Organizational Information Gain
Messy language models
Decision trees fit for purpose are kind of like actual trees in that they also consume information/water at the edge.
Plants need sunlight to make food, and decision trees need high quality information to make good decisions.
From an analytical perspective, leaves of the organizational tree represent cumulative “information gain”.
The more information the leaves have, the more "food" they can make, meaning higher-quality decisions, improved outcomes, and sustainable growth.
Automation of Broken Processes
In a municipality, enterprise SaaS, or services organization operating at scale, the digital gemba is mouse clicks, keyboard strokes, and voice conversations with client-facing colleagues in the “leaves” of the upside-down org chart.
In the leaves is where organizational values, priorities, and culture make “contact” with the market, colleagues, and prospective constituents of the platform and wider industrial ecosystem.
Any proposed AI strategy should attempt to empower customer-facing knowledge workers with better quality information at their fingertips. Rather than automate away the customer touch-point, humans benefitting from information gain can provide meaningfully better customer experiences.
Why replace humans with AI when you can empower them with better info to provide better customer experiences?
Human capital is the enterprise’s greatest asset (and liability and variable cost and talent recruiter).
Human capital has critical thinking skills, confidence, and a sense of self.
But, nobody performs perfectly every day and things can get messy.
After 3 decades in search of excellence and digital darwinism, digital skin and data muscle products were built by Brand Men for Industrial Era processes. The workflows which were fit for purpose 100 years ago.
As a result, in the Normal Now™ ennui can be pervasive.
For anyone who's ever pressed "0" to speak with the next available representative, sometimes you just want to talk to someone.
Content Processing Tsunami
At least one AI app is enabling humans to create personalized marketing content. From an audience perspective, the output of the process looks like a tidal wave of spam at scale.
The obverse of machine-augmented content generation is machine-augmented content consumption.
I've yet to come across a Sentieo-like AI app which ingests hours of podcasts, YouTube videos, or recorded webinars to identify themes of interest, and specific critical take-aways. To date, that content would need to be transcribed into text form first before ingestion by ChatGPT.
Soon, inevitably, there will be AI tools to sift through the "content" created by other AI tools:
Status Quo
Humans are engaged with dozens of collaboration apps and platforms every day.
The objective is to be "social" using technology as a "medium" to communicate. Each platform has interfaces to seamlessly connect one to another.
Expectations of software increasing isn’t just for consumers.
Enterprise expectations change, as colleagues operate like meat-based APIs engaging with dozens of SaaS products / microservices in order to kick off and manage “jobs to be done”.
Empowered Colleagues Enable Excellence
Whenever the next available representative says, “Sorry, the system is being slow” on a recorded line, I bite my tongue.
I usually resist telling them that’s because management probably had a business scorecard to manage and optimize thresholds for systems used by customer support.
Teams of people were purposefully incentivized to align on KPIs at which the business could weather customer and employee churn and rather than optimize for information gain in the leaves.
When there is threat of a drought, information (and cash flows) are optimized to retain in the roots.
Inflation is the opposite of a rainy day in this analogy.
Inflation is heat death for money supply and information quality.
Teams of executive stakeholders had multiple PowerPoint presentations to align on the conclusion that customer-facing staff could manage without. A decision was made that other priorities were more valuable than outfitting colleagues with performant tooling to better engage with customers.
Resources are finite and there were other priorities besides helping customers get jobs done.
Sometimes, I can’t resist and they get a wave of service design consulting “content” in their ear.
Calculator for Words
Simon Willison described large language models as "a calculator for words".
A model could, for instance, read through a stack of emails one by one in order to assess their relevance to a question, and then instruct itself to condense the most relevant emails into a summary.
One strategy: identify curious critical thinkers eager to engage in an ongoing dialectic of messy thoughts.
In a municipality, enterprise SaaS, or services organization, the human capital that’s endured the most and earned the most stripes has learned the most lessons of multiple failed digital transformation initiatives.
User expectations (and tolerances for wonkiness) will evolve even more very soon.
However, the business objective remains simple and universal: generate as much excess cash flow as possible using the minimum amount of capital available.
Like any investment thesis, you get what you don’t pay for.
After Governance, Risk, and Compliance-approved guardrails are in place, colleagues meeting specific criteria should be equipped with tooling to optimize information gain of human capital across the enterprise.
The criteria:
most psychologically safe
most emotionally resilient
most comfortable with ambiguity
Product-Market Hypothesis
Enterprise ecosystems can achieve sustainable growth by improving coherence, productivity, and the cash value of leaves in the org tree.
Wisdom is the ability to look forward and backward, to see both sides of life with no illusions.
— Dr. Alexander Lowen
Backpropagation allows neural networks to be applied to a much wider field of problems that were previously off-limits due to time and cost constraints.
Meanwhile, transform the org chart into an information-rich, machine-augmented decision tree.
There is an opportunity for real transformative change, rather than “helping Twitter hustlebros make SEO filler text 50% more mind-numbing.”
