The meeting involved a discussion about the advancement of the AI tool and integration into the teamās current projects. The main topics included the use of AI for creating tweets, the potential of the tool for auto-generation of text that resembles a userās tone, improving public awareness and involvement in OpenOps, and further experiments with the tool. A demonstration of the AIās ability to simplify or professionalize text inside the Mattermost interface was highlighted. More tests and public visibility are needed for OpenOps to grow beyond its insignificant open-source status. The team also discussed the potential for personalization of the AI tool and considered various use cases, like simplifying or expanding on an idea based on the context of a given conversation. Future plans include implementing changes, such as improving the visibility of work done on OpenOps by publishing updates and progress on public forums, rather than private channels.
Cloud Recording Issue
During the last meeting, a recording was made and uploaded as an MP4. However, there was difficulty in processing the file in Corpus.
Christopher was contacted for assistance but he was unavailable due to apparent PTO.
AI & Social Media
Discussed the use of AI to create social media posts, specifically for promoting a case study with the Checklist Vacuum Company.
There was a concern over the AI potentially misinterpreting language nuances, which resulted in an inappropriate output.
More vigilance mentioned is needed when using AI to generate content like this, acknowledging that sometimes the AI will provide excellent output, but at other times, it can be entirely inappropriate.
Update from Jesus
Jesus is working on integrating AI tooling into their interface.
Demos were shown where he used AI to simplify and professionalize text in a message post.
Discussion of potentially applying AI at different levels like individual productivity, team productivity, and organization-level productivity.
Open Discussion
Propose to put out example scenarios, coding, etc. on a more public platform (like forums) to engage the community in helping fill in gaps and troubleshooting.
Discuss potential ways to tweak that could be easily visible in the console for developers.
Having individual user-level override to allow individuals to fine-tune the AI to align more with their personal tone.
Action Items
Plan to socialize OpenOps publicly to garner more participation, feedback, and ultimately aid in the development of the tool.
Jesus will document all progress on OpenOps via public forums. This will include demos, issues, and other workings.
Ideas on using AI in various buckets from individual productivity to organization-level productivity will be shared via public forums.
Andrew will work on making zaps from forum channels to Corpus server.
Seth is tasked with connecting shared channels between the public AI exchange and the Corpus server.
There is an interest in squarizing videos for better visualization on mobile & social platforms. Jesus will take care of this.
Invite the Sales Engineering (SE) team to future meetings for demo discussions.
Summary generated using AI, and may contain inaccuracies. Do not take this summary as absolute truth.
In the meeting on August 3rd, key topics included preparing a demonstration of the OpenOps AI MindMeld and exploring new updates and questions. The group was joined by customers Victor, Brent, and David Arias for increased visibility and participation. The first agenda topic entailed discussing potential elements for a demo meant for a customer meeting later that day. This spurred detailed dialogue about highlight-worthy components, primarily centered on recent improvements related to summarization of GitHub issues and channel data. Team members explored the demonstration of these capabilities live during the meeting. They also discussed potential integration with personalization features. Other notable updates included the addition of AI tooling to enhance data processing and interpretation, problems with the auto-react feature in need of attention, and the possibility of using AI to generate workplace policies. For future collaboration, the team considered setting up brainstorming sessions with customers. In the next meeting, it was proposed that the team should help write an A.I. training course using A.I. and spend some time discussing foundational A.I. concepts.
Meeting Agenda and Participants
Meeting held on August 3rd
Participants include Victor, Brent, David Arias, Stu Doherty, Eric Setha, Chris, and Andrew.
Topic 1: Preparing a Demo for Customer Meeting
The aim was to brainstorm ideas for a compelling demo in the customer meeting.
Discussed focusing on the personalization features. For instance, by showing how the AI understands the user participating in the conversation.
Topic 2: New Features and Updates
Discussing new updates on the product.
Discussed the demo videos on LinkedIn were mentioned in an all-hands meeting.
Topic 3: Questions and Feedback from Guests
Fielded several questions from the guests present.
Demo for Customer Meeting
Highlight the personalization piece of the AI, specifically showing how it interprets and responds to the user based on their specific role.
This includes understanding whatās relevant for the developer advocate in the conversation.
New Features Updates
The AI now pulls information from GitHub and summarizes this information based on the userās title or perceived role.
It can also pull information related to an issue the user might have participated in on GitHub and make relevant suggestions or highlights.
Questions and Feedback from Guests
A discussion about the document extraction.
Addressed the issue of inappropriate emoji reactions by the AI. It was agreed that additional regulations should be added to avoid such instances.
Introduced a feature where the AI can generate and interpret coding data.
Action Items
Andrew: Search the town square and pin a good meeting summarization for demo purposes.
All: Do not post any confidential documents on the community server for testing purposes due to its public nature.
Christopher: Get the meeting summary for the current meeting and post it on the āshareā channel for AI exchange.
Ian, in collaboration with Victor and Eric Sethna: Commence collaboration with a customer for testing and brainstorming. (Customer details kept confidential in the transcripts.)
All for next AI MindMeld: Spending 15 minutes discussing various fundamental topics such as āWhat is LLM?ā, āWhat is the database side of AI?ā, allowing OpenAI to write the training course through question and answer format.
Summary generated using AI, and may contain inaccuracies. Do not take this summary as absolute truth.
During the meeting, the team discussed incorporating the use of Python Notebooks (also known as Jupyter Notebooks) into Matomos, which is a standard for machine learning notebook experimentation. The idea presented was creating a way for these notebooks to be previewed directly on Matomos. They are not intended to replace Jupyter or be the server that parses Python, itās more about allowing them to be viewable within Matomos system. Other discussions were based on how the team could improve change logs and make them more informative with less need for insider knowledge. Factors that could make this possible include altering the template for PRs to include a change log update draft. There was also a suggestion to start linking PRs and demos to the forums for greater visibility. The meeting ended with the potential concept of creating an AI that could simplify everyday tasks and streamline operations in Matomos being discussed.
Key Discussion Points
AI Mindmeld Discussion Points
Jupyter Notebooks Preview in Mattermost
The team has been working on a new mechanism to preview Python/Jupyter Notebooks directly within Mattermost.
Python/Jupyter Notebooks are a popular tool for machine learning and data science experiments.
Bringing this ability into Mattermost allows for easier sharing and collaboration on notebook files within the platform.
The implementation involves parsing the JSON file of the notebook and rendering it in the Mattermost interface.
Increasing Volume in Video Summaries
It was noted that the volume in video summaries is a bit low in replay.
The team will look into increasing the volume for future videos for ease of comprehension during replays.
Mattermost Built-In Features and Custom Rendering
Creating a means to easily integrate Jupyter Notebooks within Mattermost workflow is a step towards realizing the full potential of Mattermost as an āoperational hubā with custom rendering capabilities.
It is an example of functionality that could be built within Mattermost.
Understanding Jupyter Notebooks
Jupyter Notebooks is like a code pen for Python. It serves as a platform where Python programmers can build and share their demos.
However, displaying Jupyter Notebooks within Mattermost only allows preview of the final rendered Notebook file. It does not replace Jupyter or provide similar processing functionalities.
Increasing Dissemination of Information
Leveraging AI to create meeting summaries and automate minor tasks within Mattermost could improve the dissemination of information.
The more automated the summarization and documentation processes, the easier it is to keep everyone updated on matters discussed during meetings.
Incident Response
AI could be utilized to help with incident response by scanning past incidents and providing useful information and potential solutions for new incidents.
Currently, there are discussions about introducing hashtags to Mattermost posts to make it easier for AI to filter and find related issues.
Integrating Jupyter Notebooks with Mattermost
Although the current integration of Jupyter Notebooks with Mattermost provides value, team also discussed potential deeper integrations.
Actions Items
Plan for a feature that increases video sound volume in replays.
Improve the process of how changes are logged so summaries of changes are easier to produce and understand.
Consider introducing hashtags to Mattermost posts to ease the work of AI in finding related issues.
Continue exploring the possibilities of integrating Jupyter Notebooks with Mattermost for better collaboration and workflow efficiency.
Think about how AI can be used in incident response and other potential use cases within the Mattermost platform.
Summary generated using AI, and may contain inaccuracies. Do not take this summary as absolute truth.
The meeting discussed several key topics. First, a community moderator named John Combs has created a chat GPT support bot using Zapier which acts as a technical support rep. The bot has been well-received, and there was discussion on whether to make this information public. A potential demo of the bot is being scheduled and it is also suggested that Combs be invited to a future meeting.
There is also an ongoing exploration of the botās capabilities, especially how much it can remember. There was a conversation about the potential costs involved and if these should be covered if the tool becomes widely used, particularly since Combs might be using his personal account for the project.
The team evaluated whether the focus should be more on improving the AI model or adding more features to the AI plugin. They settled on the latter, pointing out that theyād mostly covered the good models and wants to move towards customization.
A demo by Jesus about a new functionality extension to the āunread message barā was shown. This feature allows users to summarize unread messages in a channel. While the frontend has been developed, the backend is not yet implemented, but a pull request has been made for its functioning.
Finally, the team discussed a potential UX issue regarding the inconsistent appearance of the āunread message barā and needing another way to trigger it. The team agreed that this could be tackled through plugin extension. The meeting ended on a positive note, with participants appreciating its insights.
Key Discussion Points
GitHub Support Bot Discussion
Community moderator John Combs created a V1 chat GPT support bot using Zapier.
The bot, Matterbot, simulates a technical support rep and tested well in initial use cases.
The bot is not perfect and requires improvements.
The botās response accuracy and possibly retaining state were impressive, providing relevant responses in the demo.
There are concerns about making the bot public and who will bear the cost if it significantly increases usage.
It is believed that Matterbot could be useful internally, serving as a draft for the support team, effectively reducing their workload.
Action Items
Invite John Combs to an upcoming meeting.
Explore how Matterbot can be improved.
Identifying Next Priorities
There is a suggestion to try out different models to heighten the excitement around the AI plugin.
Some team members believe that adding new features and customization paths would add more value than exploring new models.
There is a need to gain access to the AWS Bedrock model.
A decision must be finalized on what the next highest priority should be.
Action Items
Continue discussion on the next priority.
Make attempts to gain access to AWS Bedrock.
Invite John Combs, the creator of Matterbot, to participate in the discussions.
Unread Message Bar Updates
A demo of the newly added āSummarizeā functionality to the unread message bar was shared.
The āSummarizeā button presents a summary of unread messages.
The back-end functionality has been implemented successfully and needs to be connected to the front-end.
There are some User Experience (UX) inconsistencies with the presence of the unread message bar that need to be rectified.
Action Items
Resolving UX inconsistencies with the unread message bar.
Hooking the back-end functionality to the front-end.
Summary generated using AI, and may contain inaccuracies. Do not take this summary as absolute truth.
In the meeting, the team discussed a range of topics including new developments in the OpenOps and AI fields at Mattermost. An area of focus was on improvements to the meeting summarization tool, with an aim to avoid frequent breakdowns. This involves summarizing in chunks to manage longer meetings and to improve performance and quality. The team also spoke about their multi-streaming function which allows typing to occur in different fields simultaneously.
The possibilities of auto-generated summaries and YouTube videos were discussed, with the idea that demo excerpts and relevant information can be clipped and shared. The team recognized the potential for such features to be used for marketing, change-log updates and illustrating project velocity, potentially attracting more contributors and users.
The meeting touched on possible future capabilities, including file analysis to extract information from spreadsheets and other documents, which can be used to create graphs or further data analysis. Some participants expressed concerns about determining which features should be paid, given the underlying AI capabilities of Mattermost are already paid for. Potential options included features related to compliance and security, such as enforcing upload policies based on file types.
Furthermore, the team encouraged the sharing of non-confidential questions and discussion points, encouraging feedback from all members of the group. The conversation concluded with the team acknowledging a range of next actions and responsibilities, including reviewing potential paid features and integrating summaries of meetings into YouTube text. Future meetings suggest rotating the individual responsible for introductions to diversify YouTube image thumbnails.
Key Discussion Points
OpenOps AI Meeting Summary
New Features/Observations
The team has been tidying up some security aspects and improving the meeting summarization that keeps breaking due to the file being too large.
The team has now made it possible to transcode on the fly, reducing the quality of the audio file, and accommodating the Whisper API. The file can be summarized in chunks to avoid overwhelming the system.
The multi-stream feature allows for simultaneous responses, which has proven beneficial.
Future Possibilities
Discussion was raised on the idea of an automatic system that would create release notes from the meeting summary, post it onto YouTube, and clip parts of the demo for marketing purposes. Some parts could be easily doable with Zapier integrations.
The team suggested customizing meeting summaries to include a release notes section where necessary.
File analysis was suggested as a potential pay feature: upon file upload, the user could be asked to confirm if it contains IP or PII before proceeding.
Improved compliance and security features were discussed, like pre-scan AI checks and metadata for clear specifications of use and permissions.
File Analysis Application
A team member shared an experience of using a tool that analyzes a large data set, cleans it, and generates relevant graphs for data visualization.
This suggests potential features such as the ability to upload a document and generate analyses directly in Mattermost.
Action Items
Consider future packaging and pricing for pay features like compliance features for file upload.
Upload summaries of AI meetings to YouTube text for better SEO and discoverability (Andrew Ziegler).
Keep file analysis on the backlog for future development (Christopher Speller).
For the next meeting, have someone other than the usual facilitator (in this case, Christopher Speller) handle the intro.
Summary generated using AI, and may contain inaccuracies. Do not take this summary as absolute truth.
The AI Mindmeld meeting for August 22nd, 2023, discussed the progress and updates on AI innovations at Mattermost. One of the main points discussed was the development of content extraction from file attachments. This has worked well with text files and is still in progress for PDFs. A recent change was made to the file size limits for uploads, allowing for the breaking down of long meeting recordings and improving the usability of the AI tool.
The meeting also discussed the idea of expanding the focus of these meetings to include other product development updates and demos from different team members. While the concept was generally supported, concerns were raised about the willingness of team members to participate and the potential overlap with current R&D meetings.
Another major point was the improvement of meeting transcriptions, specifically incorporating timestamps. This new feature was demoed and discussed, highlighting the ability to navigate to specific sections of the meeting recordings. The idea of having transcripts for any audio/video file in the system was also discussed as a potential feature. The meeting concluded agreeing to present these improvements and discussing possible collaborations with other teams.
Key Discussion Points
Updates on AI Innovation and Open Ops
Content extraction and File Attachments
Continued exploration on content extraction from file attachments.
Successful extraction from random text files.
Challenges experienced in content extraction from PDF files as current content extraction methods do not extract from PDF files appropriately.
Upgrades made to the file sizes for uploading enabling extraction of summaries from longer recordings using the AI tool.
Improvements to the Mattermost Meeting Experience
Inclusion of timestamps in the transcripts to enhance navigation through different sections of the meeting.
Timestamps in the extracted summaries will enable users to navigate directly to specific parts of a meeting. This is especially useful for quick briefings on next steps.
Transcript posting with timestamps is in progress.
Development of a transcript format file for every audio or video file in the system being considered. The file would include a preview of the video and clickable timestamps.
Proposal to Expand Meeting Scope
A proposition was made to invite individuals from different parts of the product team to speak on their upcoming features.
This proposition majorly aims to incidentally turn these sessions into a form of Mattermost demo stream.
The idea was well received, with recommendations to keep them separate from AI discussions.
Potential Action Items
Consider inviting product team members to share their upcoming features.
Share more content on AI and its impact on different sections of Mattermost.
Reservations and Lessons from Previous Attempts
Concerns about the willingness of team members to participate raised due to previous unsuccessful attempts at similar initiatives.
The dev weekly meeting currently serves a somewhat similar purpose.
Test run proposed to see effectiveness of proposed scope expansion.
Potential Action Items
Consider a test run of the proposed scope expansion.
Evaluate willingness of employees to participate in public demos before full implementation.
Improvements to Transcripts
Current improvements feature changes to the format of transcript display and inclusion of timestamps.
Proposition to make timestamps clickable for easier access to sections of interest.
Working on posting transcripts without the timestamp details for a cleaner look.
Possibility of creating an inclusive transcript format file being explored.
Potential Action Items
Make timestamps clickable for easy access.
Post transcripts without timestamp details to maintain clean look.
Continue work on creating an inclusive transcript format file.
Summary generated using AI, and may contain inaccuracies. Do not take this summary as absolute truth.
The team discussed several updates during their latest meeting which revolved around AI enhancements and future plans for their projects. Improvements in the timestamp feature for meeting notes were mentioned, with the feature now consistently accurate. A new developer engagement experience was announced, where an AI-powered matter most choose your own adventure game will be created in collaboration with GitLab. There were also discussions around the addition of prompts to assist in the execution of a playbook. A potential plan of having the AI understand the contents of uploaded files was pitched, though the current version does not support this yet. The topic of generating enormous amounts of data for OpenOps was discussed, this could potentially be done by creating AI-generated chat logs. The team also mentioned the associated issues with MySQL, revealing plans of migrating users to Postgres. The conversation concluded with plans of an upcoming release and a potential discussion around public sector AI during the World Economic Forum.
Key Discussion Points
Discussion Points
Meeting Introductions
The meeting commenced with general introductions and highlights of updates and advancements in AI.
Timestamps Update
The AI is improving in creating timestamps for meeting notes. Its accuracy now is more considerable and consistent.
A release is expected to be worked on to allow others to try the new feature.
GitLab Mattermost Bot
There are plans to collaborate with GitLab to create an AI-powered Mattermost āchoose your own adventureā game for developer engagement and tutorial purposes.
The bot might include a prompts library based on the discussions from the team-up brainstorming session.
Playbook Prompts Enhancement
The team discussed enhancing prompts usage in playbooks to improve access or push information.
The prompts would serve as AI suggestions on actions to be taken at different stages of playbook execution.
AI Plugin Content Understanding
The AI plugin is currently being developed to understand and extract content from both chat and file conversations.
There are thoughts on synthesizing data for conversation from chat logs for improved AI analysis and interaction.
MySQL Support
As MySQL presents quite a challenge in scaling and requires constant checks on both platforms, the default database is soon to be Postgres for all applications.
For OpenOps, usage of MySQL is temporarily discontinued until migration to Postgres is finalized.
OpenOps Update
OpenOps usage seems to be growing, despite it still being a sandbox.
There are plans to move OpenOps into production, hence the need to migrate MySQL to Postgres effectively.
Upcoming Meetings and Events
A GovTech summit by the World Economic Forum is anticipated, with a focus on public sector AI discussion.
Action Items
Work on the release of the updated AI timestamp feature.
Collaborate with GitLab on the Mattermost bot project.
Enhance playbook prompts to better assist the execution process.
Improve AI plugin to understand and extract content from both chat and file conversations.
Plan the migration from MySQL to Postgres to allow OpenOps usage.
Prepare for the upcoming World Economic Forumās GovTech summit.
Summary generated using AI, and may contain inaccuracies. Do not take this summary as absolute truth.
