Published Work · Comparative Analyses
Comparative Analyses
AI narratives change when they meet real-world product, team, and system constraints. These analyses show how.
What these are
Frameworks that examine how AI narratives change when they meet real teams, workflows, and decision systems — one layer deeper than tools and prompts.
What they’re not
Not declarations of winners. These surface tradeoffs, assumptions, and second-order effects that only appear once AI is applied in practice.
Why it matters
Competitive advantage increasingly comes from how collaboration is designed — not just which models are used.
Jump to an analysis
Prompt Engineering vs. Context Engineering vs. Collaboration Engineering
Prompt Engineering focuses on the ask. Context Engineering expands the scope to what the AI knows — system instructions, memory, retrieval, tools. Collaboration Engineering goes further: it’s about how humans and AI actually work together, with intent, iteration, judgment, and role clarity designed into the process.
Context engineering is a meaningful step forward. It’s also the midpoint, not the destination.
Maura’s note
The shift from prompt engineering to context engineering is real — and I’m glad it’s happening. But I keep seeing the same gap: context improves what the AI sees. It doesn’t change what the human does with it. Better infrastructure. Same relationship. Collaboration engineering is about closing that gap — and the results show up in sharper thinking, stronger products, and outcomes that matter.
In Practice
Use this when teams are upgrading their RAG pipelines and context systems but haven’t rethought how humans stay engaged in the work.
In Life
If you’ve ever had a perfectly organized workspace and still couldn’t find the motivation to do the hard thinking — that’s the difference between better context and better collaboration.
Building the Plane vs. Rebuilding the Cockpit at 30,000 Feet
Teams rarely adopt AI from a clean slate. They’re already shipping, already overloaded, and already accountable. This comparison shows why AI augmentation is a collaboration design problem — context systems, trust boundaries, decision ownership, and feedback loops — not just a tooling rollout.
Maura’s note
I’ve used “building the plane while flying it” for years to explain Agile. Working hands-on with AI augmentation, I realized it no longer captured what was actually changing. AI doesn’t just affect speed — it changes how decisions are made, trusted, reviewed, and owned mid-flight.
Human-in-the-Loop vs. The Human–AI Loop
Human-in-the-Loop (HITL) usually describes oversight: a checkpoint added to improve accuracy or reduce risk. The Human–AI Loop is an authored methodology describing continuous collaboration — humans and AI learning, adapting, and improving together through shared feedback loops.
The difference is subtle but important: one treats humans as validators. The other treats humans as active collaborators and decision-makers throughout the lifecycle.
Read the full analysis →Not All AI Should Be Your Teammate
As AI tools proliferate, teams are often encouraged to treat every system as a “copilot” or teammate. In practice, that assumption breaks down. Some AI belongs in the background — executing, optimizing, or summarizing — while other systems require explicit collaboration models. Misapplied collaboration metaphors can increase cognitive load, reduce trust, and slow teams down.
Read the full analysis →Which Model Is My AI Tool Using?
Many teams interact with AI systems without visibility into which underlying models are being used — or how often those models change. That opacity affects reliability, governance, and product decision-making.
Read the full analysis →Prompt Engineering vs. Collaboration Engineering
Prompt Engineering accurately describes assigning AI a task and refining instructions for better outputs. Collaboration Engineering is the term I use for something broader: how humans and AI work together to deliver more creative, grounded, and impactful outcomes — including roles, handoffs, feedback loops, trust boundaries, memory, escalation paths, and decision ownership.
Maura’s note
I use Collaboration Engineering to name the shift from “prompting for outputs” to designing the collaboration system — the repeatable patterns that make humans + AI effective over time.
Prompt quality alone rarely scales. Teams hit limits when AI is treated as a single-use tool rather than a collaborator embedded in a system.
This distinction becomes critical as AI moves from task execution to agent-like behavior.
Read the full analysis →In Practice
Use this framing when teams say “we just need better prompts” — and the real problem is missing roles, handoffs, trust boundaries, and decision ownership.
In Life
If you’re collaborating with AI daily, the “system” matters: how you store context, review decisions, and create repeatable routines — not just what you ask once.
Why this work matters
As AI becomes embedded into products and workflows, competitive advantage increasingly comes from how collaboration is designed — not just what models are used.
The work behind the work
“For 20 years I built platforms that connected people at scale. The question I’m asking now is the same one — just with a new kind of teammate in the room.”
That question has a methodology now. The Human–AI Loop is where I document what I’ve learned, built, and proven about what humans and AI can achieve together.
Methodology
Guides & Playbooks
Literacy & Writing
Tools & Connect