Bot Technology: A Procurement Experts Thoughts
The ‘app boom’ is widely recognized to be slowing as we approach the half way mark of 2017. Success stories such as Snapchat and Uber remain (in terms of continued, steep growth), but the aggregate growth in the app market has started to decline for the first time Apple introduced the App Store in 2008. The truth is, most people have already downloaded all the apps they need. The market is already saturated with apps that satisfy our basic needs: travel/directions, calendars, messaging, social media, gaming, news, weather, etc. This fact is well known by tech giants such as Facebook and their eyes are already on the next opportunity: bot technology.
Bot technology, or a ‘web robot’, is a type of software application that runs on scrips over the internet. Companies are exploring this technology aggressively, especially for the integration of artificial intelligence. For example, Siri –Apple’s intelligent assistant - pulls data from the internet such as weather and information stored on your phone in calendars and contacts.
Some other common bots you may have used include:
- Transport for London, or TfL, which runs through the Facebook messenger app. London’s public transportation department offers the service on the Facebook Messenger app which captures data on train and tram service times, including delays and disruptions, and relays those messages via the bot to end users.
- The Dominos bot service, or ‘DOM’ the pizza bot. In 2016, Dominos introduced this bot to make ordering pizza easy and faster by having users set up their pizza preferences before messaging “Pizza” to DOM on the Facebook messenger app to get a pizza auto delivered to their address.
- Heston Bot is another recently developed program which runs on Skype and produces recommendations based on user questions regarding cooking and ingredient tips.
You’ll start to notice more and more of these applications popping up as tech companies invest in trending bot technology. Facebook in particular has been capitalizing on the rising popularity and demand for bot technology -developing over 100,000 bots for a variety of functions. Many run on the messenger app and capitalize on the commercial selling power that the communication tool is able to leverage.
In recent history, bots have operated on a combination of natural language learning and machine learning. Facebook’s Artificial Intelligence Research group (FAIR) has been working on an artificial intelligence module that enables bots to successfully complete negotiations - a significant advancement in bot technology. With the sheer amount of data at Facebook’s fingertips, along with their cutting edge AI group, they have the potential to pave the way for some very interesting technology applications. Facebook and its FAIR group recently pitted bot versus bot and human versus bot in a series of experimental mock negotiation trials where bots proved the ability to effectively negotiate.
The mock AI bot negotiations gave each bot their own value function based on a set of common items and instructed each bot, or ‘agent’, to divide the items up with the goal of maximizing value. A series of mock negotiations were conducted where the dialogue history and associated outcomes were captured and input back to the agents, effectively enabling them to anticipate future outcomes. Once this was complete, the agents were able to negotiate better and accept deals less quickly. One of the most significant findings were isolated cases where bots initially vested interest in items they did not value (based on the value function input), and later compromised or sacrificed that item to the other party. In other words, the bots were learning common negotiations tactics used today in real life negotiations.
This specific case of research and subsequent advancement in bot technology could have some drastic effects on the business world if harnessed and industrialized in an effective way. Applications of this technology could even fit into the procurement space. E-Sourcing could certainly benefit from automating negotiations. E-sourcing currently automates the RFP administration and bid evaluation process, however oftentimes that is where the automated buck stops and the human element takes over. Final negotiations and decision making rest in the hands of the e-sourcing tool’s user(s).
Bot technology could potentially play a role in the negotiation process for more tactical categories such as office supplies where price is the primary decision driver and qualitative components such as lead times and product quality play a less significant role. We’ve already seen more and more automation in the RFP and procurement processes in e-sourcing and ERP technology. Again, I don’t see a place for this technology in the sourcing process for strategic or highly complex categories, however I wouldn’t be surprised to see bot technology in the procurement process for tactical spend categories very soon.
I agree that bots will soon take control of tactical sourcing projects and the strategic sourcing projects pose more challenges. However, the strategic sourcing projects are also where the most significant gains can be made. There are several reasons why enterprises have a stronger incentive to apply AI and bots to higher value strategic sourcing events. For starters, AI can be trained to apply optimization techniques more effectively, secondly AI can also reason more effectively about larger and more complex datasets. Sourcing Robotics has already been rolled out for $100m+ sourcing events very successfully. There's a white paper on the subject that might be of interest here: https://keelvar.com/white-paper-download/