AI is quickly changing the world, including our efficiencies within the workplace. The benefits are clear, including our ability to connect and consume information; however, it’s of the utmost importance to ensure these superpowers are used with safety, security, and privacy as the top priority.
PixelMixer’s AI features are designed to help you discover the plethora of knowledge within your organization faster, while providing the right checks and balances to validate that information is accurate.
The purpose of this article is to explain the AI technologies that are available within PixelMixer so you can understand how they can be customized to meet your unique needs.
Using PixelMixer to share your videos provides a number of features which are AI enabled. The list below describes the complete list of features and how they work. For more information about how to enable/disable capabilities, please see the Security section titled Enable/Disable Individual Features.
The basis of processing the information within your videos starts with capturing a text transcript of what was discussed; therefore, all other AI features rely on the transcription feature. The PixelMixer transcription service has been fine tuned for deciphering technical jargon, acronyms and custom business terminology, as well as, the myriad of accents which are used to convey them.
Transcripts are accessible from the Transcript tab underneath the video. From this tab, you can search the transcript and jump directly to that part of the video. If the video was imported from a meeting provider that supports speaker names, you will also see the speaker name displayed next to the transcript segment. Clicking on the transcript text allows you to edit / update the transcript itself (see AI Review section for more info).
In addition to out of the box transcription capabilities, PixelMixer is continually building a customized vocabulary which is specific to your organization. Any time you correct a term within PixelMixer using our review features (see AI Review section), we update the vocabulary so it is properly deciphered by the application the next time it is encountered.
Additionally, PixelMixer Enterprise customers have the option to upload and manage existing vocabularies they might already have available within their organization.
Re-watching an entire video can be a time-consuming task, especially if you’re just looking for the meeting minutes. To that end, PixelMixer AI consolidates the entire meeting into a few sentences so you can quickly get the gist.
Summaries are generated within a few minutes of a video being available within PixelMixer and are displayed underneath the video player within the video description. AI generated summaries are labeled with the AI robot icon so others can be aware this information was derived by AI and may not be 100% accurate. We encourage you to review and update the generated summaries of your videos to ensure information is not mis-represented (see AI Review section for more info).
Let’s say there was a meeting which included a number of different discussions but only one pertained to you or your team. PixelMixer identifies the logical sections of a discussion and tags these as chapters. Each chapter is identified by a headline and a marker so you can watch just the sections which interest you.
A list of chapters is available directly from the player itself, along with a set of markers on the video timeline. The chapter list can also be edited from the Chapters tab (see AI Review section for more info).
PixelMixer AI identifies the topics discussed within videos so you can visually explore content and be automatically connected with those areas which interest you the most.
A categorized list of topics identified within a video are displayed on the Topics tab underneath the video player. Topics with a darker color indicate a heavier density of mentions. Clicking on a topic will display a list of items where the topic was mentioned so you can jump straight to that part of the video.
Topics and their categories can be updated by clicking the topic button pencil icon (see AI Review section for more info).
Action Items and Custom prompts
Arguably the most important point of a meeting is to identify and capture action items so to that end, PixelMixer AI does that for you.
Action Items are generated within a few minutes of a video being available within PixelMixer and are displayed underneath the video player within the video description. AI generated action items are labeled with the AI robot icon so others can be aware this information was derived by AI and may not be 100% accurate. We encourage you to review and update the generated action items for your videos to ensure information is not mis-represented (see AI Review section for more info).
Additionally, PixelMixer Enterprise customers have the option to define customized prompts to meet the unique generative AI requirements of their organization.
At PixelMixer, we value the trust you have instilled in us in keeping your content secure and to fulfill that obligation to you, security is our top priority. To that end, we have ensured users can clearly identify information which is AI generated and give them the ability to modify / edit anything which might be inaccurate. Additionally, we realize that some or all AI features may not be desired or appropriate for some organizations so we provide workspace administrators with the ability to enforce generative AI reviews or disable features altogether.
AI generated and derived content can be edited in place by video creators and workspace administrators. Edit buttons are displayed next to the AI generated content itself so it can be easily updated. The list below provides a reference for where generated content can be found and updated.
Transcripts – From the Transcript tab, click the text to edit the transcript segment. Editing the transcript text will update the workspace vocabulary, as appropriate. Speaker names are pulled from third party meeting providers and cannot be edited directly within PixelMixer and should be updated within the meeting provider platform (ie. Zoom).
Summary, Action Items and Custom Prompts – Edit the video description to amend generated summaries and other custom AI output. AI generated descriptions display a robot icon to let users know it was not written by a human or the video creator.
Additionally, administrators can enforce AI review which will hide/blur the description until it is reviewed by the video creator or an administrator (see Enable/Disable Individual Features for more info).
Topics – Edit detected topics from the Topics tab underneath the video player. Clicking the pencil icon will allow you to update the spelling, as well as, the topic category. Updating the topic will correct all occurrences within the video and update the workspace vocabulary as appropriate.
Chapters – Chapters are displayed within the video player but can be edited from the chapters tab. Chapters can be deleted but adding or amending the timecode is not supported at this time.
Enable/Disable Specific Features
Workspace administrators can enable/disable all of the features mentioned above, as well as, enforce generative AI review from the Workspace Management page. This page can be accessed by going to Workspace Settings within the settings menu and selecting the AI Features tab.
As a PixelMixer user with Workspace features enabled, you still have the option to opt out of using the generative AI features for newly uploaded videos. In order to enable or disable the features, please visit your user preferences page which are available after logging in to your PixelMixer workspace
Moderation and Flagging
As part of our process of continuous monitoring and improvement, AI generated content is continuously monitored and flagged across a broad range of categories. When inappropriate content is identified, PixelMixer support is alerted so action can be taken.
Additionally, PixelMixer Enterprise customers have the option to be automatically notified or receive a report of content suspected to be in violation of their corporate policies.
Third Party Processors
PixelMixer uses various 3rd party processors to support our AI feature offering. A complete list of these processors, services and certifications can be found from our Trust Center page (contact [email protected] to request access).
PixelMixer has agreements in place with our 3rd Party AI processors which prohibit them from storing, transmitting, reselling or using PixelMixer customer data for any purposes other than temporary access for processing requests.
Additionally, PixelMixer Enterprise customers have the option to select specifically which processors they wish to use in support of our services offering.
Please report any issues, threats or concerns regarding AI security to [email protected]. We value feedback and suggestions as we strive to continually optimize our security feature offering.