circumstances. Instead of “It does not work”, try: “The wrapper reports a film alarm after the cycle starts. Where does the manual explain it, and which related journal entries do we have?”

Click the Send arrow. While the answer is generated, you can see search progress and a Stop button. Some other actions may be unavailable until the response finishes.

Enlarge example ↗

Stop a response in progress

While a response is being generated, Send changes to Stop, shown as a square icon. The example highlights it in the lower-right corner of the message box.

Click it to interrupt the current response. This does not delete the chat. The message is marked as stopped, and the application states that Edmund cannot use stopped messages for later answers. To continue the request, send it again or write a new message.

Enlarge example ↗

Continue from the answer

Read the answer and open its sources. Then use the same field to ask a follow-up: “Explain the point about the sensor” or “Compare this entry with the previous similar event”. You do not need to repeat the whole chat.

Clearly distinguish new observations from assumptions: “The diagnostic screen currently shows…”. Ask Edmund to locate a source when one is missing. For another machine or unrelated problem, start a new chat in the appropriate project.

User · Working with chat

Attachments, photographs and voice input

Attach supporting information to your question and explain what Edmund should look at.

Attach a file or photograph

Click the Upload files paperclip, choose a file and wait for it to attach. Add context, such as “This photograph shows this morning's alarm; please explain its identifier”. Check the attachments before sending.

You can also attach HEIC / HEIF photographs, for example from an iPhone. Edmund converts HEIC photographs to JPEG during upload, so you do not need to convert them first. Check that the photograph has attached to the chat before sending.

Enlarge example ↗

Upload through the tab at the top of the chat

You can also add a file through Attached files at the top of an open chat.

  1. Open Attached files.

  2. Click the upload arrow in that tab and choose a file.

  3. Wait for the upload to finish. Then return to the messages and explain what Edmund should look for in the file.

This route also creates a temporary attachment for that chat, rather than a permanent project document.

Attachment readability and chat context

Edmund primarily works with digitally readable text. A document whose text can be selected and copied makes exact names and values easier to find. With a photograph or scan, resolution, sharpness and legibility matter; add the alarm, part identifier and what you need to find out in your message.

Chat context has a limited capacity. A large file, many attachments or a long chat may not fit in full. If Edmund does not refer to a relevant part or asks for less content, select the relevant pages, divide the file or question into smaller parts and identify the exact section. Start a new chat for a separate topic when needed.

From a temporary attachment to project knowledge

Attached files holds files for that chat. Attaching a file to a message does not automatically add it to permanent project Documents. To use the source in other chats and share it with the team, a User expert or administrator uploads the original file through Documents → Upload documents and waits for processing. A User passes the file to that person. This is a separate upload, not an automatic transfer of the attachment.

Dictate a message

Click Start voice input. Allow microphone access if the browser requests it for this task. Review the transcription, especially alarm numbers, part names and signal identifiers. Correct errors before sending.

If the microphone is unavailable or the surroundings are noisy, type the message instead. The message should be understandable without the original audio context.

Enlarge example ↗

You can attach a PDF or a PNG, JPEG or HEIC/HEIF photograph, as supported by the file picker. Make sure an HMI photograph clearly shows the alarm identifier. Voice input works on desktop and phone; always check the transcription before sending.

User · Working with chat

Reading answers, citations and search progress

A citation is a link to an original source supporting part of an answer.

Recognize a source in an answer

Some claims have a clickable source label, such as a document title or a dated journal entry. It may appear as a small chip, an inline link or a grouped chip showing additional sources with +1. This is a citation.

Click the source name beside the claim you want to check. A grouped chip first offers several sources; select one. The answer itself is not necessarily a verbatim quotation of the source.

Enlarge example ↗

Open and read the original source

Opening a citation displays a source panel beside the chat. Read the original content and any date, version or entry attachments. For a document, review the relevant page and its surrounding context. For a journal, check that it describes the same machine and a genuinely similar issue.

Close the panel using Close and continue in the same chat. Ask, for example, “Explain the difference between this source entry and my situation.”

Enlarge example ↗

Expand the search progress

A button such as Searching in journals or document search shows the assistant's activity. Expand it to read the available progress details. Progress explains where the search took place; a citation opens a specific supporting source.

Enlarge example ↗

How to read different source types

A database answer may also expose the database query in its search progress. Check that it matches the requested equipment and period.

A document reference opens the relevant page and available highlighting; a program source points to the corresponding part of the imported logic. A database result may have no document-style citation for every row. Check its request, period, columns and units instead.

When an expected citation is missing

An answer without a citation is based on Edmund’s general knowledge. If Edmund does not cite the document you expect, check that it is uploaded to the relevant project and has finished processing. Find its name and content in Documents; a User expert or administrator can help add the source. Then specify the document or section Edmund should use in your next message.

Rate an answer with thumbs up or down

Two rating icons appear beneath an answer. Use thumbs up, Good response, when the answer helped. Use thumbs down, Bad response, when it is incorrect, incomplete or unhelpful. Rate the particular answer directly above the icons.

Clicking thumbs up or down saves your rating immediately. A comment is optional. A time-saved estimate is offered only after thumbs up, with no value preselected.

If a follow-up form opens, describe what helped or what is wrong. For an error, identify the correct source if you know it. Ask another question in the same chat when you need further explanation.

Some chats also show a How did Edmund do? panel with Helped and Didn’t help, as in the following example. Rate it according to the actual outcome.

Enlarge example ↗
User · Working with chat

Chat history, editing and sharing

Your chats remain in My chats. Shared chats provide a reference for colleagues.

Find an earlier chat

Select an earlier chat in My chats. If you cannot see it, use Search and enter part of its name or the topic. Its previous messages and answers remain available, so you can continue with a new question.

New folder creates a personal folder for your chats. The next section shows how to move a chat into it using Chat actions.

You can also start a new chat directly inside a personal folder.

Enlarge example ↗

Open actions for a specific chat

Open the chat and click the three-dot Chat actions beside its title. The same menu appears beside the chat in the left list. Rename changes its searchable title. Move to folder organizes the chat in your folders; it does not share it with others. Delete removes the chat.

Enlarge example ↗

Rename a chat

Rename opens a small dialog with the current title. Replace it with a name describing the machine and topic, such as “Strema 5997: film unwinding”. Confirm with Save. Close dismisses the dialog without saving. Renaming does not edit earlier messages.

Enlarge example ↗

Move a chat to your own folder

  1. Create a New folder in My chats if needed.

  2. Open Chat actions → Move to folder for that chat and choose a destination. Root folder means the main list outside personal subfolders.

Enlarge example ↗

Edit your own message

Move the pointer over your own message and click the Edit pencil. Its original text opens in an editable field. Correct an alarm number or missing context and confirm with Send. Cancel keeps the original wording.

Add new observations as a new message

Editing an earlier question may reprocess the following answer. When you simply have a new observation, add it as a new message below the existing chat.

Enlarge example ↗

Share to the right folder

If Edmund is generating an answer, wait until it is complete. Sharing is unavailable during generation.

Open your own chat and click Share at the top. User and User expert can select only General. Administrators may also choose Onboarding for introductory and training chats.

Enlarge example ↗

Confirm your selection with Share to General.

Enlarge example ↗

You share the original chat, rather than a copy. As its owner, you continue in the same chat under My chats. Other people with access to the project can read new messages as well; sharing does not let them write in your chat.

Sharing started by a User or User expert lasts 30 days. An administrator can renew sharing for another 30 days, keep it open without automatic expiry or stop sharing. The owner can also stop sharing their chat.

User · Working with chat

Incident analysis from start to conclusion

Incident analysis investigates an issue using the available sources and context.

1. Open incident analysis and describe the situation

Open chat in the project and choose Analyze an incident. Use this mode to connect symptoms, source material and new findings. State the machine, exact symptom, start time, operating state and checks actually performed. Say “unknown” when a fact is missing.

Practice prompt: “The film does not unwind when wrapping starts. The issue appeared after changing the roll. The display shows [exact alarm]. The sensor and cable have not been inspected. Find related sources and ask for missing information.” Replace brackets with the actual observation. Attach photographs using the paperclip.

Enlarge example ↗

2. Choose the depth and start the investigation

Investigation effort offers Low, Medium and High. Choose how thoroughly Edmund should investigate. Review the prompt and click Analyze. Follow the search progress while the response is generated and wait for completion before drawing a conclusion.

Back to chat leaves the mode before sending. You can interrupt generation using the square Stop button in the message field; its appearance is shown in the first-question chapter.

Enlarge example ↗

3. Read the incident update

The completed response is a structured Incident Analysis. Its top strip shows investigation effort, the number of cited sources and the analysis duration. Incident update summarizes the current conclusion, possible branches and suggested continuation. It does not confirm a repair.

In the example, Edmund states that the cause is still unconfirmed. It connected roll-loading instructions, sensor fault history and PLC information. Separate source facts from conditions actually checked on the machine. Open the source names beside individual paragraphs to verify them.

Enlarge example ↗

4. Choose the next step

Next steps appears below the update. Each card has a number, title, priority and explanation. Here the first asks for the actual alarm and sequence step; others explore roll loading or the sensor. The number and content of cards depend on the analysis.

Clicking a card neither performs a machine check nor sends a message. It prepares a follow-up in the composer. Use Add your own follow-up to continue with a different observation.

Enlarge example ↗

5. Check the evidence and gaps

Supporting evidence groups relevant findings, such as PLC and database evidence. Expand a group to read it. Evidence gaps separately lists what remains unsupported; the number in brackets counts these gaps.

Separate assumptions from verified findings

Here the exact alarm, sequence step, inspection results and live values are missing. Without them, the text cannot identify a specific failed part. An empty database search does not necessarily prove that an event never occurred. Add observations that distinguish the remaining possibilities.

Enlarge example ↗

6. Inspect investigation activity

Expand Investigation activity below the result to see the types of sources Edmund searched. Individual search items can be expanded further. The list explains the search process; it does not prove that the conclusion is correct. Compare claims with their cited sources.

Enlarge example ↗

7. Add observations and continue

After choosing a card, Continuing selected analysis appears. Expand Prepared reply to read the prefilled instruction. Add your observations below; continuing combines the prepared instruction with your notes. Do not report a check that was never performed.

Your own notes are kept when you select a different suggested next step.

Here we stated that the alarm, sequence step and roll movement were still unknown and asked which HMI readings were needed. Continue submits the follow-up in the same investigation. Remove selected step clears the choice; Back to chat leaves the mode before sending. Investigation effort can also be adjusted for the continuation.

Enlarge example ↗

8. Use the updated result

The next response again contains an update and fresh continuation cards. Edmund adapts the investigation to your input. In the example, identifying the missing information leads to more specific questions about the HMI readings. This is a training-chat result, not a universal machine work instruction.

Report actual outcomes in the same chat. Clearly mark unconfirmed causes and request a summary of findings, work and open points. An analysis does not automatically create a journal entry; a User expert or administrator records it if needed.

Continue the analysis with a suggested step or your own additional observation. Opening a regular new chat starts a separate conversation outside this incident flow. A more detailed search may take longer.

Enlarge example ↗
User · Overviews and sources

Organization and project overview

The overview connects summary information with individual projects and sources.

Organization overview

Open Dashboard in the left menu. It summarizes available projects and knowledge, activity and links to items needing attention. The visible scope follows your access.

Click the document alert row or the arrow on its right. It opens the affected project’s documents with a filter matching the alert. For example, a duplicate alert opens the list filtered to Status: Duplicate. A total document count does not mean that every document has been processed successfully.

The Knowledge Base card is shown below. The other examples in this chapter show individual dashboard panels. Click Enlarge example below an illustration and view it at 100% to read small values.

Enlarge example ↗

Overview of one machine

Inside a project, select Project Dashboard. It provides a chat entry point, your recent chats, source status and the latest Maintenance journal entries. The Documents chart and table share one block, showing document types and processing status. Click a maintenance entry to open the original record.

Click a maintenance entry name, or click the alert row or the arrow on its right. Continue to chat when you need an explanation; open the document or journal list directly when you already know which item you need.

Enlarge example ↗

What needs attention

Needs attention separates issues from summary counts. The example shows 47 duplicate documents: the same files have already been uploaded. These are not 47 processing failures. Click the alert row or the arrow on its right to open the affected project’s documents filtered to Duplicate.

As a User, inspect the status and contact an expert or administrator. An alert does not grant permission to upload, edit or delete documents.

Enlarge example ↗

What the summaries mean

  • Knowledge Base shows counts of documents, maintenance entries, PLC programs and robot programs. Use the controls above the charts to change the view and period.

  • Documents shows processing status and whether files are available for search. Duplicates are counted separately.

  • Activity compares the numbers of chats and maintenance entries in the selected and preceding periods.

  • Your recent chats lets you quickly reopen your latest chats.

Enlarge example ↗
Enlarge example ↗
User · Overviews and sources

Working with project resources

You can also open a source directly, without asking in chat.

Documents and their status

Open Documents in the project and search by name. The list shows status, type, source and other details. Open a document or its menu for preview and available actions. Download the original when the action is available to you.

For a PDF marked as repaired, distinguish the normal download from downloading the original. The normal download provides the repaired version; downloading the original provides the uploaded file. Unrecoverable pages may be missing, so check the relevant page before using a particular procedure.

Processed indicates completed processing. An in-progress status does not yet mean usable knowledge. No text, errors and duplicates need expert review. If an answer cannot find a document, give the expert its name and displayed status.

Enlarge example ↗

PLC program identity and structure

Open PLC under Automation Logic and select the program. Read its name, indexing status and latest upload first. Program structure lists the available export parts; browse by device and filename.

If an expected block is absent, that does not establish that the controller does not use it. It means it has not been confirmed in the available import. Use exact program or block identifiers in questions.

Enlarge example ↗

Read the program composition chart

The PLC detail includes a composition chart beside Program Structure. In the English interface its heading is Organization Units Distribution. It is a radar chart: category names such as DB, FB and FC sit around its edge, with a colored polygon connecting their values. Use it to understand the imported program’s composition, then find the relevant type and file in Program Structure.

It describes the imported source, not the current machine state, live faults or current PLC values. The example contains the complete original chart and its surrounding labels.

Enlarge example ↗

Robot program and I/O signals

In Automation Logic, select Robots and open a program. Browse its imported files and the I/O tab. I/O means input and output signals used by the program to communicate with equipment. Search narrows the list by name or identifier. Click a signal to open its available detail.

The imported program is a source for understanding logic, not the robot’s live state. Include the exact file name or signal identifier in your question.

Enlarge example ↗

Open a maintenance entry

Open Project journals in the project. Choose a journal, then an entry by its title. A journal groups multiple events; the opened entry describes one of them. In a long list, find the event using the available names and details.

The next example explains the opened entry and its versions.

Enlarge example ↗

Read an entry’s description and solution

After opening an entry, read the event date and title. Description explains the observations; Solution records the work and outcome. Read both together. A similar fault title does not establish an identical cause.

Files contains attached material. Version history shows later changes. When using an entry cited in an answer, open the citation and compare the specific claim with the original content.

Enlarge example ↗

Find electrical and pneumatic diagrams

Look for electrical diagrams, including EPLAN exports, and pneumatic diagrams among PDF documents. Select Schematic diagrams in the Type filter, then narrow the file name if needed. This example shows a list filtered to diagrams. Follow the labels available in your environment if the type name differs.

Enlarge example ↗
User · Overviews and sources

Database tables, charts and exports

Ask a data question, read the result table, reveal more columns and download a CSV file. The project must have the appropriate data source assigned.

Specify the scope of a data question

Open chat in the relevant project. Specify the records, period and fields you need. For example: “From the connected database, show at most five latest fault records as a table with all available columns.” Add a machine identifier for one machine, or start and end dates for an exact period. Then click Send.

The following example contains the five latest database records and 28 columns. It is not automatically limited to one machine. Ask an administrator if a connection is missing or you do not know which data the project uses.

Enlarge example ↗

Read the headers and rows

The first table row contains column names; the rows below contain record values. Here each data row represents one fault. First read which records Edmund included: five displayed records are not the entire database history.

The example shows the left edge of the table. More columns continue beyond the right edge. The next example shows how to reveal them with the scrollbar.

Scroll the table left and right

Go to the bottom edge of the table itself, below its last row. The horizontal scrollbar has a track and a movable thumb. Drag the thumb right to reveal columns on the right; drag it left to return to the first columns. A horizontal touchpad gesture also works.

The example highlights the scrollbar at the bottom of the table. Try scrolling the columns directly in Edmund.

Enlarge example ↗

Sort the displayed records

Click the small arrow button beside a column name, for example Order number. The table sorts by that column and the arrow indicates direction. Sorting acts on the displayed result; it does not rewrite database records.

Enlarge example ↗

Download the result as CSV or XLSX

CSV is a table-data file that you can open in a spreadsheet application. Click Export below the table, then choose Export CSV or Export XLSX. The browser downloads the result file. Open it in a spreadsheet application and check headers, row count, period and units. The CSV belongs to this query result; it does not automatically export the entire connected database.

Per page changes the displayed row count. Ask a follow-up question for different records or columns, such as “Limit the result to last week and keep the date, machine and description.” The next example highlights Export, which opens the format menu.

Enlarge example ↗

Turn a table into a chart and download its image

Continue in the same chat, for example: “Show this table as a chart. Label the axes, units and period.” Specify the horizontal-axis value and what you want to compare. Charts are not automatically included in every database answer.

Check the chart legend, axes and units against the table. The following example shows frequency and cumulative share. Export PNG downloads a chart image; download the underlying table separately with Export CSV.

This separate chart example uses a different query and a frequency summary; it is not a conversion of the five rows shown above. The chart type and available export depend on the particular result.

Enlarge example ↗

How a database description helps

Database search can be improved over time. Alongside table and column names, Edmund can use a plain-language database description. It explains what the data means and how it is related: for example, which columns identify a machine or spare part, the units used for downtime, and which identifier links the tables.

If Edmund repeatedly misinterprets an abbreviation, value or relationship, send your administrator the specific question, the incorrect interpretation and the correct explanation. The administrator can refine the description. Better context may help Edmund select and find the right data; continue to check each answer against its results.

User · Roles and permissions

Roles, teams and permissions

Your role determines what you can do. Teams affect which projects you can access.

User, User expert and administrator

User uses chat, reads available sources, opens answer references and gives feedback. User expert also manages documents, programs and journal entries in accessible projects. Administrator additionally manages projects, people, teams, tags, suggestions, connections, organization journals and analytics.

User and User expert can share chats only to General. Sharing to Onboarding is reserved for administrators. The application may also offer a Reader role for viewing information without using chat. This manual has separate paths for the three main roles.

When a project or action is missing

Clear project search and filters. If a project is still missing, ask your administrator to check team membership and the project's assignment. If you can see the project but cannot upload or edit, ask them to check your role.

A project without an assigned team is not automatically private. Administrators can inspect the permission matrix in Users → View permissions. The administrator manual also includes User and expert workflows.

For the User role, the absence of document upload and knowledge editing is expected, not an error. Ask a User expert or administrator to add new source material.

Permission matrix

A role determines permitted actions; a team determines available projects. Readers can view source material but cannot use chat or rate answers. Administrators see all projects; other roles see their teams’ projects and projects without an assigned team.

ActionAdministratorUser expertUserReader
Use chat and rate answersYesYesYesNo
View documents, programs, data sources and journalsYesYesYesYes
Manage documents and programsYesYesNoNo
Manage maintenance entriesYesYesNoNo
Create and manage projects and their data sourcesYesNoNoNo
Manage users, teams, organization settings and analyticsYesNoNoNo
Manage connections and create organization journalsYesNoNoNo
User · Help and feedback

When something does not work

Start with what you can see on the affected screen.

Missing answers, sources or access

For an unclear answer, add an exact identifier and open the citations. Ask an expert to check a missing source and its processing status. For a missing project or action, ask an administrator to check your team and role.

Open Contact us in the left menu. The first example highlights the entry point; the following example shows the contact form.

A missing menu can also mean that an organization feature has not been configured or enabled. Ask an administrator to check your role, project settings and available modules first. If an expected connector or module is missing, the administrator can contact the Edmund team.

Enlarge example ↗

Contact support through the form

Open Contact us, choose a Reason, enter a short Subject and describe the action, expected result and actual result. For a problem in a particular chat, include its address and the time. Check the message, then use Send.

Thumbs rate an individual answer; the contact form also handles application issues and improvement requests. For example, “The source link to manual revision X is missing; the correct value is on page Y” is more useful than “It is wrong.” Rating an answer does not mean the next answer automatically learns the correction.

Enlarge example ↗