Part of my Upper Bound 2026 series — write-ups of the talks worth carrying home from Amii's AI conference in Edmonton. This one is shorter than the others; the notes are lighter. But the frameworks are immediately usable, which earns it a spot.
The Talk: "From Chaos to Clarity"
Peter Bishop of ZGM Modern Marketing Partners gave a session on AI adoption frameworks for marketing teams. His angle: most marketing teams haven't stalled on AI because they lack enthusiasm — they've stalled because they lack a decision-making framework. Everything feels urgent and nothing gets prioritized. The result is chaos: a dozen half-baked pilots, no traction, and a growing suspicion that "AI" is just hype after all.
Bishop's answer is two simple matrices that force the right questions before a team touches a tool.
What comes through in the session is that this is a practitioner framework — built from real marketing-team deployments, not a consultant's deck.
Matrix 1: Time Savings vs. Effort/Complexity
The first matrix is your prioritization filter — it answers "should we even attempt to automate this?"
The axes:
- Y-axis: Time Savings (Value). How much time does automation reclaim? High-value tasks are the ones eating hours every week — reporting, first-draft copy, social scheduling, briefing documents.
- X-axis: Effort/Complexity. How hard is it to actually automate? This covers technical complexity, integration headaches, workflow change management, and how much human judgment the task genuinely requires.
quadrantChart
title Matrix 1: Prioritization Filter
x-axis "Low Effort" --> "High Effort"
y-axis "Low Savings" --> "High Savings"
quadrant-1 "Waitlist"
quadrant-2 "Go"
quadrant-3 "Outsource + hold"
quadrant-4 "Not at this time"
That gives you four quadrants:
- High savings, low effort → "Go." These are your immediate wins. Automate them first. Don't overthink it.
- High savings, high effort → "Waitlist." Worth doing eventually, but not before you've banked some wins. The complexity means you need the experience and the credibility before you tackle these.
- Low savings, high effort → "Not at this time." This quadrant is where AI enthusiasm goes to die. The juice isn't worth the squeeze — at least not yet.
- Low savings, low effort → "Outsource + hold." Easy to automate, but low payoff for your team. If a vendor already solved it, buy the tool and move on.
This is a version of the classic effort/impact matrix — a 2x2 that shows up everywhere from product prioritization to the Eisenhower urgent/important framework. What makes Bishop's version useful for marketing specifically is the "time savings" framing on the value axis. Marketing teams feel time pressure acutely, and grounding the payoff in hours-per-week reclaimed makes the conversation concrete in a way that "strategic value" doesn't.
Matrix 2: Risk Level vs. Reversibility
The second matrix is your go/no-go check once you've picked a candidate from Matrix 1. It answers "is it safe to actually run this?"
The axes:
- Y-axis: Risk Level. What's the blast radius if this goes wrong? A rogue social post is lower risk than automated client-facing proposals. Compliance contexts, regulated industries, and brand-sensitive outputs all push the risk level up.
- X-axis: Reversibility. Can you undo it? Can you catch the mistake before it ships? A draft that a human reviews before sending is reversible. An auto-published post or an automated email sequence is not.
quadrantChart
title Matrix 2: Safety Check
x-axis "Irreversible" --> "Reversible"
y-axis "Low Risk" --> "High Risk"
quadrant-1 "Proceed carefully"
quadrant-2 "Stop"
quadrant-3 "Add a checkpoint"
quadrant-4 "Just try it"
The quadrant logic:
- Low risk + reversible → just try it. The cost of being wrong is low and you can fix it. Don't overthink the governance. Ship and learn.
- Low risk + irreversible → add a checkpoint. Low stakes but hard to take back — add one human review step before it goes out.
- High risk + reversible → proceed carefully. The downside is real but you have an escape hatch. Move slowly, build in review loops, limit the blast radius.
- High risk + irreversible → stop. This is not the automation you want to run right now. Either reduce the risk, add reversibility, or don't automate it yet.
The reversibility axis maps directly onto Jeff Bezos's one-way door / two-way door idea from Amazon's shareholder letters — decisions you can walk back deserve fast, low-ceremony calls; decisions you can't reverse demand more care. It's one of the more durable frameworks to come out of that era of thinking, and it translates cleanly here. A reversible automation is a two-way door: try it, see what happens, adjust.
For teams working in riskier contexts — regulated industries, high-stakes client communications, legal or financial copy — this matrix gives the compliance conversation a structure. The grown-up version of "rate the risk before you automate" is the NIST AI Risk Management Framework (AI RMF 1.0, January 2023, nist.gov/itl/ai-risk-management-framework), which is worth the read if your team is building anything with real consequences. Bishop's matrix is the quick cut; NIST is the rigorous version for when the stakes justify it.
What Did We Learn?
Bishop ended with a short retrospective list from teams that had gone through this process. These aren't surprising in hindsight, but they're worth hearing said plainly:
- Begin from the ground up. Don't assume your existing workflows are the right unit of automation. Go back to basics — what are we actually trying to do here?
- Build a plan quickly. Speed matters. A rough plan executed in a week beats a perfect plan started in a month.
- Some things are harder than they look. Tasks that seem mechanical often turn out to contain hidden judgment calls. The matrix helps you find these before you've already committed.
- Some things are incredibly easy. And you won't know which until you actually try. This is the argument for starting with the "Go" quadrant — the wins are real and they happen fast.
- Look for quick wins. Momentum is the resource. Early wins buy you credibility and appetite for the harder stuff.
Why It Stuck With Me
Most AI-adoption advice is either too abstract ("align AI with your strategy!") or too tactical ("here's a ChatGPT prompt for your caption"). Bishop's two matrices sit in the middle, where the actual friction lives: which tasks do we pick, and is it safe to run them? That's the conversation marketing teams are actually having in 2026, and having a shared visual for it changes how the meeting goes.
The two things I'd carry into any team conversation:
- Matrix 1 stops the wrong fights. Half the debate in AI adoption is about the wrong tasks. Running everything through effort/value first clears the table.
- Reversibility is underrated as a risk filter. Risk level is obvious to ask about. Reversibility is the one people skip — and it's the one that bites them.
Both matrices fit on a whiteboard. That's the point.
More in the Upper Bound 2026 series: the memory paradox talk, Bayesian optimization for experiment design, and a few more sessions worth unpacking.
Header photo by NEOM on Unsplash.
Content on this blog was created using human and AI-assisted workflows described here. Original ideas and editorial decisions by Justin Quaintance.