How we validated the ChatOps concept in two days using a prototype

Decision-making to touch - why PowerPoint slides are responsible for wrong decisions and why prototypes bring light into the dark.

Autor: Cyrill Rüttimann

Powerpoint slides are not only responsible for the Columbia disaster, but also for many wrong decisions in projects. For the decision basis ChatOps - Flop or Top - we have therefore created a prototype to touch. 

We make far-reaching decisions based on PowerPoint slides. However, the implementation then usually does not work as promised. But the decision is made and the project framework is outlined. Often a stop of the project is not an option.

We chose a different path. To evaluate the concept of chatOps, we built a prototype. We can touch it with our hands. We can understand it literally. Researchers discovered years ago (Sutton-Smith, Paperts, Vygotsky, Fein) that thought processes in connection with haptic perception are more profound and sustainable for our brain.

Prototypes bring light into the dark

The ipt Hackathon 2016 was the perfect opportunity for our project. Already on the first day we were able to create our Minimum Viable Product (MVP), following the recipe of Eric Ries («The Lean Startup»). In the remaining time we extended the prototype in several short cycles. We oriented ourselves to the OODA loop: Observe, Orient, Decide and Act. This strategy enabled us to create a functioning and high-quality prototype within the set time frame.

Figure 1: The OODA loop is a strategic concept. The concept supports decision-making in rapidly changing situations.

Lazy Deploy - Deployment of software from physical Kanban board

The cards on the kanban board are equipped with an RFID chip. If they are moved to the status «completed», this information is written to the chat. Our bot evaluates this information, then compiles and installs the software. It gives corresponding instructions to the IT services, which carry out these steps. The bot also controls lamps that inform people about the processes by visual signals (green lamp flashes).

The IT services in turn write the result in the chat. In case of an error, the bot evaluates the information available to it (log files, digital interfaces). If the error is a known pattern (machine learning), it automatically initiates the appropriate measures or informs the person via chat about possible actions.

Figure 2: The prototype to touch. Built up with the simplest means from the household in a few minutes.

Integration redefined - insights from the prototype

We were skeptical at first. Continuous delivery and system integration via a chat that doesn't know formalized data and mediation? The development of the prototype gave us a good insight, which we were able to demonstrate on site:

The fast integration of IT services via chat and bot convinced us immediately. New ideas were constantly emerging, which we incorporated into the prototype. Our favourite: The ability to learn from past mistakes and to react intelligently to them. This type of artificial intelligence acts as an «enabler» for new forms of integration and interaction - right up to natural entertainment with IT tools.

Understanding instead of analysing

Besides a lot of fun among colleagues, we made two important findings at the ipt Hackathon:

  1. ChatOps has potential.
  2. Without a prototype this insight would have been missing or less clear!

Traditionally, we would have formed our opinion on the basis of an internet research, a presentation and the subsequent discussion - many facts without practical experience. Thanks to our prototype, we are much better able to quantify and argue the effort, risks and benefits. This not only helps to prevent catastrophes in space travel, but also everyday ones - from missed opportunities to unnecessary waste of budget.