// welcome
Detroit
Developers
monthly meetup 
detroitdevelopers.com QR code
detroitdevelopers.com
// thank you to our sponsor
RIVET Careers QR code
We're hiring →
rivet.work/careers
// your organizers
Phil Borel
Phil Borel
Organizer
Phil Borel QR code
philborel.com
detroitdevelopers.com QR code
detroitdevelopers.com
Louis Gelinas
Louis Gelinas
Organizer
Louis Gelinas QR code
linkedin.com/in/louis-gelinas
// intro
Three Layers of
Agentic Maturity
Prompt · Context · Harness 
Phil Borel · RIVET · Detroit Developers · 2026
Slides QR code
detroitdevelopers.com/slides/agentic-coding-maturity
// the conflation problem
When engineers say "using AI," they're conflating three distinct disciplines
  • Each has its own investment curve
  • Each has its own ceiling
  • Each has its own compounding dynamics
  • The conflation makes it hard to think clearly about where to invest next
// the three layers
AI Engineering Agentic Software Engineering Harness Engineering Context Engineering Prompt Engineering
Each outer layer contains and depends on the inner ones
// layer 1: prompt engineering
Prompt Engineering
  • The craft of telling an agent what to do — precisely enough that it does it
  • CLAUDE.md, rules files, skill definitions
  • In practice: almost entirely markdown optimization
  • Tight feedback loop — edit a file, run a session, see what changed
  • Ceiling: ~+1x velocity
// layer 2: context engineering
Context Engineering
  • Managing what knowledge the agent has access to — and when
  • PILRs — Persistent Indexed Learning Repositories
  • Type 1 (Ephemeral): per-feature planning docs, test plans
  • Type 2 (Evergreen): architecture docs, system design, API contracts
  • Type 3 (Cumulative): solved problems, incident patterns, institutional memory
  • Ceiling: ~+3x velocity
PILRs blog post QR code
detroitdevelopers.com/blog/context-engineering-pilrs
// layer 3: harness engineering
Harness Engineering
  • Programming around the model — systems that run agents for you
  • Multi-step autonomous workflows with human review of outcomes
  • Deterministic steps where reasoning isn't needed
  • Expands what work gets done at all — not just speed
  • Ceiling: ~+10x velocity (S-curve)
// case study: odradek
Odradek — Bug Resolution Agent
  • Built on Claude Code SDK
  • Investigates → Fixes → Verifies → Opens PR
  • One-shot resolution rate: 60%
  • P1s: bought back ~half an engineer per sprint
  • P2s: addressed as they come in (vs. months of backlog)
  • P3s: actually getting fixed now
// odradek: what's next
Odradek Roadmap
  • Event-driven triggers — auto-fire on bug ticket creation
  • Cloud-hosted dashboard — non-engineers see work in progress & resolution status
  • MS Teams integration — ask about bugs & initiate workflows from chat
  • Parallel issue processing — git worktrees for concurrent work
  • Ephemeral test environments — CS/product verify fixes themselves sans eng involvement
  • Model routing — right model tier for each task; balances cost & effectiveness
// the three ceilings
The Three Ceilings 0 +2x +4x +6x +8x +10x Velocity Gain Investment in Tooling Prompt (+1x) Context (+3x) Harness (+10x)
// where RIVET is today
Where RIVET is
on each curve
// RIVET — prompt engineering
Prompt Engineering 0 +0.25x +0.50x +0.75x +1.0x Velocity Gain Investment RIVET
Far along, near the ceiling
  • Deep investment in CLAUDE.md, rules, skills
  • Almost entirely markdown optimization
  • Tight feedback loop — edit, run, observe
  • Most of the returns are already captured
// RIVET — context engineering
Context Engineering 0 +0.75x +1.5x +2.25x +3.0x Velocity Gain Investment RIVET
Early-mid, significant upside
  • PILRs pattern is right; infrastructure isn't finished
  • Type 1 docs solid, Type 2 in progress, Type 3 nascent
  • Starting to invest in shared hosting & data infra
  • Steepest part of the log curve is right where we are
// RIVET — harness engineering
Harness Engineering 0 +2x +4x +6x +8x +10x Velocity Gain Investment RIVET
Very early — asymmetric upside
  • Odradek is our first real harness — 60% one-shot rate
  • S-curve: slow start, steepest returns ahead
  • Each step removes an engineer from routine work
  • This is where we're investing
// where to invest
Where to invest your tokens
  • Early: Start with prompt engineering. Short feedback loop, transferable skills, necessary foundation
  • Mid-stage: Context engineering. Build the knowledge layer. Start with PILRs where agents are most confused
  • Advanced: Harness engineering. Pick a narrow, repetitive workflow. Measure the one-shot rate
  • The balance: increasing velocity - maintaining quality
// the takeaway
Unlock Capacity (Eng)
+
Unlock Capability (Product)
Capacity
Better prompt & context engineering
makes agentic tools more effective
Capability
Designers & PMs prototype,
validate & shape features directly
// read more
Read the Full Post
Blog post QR code
detroitdevelopers.com/blog/agentic-coding-maturity