Triggering the spell: The /update command - your main tool for training knowledge and configurations
Now that you've got your know-how ready in your sources folder on Google Drive, and have possibly modified your settings in Transcription settings.gdoc, it's time to instruct your BuddyPro instance to "read" all this information, process it, save it to your knowledge base, and also load your current personality and onboarding settings.
10.1 The /update command: what exactly does it do and when do you need it?β
In the chat with your BuddyPro instance (in Telegram), enter the key command:
/update
This command is the central point for updating virtually everything that defines your AI expert (except for the contents of the expert subroles themselves; there is a separate /updateRoles command for that, see Chapter 11). In the background, the /update command runs the following complex series of steps:
- Synchronize with the
sourcesfolder on Google Drive:- BuddyPro carefully checks the contents of your
sourcesfolder and all its relevant subfolders(sources/Files,sources/Texts,sources/URLs). - It identifies any new files or URLs you have added since you last ran
/update. - Recognizes changes to existing Google Docs (e.g., in
sources/Textsor theURLfileSOURCES.gdocinsources/URLs). - It notices deleted files or URLs from the
sourcesfolder.
- BuddyPro carefully checks the contents of your
- Processes new or changed knowledge sources:
- For all new or modified audio and video files (uploaded to
sources/Filesor added as URLs insources/URLs/URL SOURCES.gdoc), transcribe them to text (transcription). In doing so, it fully respects your current settings in thetranscription_settings.gdocfile (language, keywords, enhanced transcriptions). - For new PDF, DOCX, TXT files (in
sources/Files) and Google Docs (insources/Texts) it extracts their complete text content. - For new or modified URL links in
sources/URLs/URL SOURCES.gdoc, it will attempt to download and extract the textual content of the web pages in question. For YouTube and Vimeo videos, it will attempt to extract the textual transcript. - It will save the resulting plain text transcripts (transcriptions) or extracted text of these sources to the appropriate
transcriptionssubfolders in thesourcesfolder on your Google Drive (e.g.,sources/Files/transcriptions/NameFile.gdocorsources/URLs/transcriptions/url.gdoc).- Note for your review: These transcription files allow you to check exactly how your know-how was transcribed and what text base the AI will use. By default, they also include metadata at the beginning (a brief summary of the content of the resource and a suggestion of potential expert subfields that could emerge from it). We do not recommend interfering with this metadata. The transcription text itself can be manually edited on Google Drive if necessary (e.g. to correct major transcription errors). If you do so, you must re-run the
/updatecommand to have your manual changes to the transcription reflected in the AI knowledge base.
- Note for your review: These transcription files allow you to check exactly how your know-how was transcribed and what text base the AI will use. By default, they also include metadata at the beginning (a brief summary of the content of the resource and a suggestion of potential expert subfields that could emerge from it). We do not recommend interfering with this metadata. The transcription text itself can be manually edited on Google Drive if necessary (e.g. to correct major transcription errors). If you do so, you must re-run the
- For all new or modified audio and video files (uploaded to
- Update the internal vector database (the AI's knowledge "brain"):
- All newly processed or changed textual content (after transcription/extraction) is intelligently divided into smaller logical chunks ("chunks").
- Each of these chunks is converted into a vector embedding (a complex mathematical representation of its meaning and context).
- These vectors are added or updated in your AI instance's internal secure vector database. This database is unique to your AI expert.
- If you have deleted any source files or URLs from the
sources, the corresponding data and vectors are also securely removed from the vector database. This way, your AI expert "forgets" information that should no longer be part of its knowledge.
- Update the list of suggested expert subroles:
- Based on the complete, up-to-date content of your knowledge base, BuddyPro re-analyzes the possible topics and specializations contained in your know-how.
- It updates the list of names of these proposed (or already confirmed and possibly modified by you) expert sub-roles in the file
ROLES/list_of_roles.gdocon your Google Drive. - Important note: The
/updatecommand alone does not create or modify the detailed content (definitions) of these roles in their separate Google Docs in theROLESfolder. That is the job of the subsequent/updateRolescommand (see Chapter 11). The/updatecommand only sets the "stage" and updates the list of role name suggestions.
- Load the current versions of key configuration files:
- The
/updatecommand also ensures that your AI instance reloads and applies the current contents of the following Google Docs configuration files from the root folder of your linked Google Drive:- π
system_prompt.gdoc(defining your AI's main personality, communication style, and basic behavioral rules) - π
onboarding.gdoc(setting up welcome messages for new users and automated messages during conversations) - π
transcription_settings.gdoc(setting parameters for transcription of audio and video files)
- π
- This means that if you make any, even the slightest, change to these three key configuration files, you must then always run
/updateto ensure that your changes are reflected in the behavior and functioning of your AI instance! Without this, the AI will still work with the previous settings.
- The
10.2 How to track the progress of /update and when it is done?β
- When you issue the
/updatecommand, your BuddyPro instance will usually confirm that the process has been started with a short message in Telegram (e.g.Updating Buddy Pro...). - The processing time depends on the amount and type of new or changed know-how and configuration files. Processing of several small text files can be done in seconds or minutes. Transcribing hours of new audio/video material or processing a large amount of URLs can take even longer (sometimes several hours, depending on the amount of content).
- When the entire process is successfully completed, your BuddyPro instance should send you a message in Telegram, for example,
Updating Buddy Pro Finished Successfully. - What to do if the process takes too long?
- Be patient: Especially if you have uploaded a large amount of data (e.g. many videos, long audio recordings, dozens of URLs), give the system plenty of time to work. For really large volumes, this can be up to 5 hours.
- Check Google Drive: look in the
transcriptionsfolders undersources(e.g.sources/Files/transcriptionsorsources/URLs/transcriptions). If new transcribed files (.gdocs with transcripts) are gradually appearing there, it means that the system is still working and processing your know-how. - Ask your BuddyPro instance: If you are not sure whether the process has finished, you can ask your BuddyPro isntance what the status of your know-how update is and whether it is still running.
10.3 Troubleshooting common errors and processing problems with /updateβ
If BuddyPro encounters a problem with a particular file, URL, or configuration files during the /update process, it will usually notify you in a Telegram message after (partial or failed) completion and list the problematic resources or error type.
- The most common errors and their solutions:
- π Error related to subfolders (e.g.
Subfolders not allowed in sources/Filesor similar message forTextsorURLs):- Cause: You have created some custom subfolders in your
sources/Files(orsources/Textsorsources/URLs) folder. The BuddyPro platform expects all knowledge files to be placed directly in these main designated subfolders(Files,Texts,URLs), not in another directory level you created within them. - Solution: Open the appropriate folder
(sources/Files,sources/Texts, orsources/URLs) on your Google Drive. Move all the files from your own subfolders one level up (i.e. directly into theFiles,TextsorURLsfolder). Then delete the empty subfolders you created. After this structure modification, run the/updatecommand again.
- Cause: You have created some custom subfolders in your
- Error:
The user does not have sufficient permissions for this file(or other error related to file access rights):- Cause: This error most often occurs if you have accidentally created or manually uploaded a file directly to one of the system's
transcriptionssubfolders (e.g.sources/Files/transcriptionsorsources/URLs/transcriptions). These folders are exclusively for files generated automatically by the BuddyPro platform (i.e. for transcriptions of your knowledge sources). If you upload a file yourself, BuddyPro logically has no permission to overwrite, delete or otherwise manipulate such an "extraneous" file, because you are the owner of the file, not the BuddyPro system. - Solution: Carefully go through all of the
transcriptionssubfolders on your Google Drive (insources/Files,sources/Texts, andsources/URLs). Delete any files that you manually created or uploaded there. Remember - never add anything to these(transcriptions) folders yourself (except for very targeted and careful manual editing of an existing specific transcript if you know what you're doing and why - but even that is best not done unless you're 100% sure). After cleaning up these folders, run the/updatecommand again.
- Cause: This error most often occurs if you have accidentally created or manually uploaded a file directly to one of the system's
- β οΈ Error overwriting a particular audio, video file or URL:
- Cause: Can be caused by a variety of factors:
- The file is too large and the server runs out of RAM while processing it.
- The recording has extremely poor audio quality, making reliable transcription impossible.
- The file format is not fully supported or is corrupted.
- The URL is temporarily or permanently unavailable, or the page behind it requires a login.
- There is a temporary technical issue on the part of the external AI transcription service that BuddyPro uses.
- For very long videos (e.g., over about 3 hours), there may also be problems downloading from the source platform (e.g., YouTube) or processing such a large amount of data at once.
- Solution:
- Try re-uploading the problematic file (sometimes it helps if it was a temporary outage).
- For large video files, try uploading a version with a lower image quality (smaller data volume) or, if possible, extract and upload only its audio track in MP3 format (which is much smaller in data and equally valuable to the AI, since the AI does not primarily process the image information).
- For unavailable URLs, check their functionality in an anonymous browser window - they must be publicly available.
- If the problem with a particular file/URL persists after repeated attempts and verification, contact BuddyPro support
([support@buddypro.ai](mailto:support@buddypro.ai))and ideally provide them with a link to the problematic file/URL for analysis so they can investigate the problem.
- Cause: Can be caused by a variety of factors:
- π Error related to subfolders (e.g.
- After any problem fixes on your end, always run the
/updatecommand again so that the changes take effect and the system attempts to process the data correctly again.
10.4 "/update" is your most frequent sidekick: use it regularly!β
To summarize: you must use the /update command WHENEVER you make any of the following significant changes to your BuddyPro configuration or knowledge base:
- β Add new files (PDF, audio, video, text documents) or URL links to the appropriate subfolders in the
sourcesfolder on your Google Drive. - βοΈ You edit the contents of any existing file or Google Doc in your
sourcesfolder (e.g., update information in a PDF, rewrite a chapter in a Google Doc insources/Texts, add a new URL to theSOURCES.gdoc URL). - β Delete some files or URL links from the
sourcesfolder (to stop the AI from using outdated or no longer wanted information). - βοΈ Make the slightest change to the contents of the key configuration files
system_prompt.gdoc,onboarding.gdoc, ortranscription_settings.gdoc
Without running /update, your changes will not be reflected in the behavior or knowledge of your AI instance! It's like saving a document on your computer but not turning on cloud sync - others (in this case, your AI) will still see the old version and work with the outdated data or settings.
10.5 Option to abort a long-running process: /stopUpdateβ
If for some very urgent reason you need to abort the currently running /update process (for example, you realized too late that you accidentally uploaded a huge amount of data you didn't want to process and you want to stop it as soon as possible), you can use the command:
/stopUpdate
After using /stopUpdate and correcting the situation (for example, deleting the unwanted files), you will need to run /update again.
With this understanding of the /update command, you are now equipped with a key tool for training and configuring your AI Expert. In the next chapter, we'll look at how to create specific expert "hats" - that is, expert subroles - from your processed know-how using the /updateRoles command.
Frequently Asked Questions (FAQ) for this step:β
- What happens if I change
system_prompt.gdoc(oronboarding.gdoc/transcription_settings.gdoc) and don't run/update? Your AI instance will continue to use the previous, old version of this configuration file it had loaded. Your changes will not take effect until you run/update. - Do I need to run
/updateevery time I make a minor modification to a single file insources? Yes, if you want that minor modification to be reflected in the AI knowledge. The/updatecommand is designed to be relatively efficient and only process changed or new parts. - What if I accidentally delete the transcription file from the
transcriptionsfolder instead of the original file fromsources? If you delete only the.gdoctranscription file from thetranscriptionsfolder, but the original source file (e.g., PDF, MP3) remains in thesources/Filesfolder (or the URL in theSOURCES.gdoc URL), then the next time you run/update, the system will detect that the source file still exists, but its transcription is missing, and try to recreate it. However, if you want the AI to "forget" the know-how, you need to delete the original source file directly from thesourcesfolder (e.g.sources/Files/my_video.mp4) and then run/update. - What is the difference between updating the list of roles in
ROLES/list_of_roles.gdocand generating the content of the roles? The/updatecommand parses your know-how and only updates the list of suggested expert sub-role names in theROLES/list_of_roles.gdocfile. This file serves as a "menu" of available specializations. However, the detailed content (definition) of each role itself (which are separate, often multi-page Google Docs in theROLESfolder) is not created or changed by the/updatecommand. That's up to the/updateRolescommand, which generates or updates the detailed content of each individual role based on the current list and your know-how.