Personal radar

Staying current is a system — not a Twitter habit

Frontier AI moves weekly — models, regulation, agent frameworks, safety incidents. Personal radar is a sustainable information diet: trusted sources, time-boxed review, structured notes, and periodic synthesis — not infinite scroll. This lesson closes the frontier module with digest template + State of AI brief connecting trends to your capstone.

Weekly digest template

Create frontier/digest/TEMPLATE.md:

# Week of YYYY-MM-DD

## 3 things that mattered
- …

## 1 thing overrated / hype
- …

## Capstone relevance
- …

## Sources read (links)
- …

## Experiments to try (optional, ≤1h)
- …

Time-box 45–60 minutes weekly. Skip weeks intentionally rather than guilt-spiral.

Callout — Sources you will actually read: Five subscriptions beat fifty bookmarks. Quality and consistency beat completeness.

Curated source categories

Balance:

• Primary — labs (OpenAI, Anthropic, Google DeepMind blogs), arXiv summaries you trust
• Engineering — company eng blogs (Netflix, Uber, Shopify patterns you studied)
• Policy — AI Act summaries from reputable analysts, not hot takes
• Tools — release notes for stack you use (LangGraph, vLLM, MCP spec)
• Critique — safety, evals, failure postmortems

List your chosen sources in frontier/sources.md with why each stays — prune quarterly.

Avoid duplicate hype channels saying the same launch.

State of AI brief

frontier/state_of_ai_brief.md (2–4 pages) synthesizes last 4–8 weeks of digests:

• Model capability shifts affecting product (reasoning, multimodal, context length)
• Agent engineering trends (MCP adoption, harness discourse)
• Eval/safety regulatory notes (high level)
• Implications for your capstone — adopt, defer, or ignore
• Mini-demo idea — one afternoon experiment validating a trend

Brief is opinionated engineering memo, not encyclopedia.

Mini-demo

Close milestone with small artifact proving you engaged frontier hands-on:

• Reasoning model benchmark snippet (lesson 14.1)
• Vision agent eval (14.2)
• Local SLM latency chart (14.3)
• Synthetic filter pipeline (14.4)

Pick one; link from milestone-14.md.

Anti-patterns

• Saving PDFs never opened
• Replacing building with reading
• Panic pivot every new model launch — use revisit dates from design review

Signal vs noise heuristics

Prefer primary sources over aggregator threads; read release notes before hot takes; skip "AGI achieved" headlines until reproducible benchmark. Document personal rules in sources.md — evolves with experience.

Sharing digests

Optional: share one digest with study group or mentor — external accountability improves consistency. ShipAI portfolio can link one digest as artifact proving habit.

Connecting radar to capstone pivot guard

State of AI brief ends with explicit non-pivots — trends you will ignore for capstone v1 — prevents distraction from scope freeze.

Putting it together in practice

ShipAI treats this lesson as executable curriculum, not reading alone. Before marking complete, trace one real request through your portfolio stack and label where this lesson's concepts apply — even if the first pass is messy. Document what broke in the module README; that gap list becomes your next sprint.

Compare your implementation against the industry callouts cited earlier without copying their scale. Name one deliberate simplification you kept (mock auth, SQLite not Postgres, single-region deploy) and one simplification you refuse to ship without (no eval gate, no trace on mutating tools, no fail-closed guardrail on exfil cases). That contrast is what interviewers and graders look for.

Callout — Teach back: Explain this lesson's core tradeoff to a peer in five minutes without slides. If you cannot, re-read the failure modes section and add an example from your own run logs.

Common questions and misconceptions

"Is this overkill for a side project?" Side projects can skip pieces; capstones and production cannot skip knowing the pieces exist. You may waive cost accounting in v1 but your architecture diagram should still show where it would attach.

"Should I rewrite from scratch?" Extend what you built in prior modules — graders reward evolution, not parallel unused folders. Link file paths in your checklist.

"Which metric matters most?" The metric tied to user harm or revenue: policy violations, failed refunds, silent wrong answers — not vanity leaderboard scores.

Extension paths after the micro-project

After the micro-project passes smoke check, choose one extension aligned with your capstone pillar: tighten eval coverage, add a chaos or red-team case, or wire observability into SSE streams. Extensions belong in BACKLOG unless scope freeze explicitly includes them — avoids capstone death by optional polish.

Engineering problem (staff framing)

Field moves weekly. Build a personal radar process, not doomscrolling.

Diagram — Radar process

flowchart LR
  Sources --> Triage --> Notes --> Try[Small spike] --> Keep/Drop

Precise definitions & mental model

Signal vs noise, spike budgets, share-outs.

Tradeoffs — when to use what

Depth spikes vs breadth monitoring.

Failure modes (interview + on-call)

Rewriting stack monthly; never writing notes.

Production & OSS practices

Monthly radar note in portfolio.

Micro-project: Weekly digest template

Ship:

1. TEMPLATE.md + sources.md (≥8 sources categorized).
2. One completed example digest for current week.
3. state_of_ai_brief.md with capstone implications.
4. milestone-14.md linking frontier labs + brief + mini-demo.
5. Calendar reminder or documented weekly ritual (honor system OK).

Acceptance: brief stands alone; digest template reusable next week.

Checklist

• Digest template and sources list committed
• One example weekly digest filled
• State of AI brief with capstone section
• Mini-demo linked in milestone-14.md
• milestone-14.md completes frontier module index