/w=3840,quality=90,fit=scale-down)
Prem SundaramOne of the things we like to do at Flomenco is try out new ways to improve our own internal company workflow and efficiency. In this spirit we will now blog from time to time these ‘UFO’ sightings - Using Flomenco Ourselves!
This last week I was chatting with Ryan, our CEO, and saying that it would be great to figure out how I can get a daily summary of the conversations in our company Slack ‘Technical’ channel. As a co-founder and the CCO (Chief Commercial Officer) I don’t have enough hours in the day to keep up with the volume of chat that is going on between our programming team on a daily basis - but being somewhat technically inclined I do want, actually need - to know ‘what is going on’ to keep me up to date. This of course helps from the commercial side and gives me more clarity about status of projects and general progress being made.
Having experimented with similar use cases on other projects, using AI to summarize content and provide me email daily updates, and knowing how useful this can be, we decided this last Friday afternoon to try and do this for ourselves using Flomenco.
TLDR? Yup, we got this done in a couple of hours and our workflow now provides all of us a daily and weekly update! It should have been pretty simple with our Flomenco no-code workflow tool, right? The answer is Yes … and No… but not for the reasons you might think. Read on to find out how we did it.
If you are a Flomenco customer we have made this a Slack Channel Summarization template for you to use out of the box to save you the time to figure this out!
This was the end-result of our efforts - a shiny new Weekly Report!
The Goal
Each day at 10am, and each Friday at 4pm, get all the messages in our ‘Technical’ channel, including all the threaded discussions, and then summarize this into a message of no more than 200 words. Bonus points, also grab the updates made to our code base in Gitlab and summarize the releases made.
The Process
- Create a Slack ‘App’ that will provide access to our message data
- Configure Flomenco Provider for Slack and Gitlab
- Configure the Flomenco Workflow
- Sit back and watch the summaries getting generated!
While this looked ‘easy,’ it turned out that even with our technical skill and the power of the Flomenco workflow platform, it still took us a minute to get this right - well, about 2 hours, truth be told. Certainly faster than trying to do it other ways without Flomenco, and honestly not bad for what it’s doing for us. We were quite happy with the result. But there are definitely some things to know if you want to get this right.
In this blog I won’t bore you with all the steps we did - as I mentioned we’re providing this template for our customers - but I do want to discuss some of the things that took a bit of thinking, much of which is getting to know the intricacies and idiosyncrasies of how Slack works, in case you try to do this yourself.
The Gotcha’s
Creating the Slack App with Scopes
So you probably need to work with your Slack ‘Admin’ to get this right - it’s not enough to just create a bot as a user (like me) and think it will be able to get all the Slack data - your Admin (in this case Ryan) needed to create an App and give it the right ‘scopes’. Scopes are the range of data that Slack can access. Be sure to include read messages and history scopes - which allow you access to all the message history. Maybe I was on the right track, but this was the first non-obvious Slack hurdle!
Message Threads
So once we connected our data and added our ‘FLO’ app to the channel, we had the ‘Cool - we’re getting data!’ moment, but we then realized that not all the ‘Replies’ to messages were being captured. Replies to a message form a ‘thread’ of conversation. But Slack treats these differently, so be warned.
You need to get the thread ID’s of messages that have reply threads and then do a query to get that message content. And then combine these together. Yeah, not so simple right?! Luckily, once we understood what Slack was doing, we could easily create this second workflow path in Flomenco and then aggregate the data, using the Aggregate node.
User Names not ID’s
This one is technically not a ‘Gotcha’ but rather more a bonus points option. Slack returns User ID, not the name of the user, but we wanted our summary to say things like ‘Ryan promoted Dev code to Production’ not ‘A user promoted Dev code to Production’ that the AI would do if it did not have the names with each message. So with a bit of magic Ryan created the logic in Flomenco to get the User Names and ID’s from Slack and then transform our data replacing User ID’s with User Names.
Once we had worked through these items we configured our AI node in Flomenco to send this data to ChatGPT and it then returned an excellent summary. It would take in the week of messages - say like 100 messages with threads - and return a weekly summary. As Ryan said ‘That’s a pretty good summary of what happened last week!’.
We then used the ‘Blocks’ feature of Slack to format our message better (this is also optional). Then feeling pretty chuffed with ourselves, Ryan proposed to also grab data from our Gitlab repository (we connected the node in Flomenco and it pulled the data) - and then we also added that information to the payload sent to AI for summarization. Now we have a weekly message that summarizes the week’s technical chat as well as a summary of the releases that occurred!
The Flomenco Slack Channel Summarization Workflow - Available as a Template for our Customers
What’s Next?
We use Linear for Dev ticket tracking so we’re going to be integrating that data also into our weekly summary. The Daily summary of the channel we’ll keep simple and have say a 100 word summary. This means our Dev team does not have to write a daily update - so one less task for them! Beyond Dev, we see possibilities to do a similar version for our Sales - to connect to our CRM and for us to know what items changed or updated this week at an executive level - and also maybe, just maybe, we could connect the sales and technical channels so that we can get the AI to monitor coordination between requests from Sales and the progress from Technical Development - and to highlight issues and opportunities. Oh, and then there’s the opportunity for a fully automated monthly report for our growing company to include tech, sales, finance and more!
What would you like to see on a daily, weekly or monthly report? Do let us know. We hope you found this useful and we’ll keep you up to date in our next UFO sighting! 🙂