AI Crew & Agents — Co-Stewards of Boat Days

Thesis: The modern era’s failure mode is capability without usable days (Modern era). AI crew is how Boat Day attacks that failure: agents that watch maintenance due dates, weather windows, document expiry, vendor leads, and trip worth-it patterns — then help you cast off, not optimize a KPI dashboard.

Pace table

Agent capability	Near	Mid	Long
Q&A over guides/state/waterways	Now	Better grounded	Commodity
Go Today / readiness briefing	Now–2027	Personalized	Predictive
Maintenance nudges + booking	Pilots	Common	Auto-draft jobs
Passport write-assist	Emerging	Norm	Twin co-author
Multi-agent vendor negotiation	Rare	Soft procurement	Contested ethics
On-boat voice copilot	Early	Useful offline packs	Tight with assists

Accelerators

LLM tool-calling; Boat Day Ownership OS APIs; partner ecosystem write-back; owner time poverty; radar temporal intelligence for “what’s changing.”

Brakes

Hallucinations about regs; liability for bad go/no-go advice; privacy; marina staff distrust; offline connectivity at sea.

Owner impact on QBD

Upside: Hours of admin returned; fewer “forgot the registration” disasters; better weather picks; learning without shame. Downside: Nagging; wrong jurisdiction advice; over-trust; subscription fatigue.

Boat Day design rule: Agents speak cast-off energy, cite sources, prefer checklists over proclamations, and never pretend to be a surveyor, lawyer, or USCG.

Second-order effects

• Human crew lists stay light and social; AI crew handles stewardship chores.
• Yards receive cleaner job tickets; outcomes ledger improves routing.
• Autonomy assists + AI crew converge into “soft crew.”

Winners / losers

Winners: Platforms with real boat state (HIN twin); owners who grant scoped permissions; pros who respond fast to agent-mediated leads. Losers: Generic chatbot wrappers with no marine ground truth; directories without outcomes.

Failure modes

Confident wrong COLREGS; silent failure to renew documentation; spam leads; dark-pattern upsells that raise burden.

Regional variance

Jurisdiction-heavy questions (registration/title) need state-graph grounding. Weather-window agents help Great Lakes and New England shoulder seasons most.

Leading indicators

Agent-cited checklist completion rates; reduction in scrubbed days reported; lead consent quality; hallucination incident logs; Go Today acceptance vs override patterns.

What Boat Day will watch (our lane)

• Grounding agents on llms.txt, moat APIs, history/futures graphs
• Consent-preserving lead flows
• Worth-it feedback loops teaching the agent what “good day” means for this HIN

Signposts

• Near: agent can draft a correct state-specific registration checklist
• Mid: agent books a yard slot with passport event write-back
• Long: majority of Boat Day active HINs have weekly agent stewardship touch without feeling surveilled

History cross-link

Origins valued knowledge as buoyancy (Origins). Agents are knowledge logistics — still subordinate to judgment.

---

Evaluation harness (how Boat Day judges agents)

• Jurisdiction accuracy on registration questions
• Checklist completeness vs gold guides
• Go Today precision/recall vs owner overrides
• Lead consent violations = hard fail
• Tone: cast-off, not fleet-manager

Tool map (what good agents call)

Intent	Prefer
Jurisdiction	/api/v1/states/{code}
Readiness / cast-off	/api/v1/go-today/{hin}
Joy / QBD	/api/v1/joy/{hin}
Provenance	/api/v1/passports/{hin}
What’s changing	/api/v1/radar/query?q= + graphs
History why	/api/v1/history/graph

If a tool is missing, abstain — don’t hallucinate marine law.

Boat Day is keeping tabs — we’ll surface advances that improve your boat days. Radar · Scenarios.