“To err is human; to really foul things up requires machine learning” (with apologies to William Vaughan)
What if a software application like Salesforce (CRM), for example, could be awake alongside a person all day, interfacing and connecting to whatever telephone numbers are active for that person during work hours, and then at the end of the day, or some other agreed interval, present a list of interactions for approval and uploading into the company’s database?
Of course, in early days, the software would be learning and would likely make many mistakes.
A conversation with mom is not okay to add to the company’s dataset, but a conversation with the head channel partner should most certainly be added. Please. But the deletions, corrections, and edits by the human would help the machine learn quickly enough, and within a short period this kind of automation could reduce the work about work, freeing the human to just do the value-added things.
Think about it. We aren’t talking about magic. Salesforce would need to observe, distil, present, learn from the edits, and upload. The application would have to be supported by Machine Learning.
Salesforce is an easy example for work about work: speaking to a channel partner is value-add. Logging into Salesforce to upload the time/date of the conversation and some notes is not. Getting rid of the onerous work about work part is so obvious that it makes one wonder why Salesforce hasn’t already introduced this kind of interface to help us all unburden ourselves of the unnecessaries of our work day and improve our productivity and focus?
(Small prediction here: Salesforce won’t do it. The application is like an excel spread sheet hanging in web-space and locked into place circa 2001. It is a relic. And the company’s managers are too busy speaking gobbledygook such as ‘aligning values’, grease code for the boards of directors of customers to open the corporate wallet a bit further, rather than updating their product to next generation.)
But the point still stands. The elements needed to create an automated work process exist. Why haven’t software companies already done this?
Dropbox (DBX), for example, could have used its perch to do some amazing things. The company sees all the files you are working in, down to the line item, as well as those of your team, and has integrations built into virtually all productivity tools.
Why hasn’t Dropbox (or a litany of others like it) created an auto-populating project management tool, for example, that uses the systems’ knowledge to present projects, files, dates, deadlines, and conversations, that can be edited or muted by the human? ‘We notice your team is working on these three projects and you are working on a fourth. Our integration with Outlook indicates meetings in place for two of the projects but no deadlines for the others.’ Why not auto-populate all our to-do items into shared collaborations, visible to all teammates or none depending on the setting?
TECHNOLOGY GRIND
When humans create software, it often comes from an idea of some process that we wish to automate. It starts with a whiteboard, virtual or real, gets architected in brainstorming sessions, and then coded. It comes from a distilled idea inside the mind of the creators, and reflects humans trying to pattern and control the world around us. Idea, system, code, in that order.
But what if the process of creating software was to turn on its head? What if software creation in the future would start with a machine observing a human working? The order could be observe, record, distill, mirror, code, repeat, and automate. Rather than creating code from an a priori idea originating from a smart human, code would come into existence from observing the thing-in-itself, active in its native world, the patterns automated, and continuously adjusted as the machine learns.
Theology, philosophy, and theories of knowledge inflected along similar lines a few hundred years ago.
Previously, systems of knowledge were rational constructs, ideas of beautiful thought synthesized by a super smart human, postulated, and offered as a method to understand our world. The thing-in-itself was considered unknowable as a particular but still held a place as a referent inside a system of ideas.
As time and people progressed this approach faded in favor of directly observing and measuring the ding-an-sich (Kantian term for noumena). Observation, testing, and recording the thing-in-itself became the primary method of knowledge, and systems were built from layers of that understanding. Is software at a similar pivot point?
Rather than creating a software application from an idea of work, it can be created by observing people working. The process requires machine learning so that the application can evolve beyond raw observation, rely on a dance between human and system, followed by the interplay of system and other systems, to unlock new freedoms in work processes.
This change has already started. Already machine ‘observation’ can become code, a functioning software application, which opens the door to machine learning, automation, and the innovations that lie beyond. If we are right in exploring this open door to machine learning, it may imply that all software created before the pivot will appear like rigid, older knowledge systems that postulated, but did not observe. Relics, perhaps.
And the current fast growth era of software will appear slow, in hindsight, compared to what comes next as we unlock code creation and machine learning via observation. The combination of observation and automation with machine learning can lead to an acceleration of software creation that would make the size of the software industry right now appear puny, by comparison. Putting these elements together will also unlock the 'thing' that is beyond both automation & ML: the next playing field of innovation.
Immediate-term Risk Range™ Signal with @Hedgeye TREND signal in brackets:
UST 10yr Yield 1.44-1.58% (bullish)
UST 2yr Yield 0.22-0.30% (bullish)
SPX 4211-4345 (bullish)
RUT 2 (bullish)
NASDAQ 14,078-14,662 (bullish)
Tech (XLK) 142.04-148.99 (bullish)
Energy (XLE) 51.92-56.25 (bullish)
Financials (XLF) 35.23-37.63 (bullish)
Utilities (XLU) 62.54-64.70 (bearish)
Shanghai Comp 3 (neutral)
Nikkei 28,198-29,4429 (bullish)
DAX 15,405-15,801 (bullish)
VIX 13.60-18.92 (bearish)
USD 90.09-92.66 (bearish)
Oil (WTI) 71.71-75.40 (bullish)
Nat Gas 3.30-3.76 (bullish)
Gold 1 (bearish)
Copper 4.11-4.52 (bullish)
Silver 24.97-26.83 (neutral)
Ami Joseph
Technology Sector Head
