EC/
Resume
← Field Guide

Framework atlas

Models for building intelligence in the physical world.

Four small frameworks make the practice legible: where intelligence sits, how authority is bounded, how an investigation moves, and how a result proves itself.

01Macroarchitecture

AI for Engineering Stack

From the physical world to outcomes, each layer turns engineering reality into something an intelligent system can inspect, govern and ship.

  1. 08OutcomesQuality · productivity · risk · decisions
  2. 07DeploymentUAT · production · adoption · feedback
  3. 06Engineering productDesktop · web · APIs · digital twins
  4. 05Evidence & governanceProvenance · evals · rules · human review
  5. 04IntelligenceAgents · retrieval · models · reasoning
  6. 03FormalizationEntities · graphs · populations · predicates
  7. 02Engineering contextModels · drawings · specs · requirements
  8. 01Physical worldBuildings · systems · assets · space
Physical → Context → Structure → Intelligence → Evidence → Verification → Product → Deploymentv1.0 · 2026
02Authority boundary

AI Authority Ladder

AI can do more than answer questions, but investigation and engineering authority are different things.

Governed zoneVerify · Determine · ApproveFixed rules, review and accountable authority
— authority boundary —
AI-assisted zoneDiscover · Classify · Interpret · Investigate · ProposeRetrieval, traversal, reasoning and candidate findings
Read the boundary in AEC →
03Operating loop

Engineering Intelligence Loop

The system is not a one-way pipeline. What happens in the field returns as feedback to improve the next evaluation.

  1. 01Requirement
  2. 02Interpretation
  3. 03Formalization
  4. 04Population
  5. 05Evidence
  6. 06Investigation
  7. 07Verification
  8. 08Decision → production → feedback
Read the operating model →
04Evidence model

Graph → proof

Engineering knowledge becomes useful when entities, relationships, sources and evidence can be traversed into a formal claim.

GraphEntities + relationships
→
PopulationWho the rule applies to
→
PredicateWhat must hold
→
CounterexampleWhere it fails
→
DeterminationNOT MET · inspectable
Trace the determination →

Use the frameworks

They are working models: useful when they sharpen a decision, revisable when the field teaches something new.

Back to the Field Guide →See them applied →