The FRICT Method: A Not-Quite-Random Way to Spot Automation Gold

February 23, 2026 / lbhuston / Leave a comment

There’s a certain kind of exhaustion that doesn’t come from hard problems.

It comes from repeated problems.

The kind you’ve solved before. The kind you’ll solve again tomorrow. The kind that makes you think, “Why am I still doing this by hand?”

Over the past few years—whether in cybersecurity operations, advisory work, or just wrangling my own digital life—I’ve noticed something: most people don’t struggle to build automation.

They struggle to choose the right things to automate.

A mental model can be used to develop strategies for achieving goals By understanding how different parts of a system interact strategies can be created that take advantage of synergies and identify areas where improvements are needed 3981588

So here’s a methodology I’ve been refining. It’s practical. It’s testable. And it’s surprisingly reliable.

I call it FRICT.

Step 1: Run the FRICT Filter

Before you automate anything, run it through this filter.

If a task is:

Frequent (weekly or more often)
Rules-based (clear decision criteria)
Information-moving (copy/paste, reformatting, summarizing, transforming)
Checklist-driven (same steps each time)
Templated (same structure, different inputs)

…it’s a strong automation candidate.

Why This Works

High leverage tends to live inside repeated, structured work.

Think about your week:

Generating recurring reports
Moving data between systems
Creating customer follow-ups
Reviewing logs for defined patterns
Reformatting notes into documentation

These aren’t “hard” problems. They’re structured problems. And structured problems are automation-friendly by nature.

In cybersecurity operations, we’ve seen this repeatedly. Log triage. Ticket enrichment. Asset tagging. Compliance evidence collection. They’re not intellectually trivial—but they are structured.

And structure is oxygen for automation.

The Caveat

Some frequent tasks still require deep contextual judgment. Executive communications. Incident response war rooms. Strategic advisory decisions.

Those may be frequent—but they’re not always safely automatable.

FRICT gets you to the right neighborhood. It doesn’t mean you bulldoze the house.

Step 2: Score Before You Build

This is where most people go wrong.

They automate what’s annoying, not what’s valuable.

Before building anything, score the candidate task across five axes, 0–5 each:

Time saved per month
Error reduction
Risk if wrong (invert this—lower is better)
Data access feasibility
Repeatability

Then use this formula:

(Time + Error + Repeatability + Feasibility) − Risk ≥ 10

If it scores 10 or higher, it’s worth serious consideration.

Why This Works

This forces you to think in terms of:

ROI
Operational safety
Feasibility
System access realities

In security consulting, we’ve learned this lesson the hard way. Automating the wrong control can introduce more risk than it removes. Automating something that saves 20 minutes a month but takes 12 hours to build? That’s hobby work, not leverage.

This scoring model prevents premature enthusiasm.

It also forces you to confront a truth:

Just because something is automatable doesn’t mean it’s worth automating.

A Quick Example

Let’s say you generate a weekly client status report.

FRICT check:

Frequent? ✔ Weekly
Rules-based? ✔ Same metrics
Information-moving? ✔ Pulling data from systems
Checklist-driven? ✔ Same sections
Templated? ✔ Same structure

Score it:

Time saved/month: 4
Error reduction: 3
Risk if wrong: 2
Data feasibility: 4
Repeatability: 5

Formula:

(4 + 3 + 5 + 4) − 2 = 14

That’s automation gold.

Now compare that to “automate strategic roadmap planning.”

FRICT? Weak.
Score? Probably low repeatability, high risk.

That’s a human job.

The Subtle Insight: Automation Is Risk Management

In cybersecurity, we obsess over reducing human error.

But here’s the uncomfortable truth:

Most organizations still rely heavily on manual, repetitive, error-prone workflows.

Automation isn’t about convenience.

It’s about:

Reducing variance
Increasing consistency
Making controls measurable
Freeing human judgment for non-templated work

The irony? The more strategic your role becomes, the more your value depends on eliminating the structured tasks beneath you.

FRICT helps you find them.

The scoring model helps you prioritize them.

Together, they create something better than random automation experiments.

They create a system.

What This Looks Like in Practice

If you want to apply this method this week:

List every recurring task you do for 7 days.
Mark the ones that pass FRICT.
Score the top five.
Only build the ones that cross the ≥10 threshold.
Re-evaluate quarterly.

You’ll be surprised how quickly this surfaces 2–3 high-leverage opportunities.

And here’s the part people don’t expect:

Once you start doing this intentionally, you begin redesigning your work to be more automatable.

That’s when things get interesting.

The Contrary View

There’s one important caveat.

Some strategic automations score low at first—but unlock long-term leverage.

Examples:

Building a normalized data model
Creating unified dashboards
Establishing an API integration layer

They may not immediately score ≥10.

But they create compounding effects.

That’s where experience comes in. Use the formula as a guardrail—not a prison.

Final Thought: Automate the Machine, Not the Mind

If you automate everything, you lose your edge.

If you automate nothing, you waste your edge.

The sweet spot is this:

Automate the predictable.
Protect the contextual.
Elevate the human.

FRICT isn’t magic.

But it’s not random either.

And in a world racing toward AI-first everything, having a disciplined way to decide what should be automated may be the most valuable skill of all.

Method Summary

FRICT Filter
Frequent + Rules-based + Information-moving + Checklist-driven + Templated

Scoring Formula
(Time + Error + Repeatability + Feasibility) − Risk ≥ 10

Now I’m curious:

What’s one task you’ve been doing repeatedly that probably shouldn’t require your brain anymore?

* AI tools were used as a research assistant for this content, but human moderation and writing are also included. The included images are AI-generated.

Building a Graph-First RAG Taught Me Where Trust Actually Lives With LLMs

January 14, 2026 / lbhuston / Leave a comment

I didn’t build this because I thought the world needed another RAG framework.

I built it because I didn’t trust the answers I was getting—and I didn’t trust my own understanding of why those answers existed.

ChatGPT Image Jan 14 2026 at 04 07 59 PM

Reading about knowledge graphs and retrieval-augmented generation is easy. Nodding along to architecture diagrams is easy. Believing that “this reduces hallucinations” is easy.

Understanding where trust actually comes from is not.

So I built KnowGraphRAG, not as a product, but as an experiment: What happens if you stop treating the LLM as the center of intelligence, and instead force it to speak only from a structure you can inspect?

Why Chunk-Based RAG Breaks Down in Real Work

Traditional RAG systems tend to look like this:

Break documents into chunks
Embed those chunks
Retrieve “similar” chunks at query time
Hand them to an LLM and hope it behaves

This works surprisingly well—until it doesn’t.

The failure modes show up fast when:

you’re using smaller local models
your data isn’t clean prose (logs, configs, dumps, CSVs)
you care why an answer exists, not just what it says

Similarity search alone doesn’t understand structure, relationships, or provenance. Two chunks can be “similar” and still be misleading when taken together. And once the LLM starts bridging gaps on its own, hallucinations creep in—especially on constrained hardware.

I wasn’t interested in making the model smarter.
I was interested in making it more constrained.

Flipping the Model: The Graph Comes First

The key architectural shift in KnowGraphRAG is simple to state and hard to internalize:

The knowledge graph is the system of record.
The LLM is just a renderer.

Under the hood, ingestion looks roughly like this:

Documents are ingested whole, regardless of format
- PDFs, DOCX, CSV, JSON, XML, network configs, logs
They are chunked, but chunks are not treated as isolated facts
Entities are extracted (IPs, orgs, people, hosts, dates, etc.)
Relationships are created
- document → chunk
- chunk → chunk (sequence)
- document → entity
- entity → entity (when relationships can be inferred)
Everything is stored in a graph, not a vector index

Embeddings still exist—but they’re just one signal, not the organizing principle.

The result is a graph where:

documents know what they contain
chunks know where they came from
entities know who mentions them
relationships are explicit, not inferred on the fly

That structure turns out to matter a lot.

What “Retrieval” Means in a Graph-Based RAG

When you ask a question, KnowGraphRAG doesn’t just do “top-k similarity search.”

Instead, it roughly follows this flow:

Extract entities from the query
- Not embeddings yet—actual concepts
Anchor the search in the graph
- Find documents, chunks, and entities already connected
Traverse outward
- Follow relationships to build a connected subgraph
Use embeddings to rank, not invent
- Similarity helps order candidates, not define truth
Expand context deliberately
- Adjacent chunks, related entities, structural neighbors

Only after that context is assembled does the LLM get involved.

And when it does, it gets a very constrained prompt:

Here is the context
Here are the citations
Do not answer outside of this

This is how hallucinations get starved—not eliminated, but suffocated.

Why This Works Especially Well with Local LLMs

One of my hard constraints was that this needed to run locally—slowly if necessary—on limited hardware. Even something like a Raspberry Pi.

That constraint forced an architectural honesty check.

Small, non-reasoning models are actually very good at:

summarizing known facts
rephrasing structured input
correlating already-adjacent information

They are terrible at inventing missing links responsibly.

By moving correlation, traversal, and selection into the graph layer, the LLM no longer has to “figure things out.” It just has to talk.

That shift made local models dramatically more useful—and far more predictable.

The Part I Didn’t Expect: Auditability Becomes the Feature

The biggest surprise wasn’t retrieval quality.

It was auditability.

Because every answer is derived from:

specific graph nodes
specific relationships
specific documents and chunks

…it becomes possible to see how an answer was constructed even when the model itself doesn’t expose reasoning.

That turns out to be incredibly valuable for:

compliance work
risk analysis
explaining decisions to humans who don’t care about embeddings

Instead of saying “the model thinks,” you can say:

these entities were involved
these documents contributed
this is the retrieval path

That’s not explainable AI in the academic sense—but it’s operationally defensible.

What KnowGraphRAG Actually Is (and Isn’t)

KnowGraphRAG ended up being a full system, not a demo:

Graph-backed storage (in-memory + persistent)
Entity and relationship extraction
Hybrid retrieval (graph-first, embeddings second)
Document versioning and change tracking
Query history and audit trails
Batch ingestion with guardrails
Visualization so you can see the graph
Support for local and remote LLM backends
An MCP interface so other tools can drive it

But it’s not a silver bullet.

It won’t magically make bad data good.
It won’t remove all hallucinations.
It won’t replace judgment.

What it does do is move responsibility out of the model and back into the system you control.

The Mindset Shift That Matters

If there’s one lesson I’d pass on, it’s this:

Don’t ask LLMs to be trustworthy.
Architect systems where trust is unavoidable.

Knowledge graphs and RAG aren’t a panacea—but together, they create boundaries. And boundaries are what make local LLMs useful for serious work.

I didn’t fully understand that until I built it.

And now that I have, I don’t think I could go back.

Support My Work

Support the creation of high-impact content and research. Sponsorship opportunities are available for specific topics, whitepapers, tools, or advisory insights. Learn more or contribute here: Buy Me A Coffee

**Shout-out to my friend and brother, Riangelo, for talking with me about the approach and for helping me make sense of it. He is building an enterprise version with much more capability.

Your First AI‑Assisted Research Project: A Step‑by‑Step Guide

January 8, 2026 / lbhuston / Leave a comment

Transforming Knowledge Work from Chaos to Clarity

Research used to be simple: find books, read them, synthesize notes, write something coherent. But in the era of abundant information — and even more abundant tools — the core challenge isn’t a lack of sources; it’s context switching. Modern research paralysis often results from bouncing between gathering information and trying to make sense of it. That constant mental wrangling drains our capacity to think deeply.

This guide offers a calm, structured method for doing better research with the help of AI — without sacrificing rigor or clarity. You’ll learn how to use two specialized assistants — one for discovery and one for synthesis — to move from scattered facts to meaningful insights.

Unnamed 3

1. The Core Idea: Two Phases, Two Brains, One Workflow

The secret to better research isn’t more tools — it’s tool specialization. In this process, you separate your work into two clearly defined phases, each driven by a specific AI assistant:

Phase	Goal	Tool	Role
Discovery	Find the best materials	Perplexity	Live web researcher that retrieves authoritative sources
Synthesis	Generate deep insights	NotebookLM	Context‑bound reasoning and structured analysis

The fundamental insight is that searching for information and understanding information are two distinct cognitive tasks. Conflating them creates mental noise that slows us down.

2. Why This Matters (and the AI Context)

Before we dive into the workflow, it’s worth grounding this methodology in what we currently know about AI’s real impact on knowledge work.

Recent economic research finds that access to generative AI can materially increase productivity for knowledge workers. For example:

Workers using AI tools reported saving an average of 5.4% of their work hours — roughly 2.2 hours per week — by reducing time spent on repetitive tasks, which corresponds to a roughly 1.1% increase in overall productivity.
Field experiments have shown that when knowledge workers — such as customer support agents — have access to AI assistants, they resolve about 15% more issues per hour on average.
Empirical studies also indicate that AI adoption is broad and growing: a majority of knowledge workers use generative AI tools in everyday work tasks like summarization, brainstorming, or information consolidation.

Yet, productivity is not automatic. These tools augment human capability — they don’t replace judgment. The structured process below helps you keep control over quality while leveraging AI’s strengths.

3. The Workflow in Action

Let’s walk through the five steps of a real project. Our example research question:
What is the impact of AI on knowledge worker productivity?

Step 1: Framing the Quest with Perplexity (Discovery)

Objective: Collect high‑quality materials — not conclusions.

This is pure discovery. Carefully construct your prompt in Perplexity to gather:

Recent reports and academic research
Meta‑analyses and surveys
Long‑form PDFs and authoritative sources

Use constraints like filetype:pdf or site:.edu to surface formal research rather than repackaged content.

Why it works: Perplexity excels at scanning the live web and ranking sources by authority. It shouldn’t be asked to synthesize — that comes later.

Step 2: Curating Your Treasure (Human Judgment)

Objective: Vet and refine.

This is where your expertise matters most. Review each source for:

Recency: Is it up‑to‑date? AI and productivity research moves fast.
Credibility: Is it from a reputable institution or peer‑reviewed?
Relevance: Does it directly address your question?
Novelty: Does it offer unique insight or data?

Outcome: A curated set of URLs and a Perplexity results export (PDF) that documents your initial research map.

Step 3: Building Your Private Library in NotebookLM

Objective: Upload both context and evidence into a dedicated workspace.

What to upload:

Your Perplexity export (for orientation)
The original source documents (full depth)

Pro tip: Avoid uploading summaries only or raw sources without context. The first leads to shallow reasoning; the second leads to incoherent synthesis.

NotebookLM becomes your private, bounded reasoning space.

Step 4: Finding Hidden Connections (Synthesis)

Objective: Treat the AI as a reasoning partner — not an autopilot.

Ask NotebookLM questions like:

Where do these sources disagree on productivity impact?
What assumptions are baked into definitions of “productivity”?
Which sources offer the strongest evidence — and why?
What’s missing from these materials?

This step is where your analysis turns into insight.

Step 5: Trust, but Verify (Verification & Iteration)

Objective: Ensure accuracy and preserve nuance.

As NotebookLM provides answers with inline citations, click through to the original sources and confirm context integrity. Correct over‑generalizations or distortions before finalizing your conclusions.

This human‑in‑the‑loop verification is what separates authentic research from hallucinated summaries.

4. The Payoff: What You’ve Gained

A disciplined, AI‑assisted workflow isn’t about speed alone — though it does save time. It’s about quality, confidence, and clarity.

Here’s what this workflow delivers:

Improvement Area	Expected Outcome
Time Efficiency	Research cycles reduced by ~50–60% — from hours to under an hour when done well
Citation Integrity	Claims backed by vetted sources
Analytical Rigor	Contradictions and gaps are surfaced explicitly
Cognitive Load	Less context switching means less burnout and clearer thinking

By the end of the process, you aren’t just informed — you’re oriented.

5. A Final Word of Advice

This structured workflow is powerful — but it’s not a replacement for thinking. Treat it as a discipline, not a shortcut.

Keep some time aside for creative wandering. Not all insights come from structured paths.
Understand your tools’ limits. AI is excellent at retrieval and pattern recognition — not at replacing judgment.
You’re still the one who decides what matters.

Conclusion: Calm, Structured Research Wins

By separating discovery from synthesis and assigning each task to the best available tool, you create a workflow that’s both efficient and rigorous. You emerge with insights grounded in evidence — and a process you can repeat.

In an age of information complexity, calm structure isn’t just a workflow choice — it’s a competitive advantage.

Apply this method to your next research project and experience the clarity for yourself.

Support My Work

* AI tools were used as a research assistant for this content, but human moderation and writing are also included. The included images are AI-generated.

Future Brent – A Mental Model: A 1% Nudge Toward a Kinder Tomorrow

December 9, 2025December 9, 2025 / lbhuston / Leave a comment

On Not Quite Random, we often wander through the intersections of the personal and the technical, and today is no different. Let me share with you a little mental model I like to call “Future Brent.” It’s a simple yet powerful approach: every time I have a sliver of free time, I ask, “What can I do right now that will make things a little easier for future Brent?”

ChatGPT Image Dec 9 2025 at 10 23 45 AM

It’s built on three pillars. First, optimizing for optionality. That means creating flexibility and space so that future Brent has more choices and less friction. Second, it’s about that 1% improvement each day—like the old adage says, just nudging life forward a tiny bit at a time. And finally, it’s about kindness and compassion for your future self.

Just the other day, I spent 20 minutes clearing out an overcrowded closet. That little investment meant that future mornings were smoother and simpler—future Brent didn’t have to wrestle with a mountain of clothes. And right now, as I chat with you, I’m out on a walk—because a little fresh air is a gift to future Brent’s health and mood.

In the end, this mental model is about blending a bit of personal reflection with a dash of practical action. It’s a reminder that the smallest acts of kindness to ourselves today can create a more flexible, happier, and more empowered tomorrow. So here’s to all of us finding those little 1% opportunities and giving future us a reason to smile.

Hybrid Work, Cognitive Fragmentation, and the Rise of Flow‑Design

December 8, 2025 / lbhuston / Leave a comment

Context: Why hybrid work isn’t just a convenience

Hybrid work isn’t a fringe experiment anymore — it’s quickly becoming the baseline. A 2024–25 survey in the U.S. shows that 52% of employees whose jobs can be remote work in a hybrid mode, and another 27% are fully remote.

Other recent studies reinforce the upsides: hybrid arrangements often deliver similar productivity and career‑advancement outcomes as fully on-site roles, while improving employee retention and satisfaction.

Redheadcoffee

In short: hybrid work is now normal — and that normalization brings new challenges that go beyond “working from home vs. office.”

The Hidden Cost: Cognitive Fragmentation as an Engineering Problem

When organizations shift to hybrid work, they often celebrate autonomy, flexibility, and freedom from commutes. What gets less attention is how hybrid systems — built around multiple apps, asynchronous communication, decentralized teams, shifting time zones — cause constant context switching.

Each time we jump from an email thread to a project board, then to a chat, then to a doc — that’s not just a change in window or tab. It is a mental task switch.
Such switches can consume as much as 40% of productive time.
Beyond lost time, there’s a deeper toll: the phenomenon of “attention residue.” That’s when remnants of the previous task linger in your mind, degrading focus and decreasing performance on the current task — especially harmful for cognitively demanding or creative work.

If we think about hybrid work as an engineered system, context switching is a kind of “friction” — not in code or infrastructure, but in human attention. And like any engineering problem, friction can — and should — be minimized.

Second‑Order Effects: Why Cognitive Fragmentation Matters

Cognitive fragmentation doesn’t just reduce throughput or add stress. Its effects ripple deeper, with impacts on:

Quality of output: When attention is fragmented, even small tasks suffer. Mistakes creep in, thoughtfulness erodes, and deep work becomes rare.
Long-term mental fatigue and burnout: Constant switching wears down cognitive reserves. It’s no longer just “too much work,” but “too many contexts” demanding attention.
Team performance and morale: At the organizational level, teams that minimize context switching report stronger morale, better retention, and fewer “after‑hours” overloads.
Loss of strategic thinking and flow states: When individuals rarely stay in one mental context long enough, opportunities for deep reflection, creative thinking, or coherent planning erode.

In short, hybrid work doesn’t just shift “where” work happens — it fundamentally alters how work happens.

Why Current Solutions Fall Short

There are many popular “help me focus” strategies:

The classic — Pomodoro Technique / “deep work” blocks / browser blockers.
Calendar-based time blocking to carve out uninterrupted hours.
Productivity suites: project/task trackers like Asana, Notion, Linear and other collaboration tools — designed to organize work across contexts.

And yet — these often treat only the symptoms, not the underlying architecture of distraction. What’s missing is a system‑level guidance on:

Mapping cognitive load across workflow architecture (not just “my calendar,” but “how many systems/platforms/contexts am I juggling?”).
Designing environments (digital and physical) that reduce cross‑system interference instead of piling more tools.
Considering second‑ and third‑order consequences — not just “did I get tasks done?” but “did I preserve attention capacity, quality, and mental energy?”

In other words: we lack a rationalist, engineered approach to hybrid‑work life hacking.

Toward Flow‑Preserving Systems: A Pareto Model of Attention

If we treat attention as a finite resource — and work systems as pipelines — then hybrid work demands more than discipline: it demands architecture. Here’s a framework rooted in the 80/20 (Pareto) principle and “flow‑preserving design.”

1. Identify your “attention vector” — where does your attention go?

List the systems, tools, communication modes, and contexts you interact with daily. How many platforms? How many distinct contexts (e.g., team A chat, team B ticket board, email, docs, meetings)? Rank them by frequency and friction.

2. Cull ruthlessly. Apply the 80/20 test to contexts:

Which 20% of contexts produce 80% of meaningful value? Those deserve high-bandwidth attention and uninterrupted time. Everything else — low‑value, context‑switch‑heavy noise — may be candidates for elimination, batching, or delegation.

3. Build “flow windows,” not just “focus zones.”

Rather than hoping “deep work days” will save you, build structural constraints: e.g., merge related contexts (use fewer overlapping tools), group similar tasks, minimize simultaneous cross-team demands, push meetings into consolidated blocks, silence cross‑context notifications when in flow windows.

4. Design both digital and physical environments for flow.

Digital: reduce number of apps, unify communications, use integrated platforms intelligently.
Physical: fight “always on” posture — treat work zones as environments with their own constraints.

5. Monitor second‑order effects.

Track not just output quantity, but quality, mental fatigue, clarity, creativity, and subjective well‑being. Use “collaboration analytics” if available (e.g., data on meeting load, communication frequency) to understand when fragmentation creeps up.

Conclusion: Hybrid Work Needs More Than Tools — It Needs Architecture

Hybrid work is now the baseline for millions of professionals. But with that shift comes a subtle and pervasive risk: cognitive fragmentation. Like a system under high load without proper caching or resource pooling, our brains start thrashing — switching, reloading, groggy, inefficient.

We can fight that not (only) through willpower, but through design. Treat your mental bandwidth as a resource. Treat hybrid work as an engineered system. Apply Pareto-style pruning. Consolidate contexts. Build flow‑preserving constraints. Track not just tasks — but cognitive load, quality, and fatigue.

If done intentionally, you might discover that hybrid work doesn’t just offer flexibility — it offers the potential for deeper focus, higher quality, and less mental burnout.

References

Great Place to Work, Remote Work Productivity Study: greatplacetowork.com
Stanford University Research on Hybrid Work: news.stanford.edu
Reclaim.ai on Context Switching: reclaim.ai
Conclude.io on Context Switching and Productivity Loss: conclude.io
Software.com DevOps Guide: software.com
BasicOps on Context Switching Impact: basicops.com
RSIS International Study on Collaboration Analytics: rsisinternational.org

Support My Work

If this post resonated with you, and you’d like to support further writing like this — analyses of digital work, cognition, and designing for flow — consider buying me a coffee: Buy Me a Coffee ☕

* AI tools were used as a research assistant for this content, but human moderation and writing are also included. The included images are AI-generated.

When the Machine Does (Too Much of) the Thinking: Preserving Human Judgment and Skill in the Age of AI

December 1, 2025 / lbhuston / Leave a comment

We’re entering an age where artificial intelligence is no longer just another tool — it’s quickly becoming the path of least resistance. AI drafts our messages, summarizes our meetings, writes our reports, refines our images, and even offers us creative ideas before we’ve had a chance to think of any ourselves.

Convenience is powerful. But convenience has a cost.

As we let AI take over more and more of the cognitive load, something subtle but profound is at risk: the slow erosion of our own human skills, craft, judgment, and agency. This article explores that risk — drawing on emerging research — and offers mental models and methodologies for using AI without losing ourselves in the process.

SqueezedByAI3

The Quiet Creep of Cognitive Erosion

Automation and the “Out-of-the-Loop” Problem

History shows us what happens when humans rely too heavily on automation. In aviation and other high-stakes fields, operators who relied on autopilot for long periods became less capable of manual control and situational awareness. This degradation is sometimes called the “out-of-the-loop performance problem.”

AI magnifies this. While traditional automation replaced physical tasks, AI increasingly replaces cognitive ones — reasoning, drafting, synthesizing, deciding.

Cognitive Offloading

Cognitive offloading is when we delegate thinking, remembering, or problem-solving to external systems. Offloading basic memory to calendars or calculators is one thing; offloading judgment, analysis, and creativity to AI is another.

Research shows that when AI assists with writing, analysis, and decision-making, users expend less mental effort. Less effort means fewer opportunities for deep learning, reflection, and mastery. Over time, this creates measurable declines in memory, reasoning, and problem-solving ability.

Automation Bias

There is also the subtle psychological tendency to trust automated outputs even when the automation is wrong — a phenomenon known as automation bias. As AI becomes more fluent, more human-like, and more authoritative, the risk of uncritical acceptance increases. This diminishes skepticism, undermines oversight, and trains us to defer rather than interrogate.

Distributed Cognitive Atrophy

Some researchers propose an even broader idea: distributed cognitive atrophy. As humans rely on AI for more of the “thinking work,” the cognitive load shifts from individuals to systems. The result isn’t just weaker skills — it’s a change in how we think, emphasizing efficiency and speed over depth, nuance, curiosity, or ambiguity tolerance.

Why It Matters

Loss of Craft and Mastery

Skills like writing, design, analysis, and diagnosis come from consistent practice. If AI automates practice, it also automates atrophy. Craftsmanship — the deep, intuitive, embodied knowledge that separates experts from novices — cannot survive on “review mode” alone.

Fragility and Over-Dependence

AI is powerful, but it is not infallible. Systems fail. Context shifts. Edge cases emerge. Regulations change. When that happens, human expertise must be capable — not dormant.

An over-automated society is efficient — but brittle.

Decline of Critical Thinking

When algorithms become our source of answers, humans risk becoming passive consumers rather than active thinkers. Critical thinking, skepticism, and curiosity diminish unless intentionally cultivated.

Society-Scale Consequences

If entire generations grow up doing less cognitive work, relying more on AI for thinking, writing, and deciding, the long-term societal cost may be profound: fewer innovators, weaker democratic deliberation, and an erosion of collective intellectual capital.

Mental Models for AI-Era Thinking

To navigate a world saturated with AI without surrendering autonomy or skill, we need deliberate mental frameworks:

1. AI as Co-Pilot, Not Autopilot

AI should support, not replace. Treat outputs as suggestions, not solutions. The human remains responsible for direction, reasoning, and final verification.

2. The Cognitive Gym Model

Just as muscles atrophy without resistance, cognitive abilities decline without challenge. Integrate “manual cognitive workouts” into your routine: writing without AI, solving problems from scratch, synthesizing information yourself.

3. Dual-Track Workflow (“With AI / Without AI”)

Maintain two parallel modes of working: one with AI enabled for efficiency, and another deliberately unplugged to keep craft and judgment sharp.

4. Critical-First Thinking

Assume AI could be wrong. Ask:

What assumptions might this contain?
What’s missing?
What data or reasoning would I need to trust this?
This keeps skepticism alive.

5. Meta-Cognitive Awareness

Ease of output does not equal understanding. Actively track what you actually know versus what the AI merely gives you.

6. Progressive Autonomy

Borrowing from educational scaffolding: use AI to support learning early, but gradually remove dependence as expertise grows.

Practical Methodologies

These practices help preserve human skill while still benefiting from AI:

Personal Practices

Manual Days or Sessions: Dedicate regular time to perform tasks without AI.
Delayed AI Use: Attempt the task first, then use AI to refine or compare.
AI-Pull, Not AI-Push: Use AI only when you intentionally decide it is needed.

Team or Organizational Practices

Explain-Your-Reasoning Requirements: Even if AI assists, humans must articulate the rationale behind decisions.
Challenge-and-Verify Pass: Explicitly review AI outputs for flaws or blind spots.
Assign Human-Only Tasks: Preserve areas where human judgment, ethics, risk assessment, or creativity are indispensable.

Educational or Skill-Building Practices

Scaffold AI Use: Early support, later independence.
Complex, Ambiguous Problem Sets: Encourage tasks that require nuance and cannot be easily automated.

Design & Cultural Practices

Build AI as Mentor or Thought Partner: Tools should encourage reflection, not replacement.
Value Human Expertise: Track and reward critical thinking, creativity, and manual competence — not just AI-accelerated throughput.

Why This Moment Matters

AI is becoming ubiquitous faster than any cognitive technology in human history. Without intentional safeguards, the path of least resistance becomes the path of most cognitive loss. The more powerful AI becomes, the more conscious we must be in preserving the very skills that make us adaptable, creative, and resilient.

A Personal Commitment

Before reaching for AI, pause and ask:

“Is this something I want the machine to do — or something I still need to practice myself?”

If it’s the latter, do it yourself.
If it’s the former, use the AI — but verify the output, reflect on it, and understand it fully.

Convenience should not come at the cost of capability.

Support My Work

References

Macnamara, B. N. (2024). Research on automation-related skill decay and AI-assisted performance.
Gerlich, M. (2025). Studies on cognitive offloading and the effects of AI on memory and critical thinking.
Jadhav, A. (2025). Work on distributed cognitive atrophy and how AI reshapes thought.
Chirayath, G. (2025). Analysis of cognitive trade-offs in AI-assisted work.
Chen, Y., et al. (2025). Experimental results on the reduction of cognitive effort when using AI tools.
Jose, B., et al. (2025). Cognitive paradoxes in human-AI interaction and reduced higher-order thinking.
Kumar, M., et al. (2025). Evidence of cognitive consequences and skill degradation linked to AI use.
Riley, C., et al. (2025). Survey of cognitive, behavioral, and emotional impacts of AI interactions.
Endsley, M. R., Kiris, E. O. (1995). Foundational work on the out-of-the-loop performance problem.
Research on automation bias and its effects on human decision-making.
Discussions on the Turing Trap and the risks of designing AI primarily for human replacement.
Natali, C., et al. (2025). AI-induced deskilling in medical diagnostics.
Commentary on societal-scale cognitive decline associated with AI use.

* AI tools were used as a research assistant for this content, but human moderation and writing are also included. The included images are AI-generated.

System Hacking Your Tech Career: From Surviving to Thriving Amid Automation

November 17, 2025November 14, 2025 / lbhuston / Leave a comment

There I was, halfway through a Monday that felt like déjà-vu: a calendar packed with back-to-back video calls, an inbox expanding in real-time, a new AI-tool pilot landing without warning, and a growing sense that the workflows I’d honed over years were quietly becoming obsolete. As a tech advisor accustomed to making rational, evidence-based decisions, it hit me that the same forces transforming my clients’ operations—AI, hybrid work, and automation—were rapidly reshaping my own career architecture.

WorkingWithRobot1

The shift is no longer theoretical. Hybrid work is now a structural expectation across the tech industry. AI tools have moved from “experimental curiosity” to “baseline requirement.” Client expectations are accelerating, not stabilising. For rational professionals who have always relied on clarity, systems, and repeatable processes, this era can feel like a constant game of catch-up.

But the problem isn’t the pace of change. It’s the lack of a system for navigating it.
That’s where life-hacking your tech career becomes essential: clear thinking, deliberate tooling, and habits that generate leverage instead of exhaustion.

Problem Statement

The Changing Landscape: Hybrid Work, AI, and the Referral Economy

Hybrid work is now the dominant operating model for many organisations, and the debate has shifted from “whether it works” to “how to optimise it.” Tech advisors, consultants, and rational professionals must now operate across asynchronous channels, distributed teams, and multiple modes of presence.

Meanwhile, AI tools are no longer optional. They’ve become embedded in daily workflows—from research and summarisation to code support, writing, data analysis, and client-facing preparation. They reduce friction and remove repetitive tasks, but only if used strategically rather than reactively.

The referral economy completes the shift. Reputation, responsiveness, and adaptability now outweigh tenure and static portfolios. The professionals who win are those who can evolve quickly and apply insight where others rely on old playbooks.

Key Threats

Skills Obsolescence: Technical and advisory skills age faster than ever. The shelf life of “expertise” is shrinking.
Distraction & Overload: Hybrid environments introduce more communication channels, more noise, and more context-switching.
Burnout Risk: Without boundaries, remote and hybrid work can quietly become “always-on.”
Misalignment: Many professionals drift into reactive cycles—meetings, inboxes, escalations—rather than strategic, high-impact advisory work.

Gaps in Existing Advice

Most productivity guidance is generic: “time-block better,” “take breaks,” “use tools.”
Very little addresses the specific operating environment of high-impact tech advisors:

complex client ecosystems
constant learning demands
hybrid workflows
and the increasing presence of AI as a collaborator

Even less addresses how to build a future-resilient career using rational decision-making and system-thinking.

Life-Hack Framework: The Three Pillars

To build a durable, adaptive, and high-leverage tech career, focus on three pillars: Mindset, Tools, and Habits.
These form a simple but powerful “tech advisor life-hack canvas.”

Pillar 1: Mindset

Why It Matters

Tools evolve. Environments shift. But your approach to learning and problem-solving is the invariant that keeps you ahead.

Core Ideas

Adaptability as a professional baseline
First-principles thinking for problem framing and value creation
Continuous learning as an embedded part of your work week

Actions

Weekly Meta-Review: 30 minutes every Friday to reflect on what changed and what needs to change next.
Skills Radar: A running list of emerging tools and skills with one shallow-dive each week.

Pillar 2: Tools

Why It Matters

The right tools amplify your cognition. The wrong ones drown you.

Core Ideas

Use AI as a partner, not a replacement or a distraction.
Invest in remote/hybrid infrastructure that supports clarity and high-signal communication.
Treat knowledge-management as career-management—capture insights, patterns, and client learning.

Actions

Build your Career Tool-Stack (AI assistant, meeting-summary tool, personal wiki, task manager).
Automate at least one repetitive task this month.
Conduct a monthly tool-prune to remove anything that adds friction.

Pillar 3: Habits

Why It Matters

Even the best system collapses without consistent execution. Habits translate potential into results.

Core Ideas

Deep-work time-blocking that protects high-value thinking
Energy management rather than pure time management
Boundary-setting in hybrid/remote environments
Reflection loops that keep the system aligned

Actions

A simple morning ritual: priority review + 5-minute journal.
A daily done list to reinforce progress.
A consistent weekly review to adjust tools, goals, and focus.
A quarterly career sprint: one theme, three skills, one major output.

Implementation: 30-Day Ramp-Up Plan

Week 1

Map a one-year vision of your advisory role.
Pick one AI tool and integrate it into your workflow.
Start the morning ritual and daily “done list.”

Week 2

Build your skills radar in your personal wiki.
Audit your tool-stack; remove at least one distraction.
Protect two deep-work sessions this week.

Week 3

Revisit your vision and refine it.
Automate one repetitive task using an AI-based workflow.
Practice a clear boundary for end-of-day shutdown.

Week 4

Reflect on gains and friction.
Establish your knowledge-management schema.
Identify your first 90-day career sprint.

Example Profiles

Advisor A – The Adaptive Professional

An advisor who aggressively integrated AI tools freed multiple hours weekly by automating summaries, research, and documentation. That reclaimed time became strategic insight time. Within six months, they delivered more impactful client work and increased referrals.

Advisor B – The Old-Model Technician

An advisor who relied solely on traditional methods stayed reactive, fatigued, and mismatched to client expectations. While capable, they couldn’t scale insight or respond to emerging needs. The gap widened month after month until they were forced into a reactive job search.

Next Steps

Commit to one meaningful habit from the pillars above.
Use the 30-day plan to stabilise your system.
Download and use a life-hack canvas to define your personal Mindset, Tools, and Habits.
Stay alert to new signals—AI-mediated workflows, hybrid advisory models, and emerging skill-stacks are already reshaping the next decade.

Support My Work

If you want to support ongoing writing, research, and experimentation, you can do so here:
https://buymeacoffee.com/lbhuston

References

Tech industry reporting on hybrid-work productivity trends (2025).
Productivity research on context switching, overload, and hybrid-team dysfunction (2025).
AI-tool adoption studies and practitioner guides (2024–2025).
Lifecycle analyses of hybrid software teams and distributed workflows (2023–2025).
Continuous learning and skill-half-life research in technical professions (2024–2025).

* AI tools were used as a research assistant for this content, but human moderation and writing are also included. The included images are AI-generated.

TEEs for Confidential AI Training

November 14, 2025November 14, 2025 / lbhuston / Leave a comment

Training AI models on regulated, sensitive, or proprietary datasets is becoming a high-stakes challenge. Organizations want the benefits of large-scale learning without compromising confidentiality or violating compliance boundaries. Trusted Execution Environments (TEEs) are increasingly promoted as a way to enable confidential AI training, where data stays protected even while in active use. This post examines what TEEs actually deliver, where they struggle, and how realistic confidential training is today.

Nodes

Why Confidential Training Matters

AI training requires large amounts of high-value data. In healthcare, finance, defense, and critical infrastructure, exposing such data — even to internal administrators or cloud operators — is unacceptable. Conventional protections such as encryption at rest or in transit fail to address the core exposure: data must be decrypted while training models.

TEEs attempt to change that by ensuring data remains shielded from infrastructure operators, hypervisors, cloud admins, and co-tenants. This makes them particularly attractive when multiple organizations want to train joint models without sharing raw data. TEEs can, in theory, provide a cryptographic and hardware-backed guarantee that each participant contributes data securely and privately.

What TEEs Bring (and How They Work)

A Trusted Execution Environment is a hardware-isolated enclave within a CPU, GPU, or accelerator. Code and data inside the enclave remain confidential and tamper-resistant even if the surrounding system is compromised.

Key capabilities relevant to AI training:

Isolated execution and encryption-in-use: Data entering the enclave is decrypted only inside the hardware boundary. Training data and model states are protected from the host environment.
Remote attestation: Participants can verify that training code is running inside authentic TEE hardware with a known measurement.
Collaborative learning support: TEEs can be paired with federated learning or multi-party architectures to support joint training without raw data exchange.
Vendor ecosystem support: CPU and GPU vendors are building confidential computing features intended to support model training, providing secure memory, protected execution, and attestation flows.

These features theoretically enable cross-enterprise or outsourced training with strong privacy guarantees.

The Friction: Why Adoption Is Still Limited

While compelling on paper, confidential training at scale remains rare. Several factors contribute:

Performance and Scalability

Training large models is compute-heavy and bandwidth-intensive. TEEs introduce overhead from encryption, isolation, and secure communication. Independent studies report 8× to 41× slowdowns in some GPU-TEE training scenarios. Even optimistic vendor claims place overhead in the 5–15% range, but results vary substantially.

My earlier estimate of 10–35% overhead carries ~40% uncertainty due to model size, distributed workload characteristics, framework maturity, and hardware design. In practice, real workloads often exceed these estimates.

Hardware and Ecosystem Maturity

TEE support historically focused on CPUs. Extending TEEs to GPUs and AI accelerators is still in early stages. GPU TEEs currently face challenges such as:

Limited secure memory availability
Restricted instruction support
Weak integration with distributed training frameworks
Immature cross-node attestation and secure collective communication

Debugging, tooling, and developer familiarity also lag behind mainstream AI training stacks.

Practical Deployment and Governance

Organizations evaluating TEE-based training must still trust:

Hardware vendors
Attestation infrastructure
Enclave code supply chains
Side-channel mitigations

TEEs reduce attack surface but do not eliminate trust dependencies. In many cases, alternative approaches — differential privacy, federated learning without TEEs, multiparty computation, or strictly controlled on-prem environments — are operationally simpler.

Legal, governance, and incentive alignment across organizations further complicate multi-party training scenarios.

Implications and the Path Forward

Technically feasible but not widespread: Confidential training works in pilot environments, but large-scale enterprise adoption is limited today. Confidence ≈ 70%.
Native accelerator support is pivotal: Once GPUs and AI accelerators include built-in secure enclaves with minimal overhead, adoption will accelerate.
Collaborative use-cases drive value: TEEs shine when multiple organizations want to train shared models without disclosing raw data.
Hybrid approaches dominate: Organizations will likely use TEEs selectively, combining them with differential privacy or secure multiparty computation for balanced protection.
Trust and governance remain central: Hardware trust, supply-chain integrity, and side-channel resilience cannot be ignored.
Vendors are investing heavily: Cloud providers and chip manufacturers clearly view confidential computing as a future baseline for regulated AI workloads.

In short: the technology is real and improving, but the operational cost is still high. The industry is moving toward confidential training — just not as fast as the marketing suggests.

More Info and Getting Help

If your organization is evaluating confidential AI training, TEEs, or cross-enterprise data-sharing architectures, I can help you determine what’s practical, what’s hype, and how these technologies fit into your risk and compliance requirements. Typical engagements include:

Assessing whether TEEs meaningfully reduce real-world risk
Evaluating training-pipeline exposure and data-governance gaps
Designing pilot deployments for regulated environments
Developing architectures for secure multi-party model training
Advising leadership on performance, cost, and legal trade-offs

For support or consultation:
Email: bhuston@microsolved.com
Phone: 614-351-1237

References

Google Cloud, “Confidential Computing: Analytics and AI Overview.”
Phala Network, “How NVIDIA Enables Confidential AI.”
Microsoft Azure, “Trusted Execution Environment Overview.”
Intel, “Confidential Computing and AI Whitepaper.”
MDPI, “Federated Learning with Trusted Execution Environments.”
Academic Study, “GPU TEEs for Distributed Data-Parallel Training (2024–2025).”
Duality Technologies, “Confidential Computing and TEEs in 2025.”
Bagel Labs, “With Great Data Comes Great Responsibility.”

* AI tools were used as a research assistant for this content, but human moderation and writing are also included. The included images are AI-generated.

Introducing The Workday Effectiveness Index

November 6, 2025 / lbhuston / Leave a comment

Introduction:

I recently wrote about building systems for your worst days here.

J0309621

That got me thinking that I need a system to measure how my systems and optimizations are performing on my worst (and average days for that matter) days. Thus:

WDEI: Workday Effectiveness Index

What it is:

A quick metric for packed days so you know if your systems are carrying you or if there’s a bottleneck to fix.

Formula:

WDEI = (top‑leverage tasks completed ÷ top‑leverage tasks planned) × (focused minutes ÷ available “maker” minutes)

How to use (2‑minute setup):

Define top‑leverage tasks (3 max for the day).

Estimate maker minutes (non‑meeting, potentially focusable time).

Log focused minutes (actual deep‑work blocks ≥15 min, no context switches).

Compute WDEI at day end.

Interpretation:

≥ 0.60 → Systems working; keep current routines.

0.40–0.59 → Friction; tune meeting hygiene, buffers, or task slicing.

< 0.40 → Bottleneck; fix in the next weekly review (reprioritize, delegate, or automate).

Example (fast math):

Planned top‑leverage tasks: 3; completed: 2 → 2/3 = 0.67

Maker minutes: 90; focused minutes: 55 → 55/90 = 0.61

WDEI = 0.67 × 0.61 = 0.41 → bottleneck detected

Common fixes (pick one):

Reduce same‑day commitment: drop to 1–2 top‑leverage tasks on heavy days.

Pre‑build micro‑blocks: 3×20 min protected focus slots.

Convert meetings → async briefs; bundle decisions.

Pre‑stage work: checklist, files open, first keystroke defined.

Tiny tracker (copy/paste):

Date: __

TL planned: __ | TL done: __ | TL ratio: __

Maker min: __ | Focused min: __ | Focus ratio: __

WDEI = __ × __ = __

One friction to remove tomorrow: __

Support My Work:

* AI tools were used as a research assistant for this content, but human moderation and writing are also included. The included images are AI-generated.

“Project Suncatcher”: Google’s Bold Leap to Space‑Based AI

November 5, 2025 / lbhuston / Leave a comment

Every day, we hear about the massive energy demands of AI models: towering racks of accelerators, huge data‑centres sweltering under cooling systems, and power bills climbing as the compute hunger grows. What if the next frontier for AI infrastructure wasn’t on Earth at all, but in space? That’s the provocative vision behind Project Suncatcher, a new research initiative announced by Google to explore a space‑based, solar‑powered AI infrastructure using satellite constellations.

What is Project Suncatcher?

In a nutshell: Google’s researchers have proposed a system in which instead of sprawling Earth‑based data centres, AI compute is shifted to a network (constellation) of satellites in low Earth orbit (LEO), powered by sunlight, linked via optical (laser) inter‑satellite communications, and designed for the compute‑intensive workloads of modern machine‑learning.

The orbit: A dawn–dusk sun‑synchronous LEO to maintain continuous sunlight exposure.
Solar productivity: Up to 8x more effective than Earth-based panels due to absence of atmosphere and constant sunlight.
Compute units: Specialized hardware like Google’s TPUs, tested for space conditions and radiation.
Inter-satellite links: Optical links at tens of terabits per second, operating over short distances in tight orbital clusters.
Prototyping: First satellite tests planned for 2027 in collaboration with Planet.

Why is Google Doing This?

1. Power & Cooling Bottlenecks

Terrestrial data centres are increasingly constrained by power, cooling, and environmental impact. Space offers an abundant solar supply and reduces many of these bottlenecks.

2. Efficiency Advantage

Solar panels in orbit are drastically more efficient, yielding higher power per square meter than ground systems.

3. Strategic Bet

This is a moonshot—an early move in what could become a key infrastructure play if space-based compute proves viable.

4. Economic Viability

Launch costs dropping to $200/kg to LEO would make orbital AI compute cost-competitive with Earth-based data centres on a power basis.

Major Technical & Operational Challenges

Formation flying & optical links: High-precision orbital positioning and reliable laser communications are technically complex.
Radiation tolerance: Space radiation threatens hardware longevity; early tests show promise but long-term viability is uncertain.
Thermal management: Heat dissipation without convection is a core engineering challenge.
Ground links & latency: High-bandwidth optical Earth links are essential but still developing.
Debris & regulatory risks: Space congestion and environmental impact from satellites remain hot-button issues.
Economic timing: Launch cost reductions are necessary to reach competitive viability.

Implications & Why It Matters

Shifts in compute geography: Expands infrastructure beyond Earth, introducing new attack and failure surfaces.
Cybersecurity challenges: Optical link interception, satellite jamming, and AI misuse must be considered.
Environmental tradeoffs: Reduces land and power use on Earth but may increase orbital debris and launch emissions.
Access disparity: Could create gaps between those who control orbital compute and those who don’t.
AI model architecture: Suggests future models may rely on hybrid Earth-space compute paradigms.

My Reflections

I’ve followed large-scale compute for years, and the idea of AI infrastructure in orbit feels like sci-fi—but is inching toward reality. Google’s candid technical paper acknowledges hurdles, but finds no physics-based showstoppers. Key takeaway? As AI pushes physical boundaries, security and architecture need to scale beyond the stratosphere.

Conclusion

Project Suncatcher hints at a future where data centres orbit Earth, soaking up sunlight, and coordinating massive ML workloads across space. The prototype is still years off, but the signal is clear: the age of terrestrial-only infrastructure is ending. We must begin securing and architecting for a space-based AI future now—before the satellites go live.

What to Watch

Google’s 2027 prototype satellite launch
Performance of space-grade optical interconnects
Launch cost trends (< $200/kg)
Regulatory and environmental responses
Moves by competitors like SpaceX, NVIDIA, or governments