Day 1: The Landscape

Overview

Day 1 answers the question: where are we? Attendees arrive with varying levels of experience — some have used Copilot for months, others are new to AI coding tools entirely. This session establishes a shared mental model of the agentic coding ecosystem, quantifies the gap between tool capability and practitioner utilization, and frames the purpose of the remaining four days.

The day's ACIs are mostly foundational. This is intentional. Day 1 is orientation, not skill-building. The goal is not to teach techniques but to change how attendees think about what these tools are and what they are leaving on the table. The one harder note is the productivity claim (ACI-025) — an honest look at where the speedups are real and where they are not — which sets up the map of the whole week.

By the end of this day, every attendee should be able to place themselves on the agentic-assistive spectrum and articulate what "using 10% of the capability" means for their own workflow.

Setup

Prerequisites: None beyond basic software development experience. No prior AI coding tool experience required.

Materials: reference/landscape.md (reference handout), reference/evaluation-framework.md (preview copy — scored as Day 4's self-assessment, then extended with a multi-agent dimension on Day 5).

Time allocation: 100 minutes theory, 75 minutes practical.

Session Flow

ACI-015: The Agentic-Assistive Spectrum (20 minutes)

Key point: There is a spectrum from autocomplete (assistive) to multi-step autonomous execution (agentic). Most practitioners are clustered at the assistive end. The spectrum is not a quality judgement — it is a capability map.

Teaching notes: Start with a show of hands: "Who has used Copilot, Cursor, or similar? Who has used it for anything beyond autocomplete?" This establishes the room's baseline. Present the spectrum as a map, not a ladder. Assistive use is appropriate for some tasks. The insight is that most practitioners do not know the agentic end exists.

Transition: "Now that we have a map, let's look at what's missing from it."

ACI-013: The Methodology Desert (15 minutes)

Key point: The ecosystem has tools and documentation. It does not have methodology — the structured practices that turn tool features into consistent outcomes. This course exists because the desert exists.

Teaching notes: Distinguish between tool documentation ("here is how to use feature X") and methodology ("here is how to organise your work so that feature X produces good results over months"). Show examples from the landscape survey: vendor docs explain commands, not workflows. The gap is not in the tools — it is in the practice layer above them.

Transition: "The desert is not just an intellectual gap. It has a measurable cost."

ACI-018: The Ten Percent Problem (20 minutes)

Key point: Most developers use AI coding tools at roughly 10% of their capability. The gap between 1.2x and 5x productivity is not the tool — it is the practice.

Teaching notes: This is the most data-heavy ACI of the day. Present the utilization-by-capability-level table (autocomplete ~80% of users, chat ~15%, multi-file ~4%, autonomous ~1%). Connect to enterprise metrics: McKinsey's 20-45% improvement range masks the same utilization gap. The aggregate number hides the actionable insight.

Transition: "So why does the gap persist? Why don't practitioners move up the spectrum on their own?"

ACI-020: The Tool-Practice Gap (20 minutes)

Key point: Tool documentation is necessary but not sufficient. The gap between "I know the features" and "I get consistent results" is a practice gap, not a knowledge gap.

Teaching notes: Close the loop. The spectrum (015) shows where you could be. The desert (013) shows why there is no map. The ten percent problem (018) quantifies the cost. The tool-practice gap (020) explains the mechanism: practitioners who read the docs and understand the features still cannot close the gap without methodology. This is the practice gap the rest of the course closes.

Transition: "We know the gap is real. The next question is how big the payoff actually is — and where."

ACI-025: The Productivity Claim (15 minutes)

Key point: Agentic coding is a variable multiplier, not a uniform speedup. Mechanical, systematic work (audits, refactoring, boilerplate) accelerates 10-50x; creative and architectural work barely moves; scope-sensitive work can go negative when the agent optimises locally without global awareness. "A different distribution of effort" is more honest than "10x developer."

Teaching notes: This is the honest-data counterweight to the ten percent problem. Present the multiplier-by-task-type picture from the cross-repo dataset (six repositories, one developer, 18 months): 10-50x on systematic audits, 2-3x on standard feature work, ~1x on architecture and scoping, negative on scope-sensitive work where the agent builds the wrong thing faster. The enterprise implication is the headline — a single "productivity" number averages these cases and hides the actionable insight, and commit or PR counts cannot tell extraordinary value from extraordinary waste. The honest ROI story is a distribution, not a point estimate.

Transition: "That is the payoff and its limits. One last piece before we start the methodology — the shape of the whole week."

The Two Tiers: SAAC and MAAC (10 minutes)

Key point: The course has two tiers on two different axes. Days 1-4 teach SAAC — Single-Agent Agentic Coding: one human, one agent, one session, measured on a depth axis (how autonomous is the agent?). Day 5 teaches MAAC — Multi-Agent Agentic Coding: one human coordinating a fleet, measured on a breadth axis (how many agents can one human run?).

Teaching notes: Draw the two axes (see reference/taxonomy.md). Depth is the spectrum from ACI-015 — assistive to agentic, topping out at Delegator/Verifier. Breadth is the new axis: one agent to a coordinated fleet under a human hypervisor, topping out at Coordinator/Arbiter. The load-bearing relationship is that MAAC is built ON SAAC — you earn the fleet by mastering the single agent first, which is why MAAC is the last day, gated, not woven through the early ones. Have attendees place themselves: most are early on the depth axis and have not touched breadth at all. Days 2-4 build the depth; Day 5 is the step up to breadth.

Transition: "Let's start closing the practice gap. Day 2 is the first concrete skill: how to talk to an agent."

Individual pathway notes

Emphasise personal productivity positioning. "Where are you on the spectrum? What are you leaving on the table?" Give ACI-015 and ACI-018 the most airtime. Ask each attendee to self-assess their current utilization level. The question to leave them with: "What would it look like to move one step up the spectrum this week?"

Enterprise pathway notes

Emphasise the organisational gap. "Your team has licenses. How much capability are they actually using?" Give ACI-018 and ACI-020 the most airtime. Tie to ROI: the difference between 1.2x and 5x is not the tool, it is the practice. Frame the rest of the course as the methodology that turns adoption into utilization. Enterprises measure adoption (how many seats?). This course measures utilization (how effectively are those seats used?).

Connections

To Day 2: Day 1 establishes what is possible and what is missing. Day 2 teaches the first concrete skill: how to communicate with an agent. The bridge is ACI-020's conclusion — the gap is not knowledge, it is practice, and practice starts with communication.

To Day 5: The two-tier map planted today is the week's through-line. Day 5 is the MAAC tier it previews — the step up from one agent to a fleet — and attendees earn it by mastering the single-agent depth axis first across Days 1-4. Preview the evaluation framework today too: they score themselves on it as Day 4's self-assessment and extend it with a multi-agent dimension on Day 5.

Deep-Dive reading: ACI-012 (Methodology Transfer Velocity) is today's self-directed reading — how fast methodology spreads across projects, a landscape-level adoption signal that complements the productivity claim.