Jarvis is getting smarter and how our AI assistant in Slack learned to manage weekly project plans

Let us remind you that earlier at Skylex we already talked about our virtual assistant Jarvis. This invisible AI assistant helps us automatically transform the chaos of work conversations into clear tasks and maintain an internal database, saving every important thought of the developers. The bot simply stays in our work chats, listens carefully to discussions, structures conversations, and generates ready-made suggestions for the task manager. This allows us to maintain a continuous and fast exchange of information between teams, while managers no longer have to collect reports piece by piece every evening and fill out tables manually.

But we are not stopping there. Recently, we added an interesting and very useful update to Jarvis. Now it can independently manage project work plans, collect daily updates from the team, and generate an up-to-date plan. This update fundamentally changes how we coordinate processes, as the routine work of collecting statuses has completely shifted to artificial intelligence.

Let's look into how this new feature works in plain language, how it makes life easier for the team, and why it makes process coordination transparent and clear for every participant.

Why chat-based planning often turns into chaos

In any company, planning usually looks like this. On Monday, the team gathers for a discussion and plans tasks, and then throughout the week the status of the work constantly changes. Someone writes about updates in the general chat, someone clarifies technical details in direct messages, and some important decisions get completely buried in the stream of daily chatter. Finding the loose ends afterwards is simply impossible, and the current picture of the project gets lost in a pile of messages again.

A manager has to run around chats, gather information from different conversation branches (which we call threads), and manually copy everything into a report. This takes a huge amount of time that could be spent solving more complex strategic tasks. In addition, when information is collected manually, there is always a risk of missing something or misunderstanding a developer's technical comment.

To forget about this routine forever, we taught Jarvis to independently monitor plans and collect work statuses right in Slack. Now the human factor is completely eliminated from the process of collecting intermediate reporting.

How Jarvis manages plans and collects updates

The new operating algorithm of our AI assistant is structured very simply and logically. We designed the system so that it does not interfere with developers writing code and does not distract them with constant queries throughout the day, while simultaneously keeping a pulse on every task. The entire process is divided into three major stages that work automatically in the background mode.

1. Storing the plan for each project

Jarvis now has access to a special section of our internal database where the current work plan for each active project is recorded. It clearly knows which tasks the team must complete in the near future, what priorities the manager set, and what deadlines are assigned to each development stage. This is full contextual understanding of the entire project based on direct synchronization with our CRM database.

The AI assistant stores the task structure and daily compares it with the team's actual progress visible in the chats. If a task changes or new details appear, Jarvis instantly enters these edits into its internal project map.

2. Collecting and analyzing updates through the database

To collect up-to-date statuses, we implemented a reliable data storage architecture. All messages, details, and work discussions of developers from Slack are automatically saved to our internal database. Each channel in Slack is clearly linked to a separate project in the CRM, so the artificial intelligence always precisely understands the context of the conversation and knows which exact task a particular message belongs to.

Next, Jarvis comprehensively analyzes all stored messages in the database together with the overall project plan and the list of active tasks. For the algorithm to draw accurate and logical conclusions, we prepared a number of specialized prompts. The AI algorithm extracts task context from correspondence, separates work statuses from regular chatter, and records actual execution progress. For example, when a comment about successfully configuring a database or API readiness appears in the database, Jarvis automatically matches this information with the current plan and updates the task status.

3. Daily report requests and operational plan adjustments

Reporting now happens daily rather than once a week. Exactly 30 minutes before the end of the workday, Jarvis sends an automatic status update request to the relevant Slack channels.

Upon receiving new comments from developers, the system immediately analyzes the provided information, checks it against the current schedule, and automatically draws conclusions about the pace of work. Based on this daily analysis, Jarvis instantly makes operational changes to project plans. This allows the team and the manager to see the real state of affairs every evening and start a new workday with a clear understanding of current priorities without unnecessary sync calls and long status updates.

Fixing a new plan with a single command

We perfectly understand that software development is a living process. There are moments when plans change radically. For example, after a strategic call with a client, it turns out that priorities urgently need to shift, a new feature needs to be added, or current tasks must be postponed to urgently fix bugs. For such cases, we added the ability to quickly set a completely new benchmark for our artificial intelligence.

Now there is a simple and convenient command /plan in Slack. When a manager enters it in the relevant project channel, Jarvis immediately switches to new plan recording mode. The manager can write an updated task list directly in the chat in free format, without using complex templates or special formatting.

Jarvis instantly reads this text, updates the information in the database, and will subsequently conduct daily analysis and generate reports based on this new plan. This allows the team to stay flexible and instantly adapt to any changes without long paperwork or reconfiguring task managers.

How custom AI solutions help grow business

The example of our Jarvis clearly demonstrates why custom development of automation tools is far more effective than using ready-made standard bots available on the market. No off-the-shelf template builder is capable of integrating so deeply into internal company processes and understanding unique development business logic.

When we build tools for our own needs, we get full freedom of action. We can teach the AI to think exactly as our managers need, connect it to any databases, and not overpay for every new user or message. This is a long-term investment that becomes smarter with every update, takes on more routine tasks, and allows the team to focus exclusively on creating quality code and developing projects.

Conclusion

Thanks to this extraordinary Jarvis update, we took another important step toward deep project management automation. We continue to eliminate the need to manually keep heaps of reports, fill out endless spreadsheets, and constantly distract developers with questions about how work on tasks is progressing.

Now our Slack works as a single living ecosystem. The team simply discusses tasks in threads, calmly does their work, and communicates in chats, while artificial intelligence gathers this data, neatly organizes it, updates statuses in the database, and performs daily plan analysis. This makes work transparent for clients, calm for managers, and maximally productive for developers.