Machine Design Knowledge Lab
ABOUT MECHALOGICS
MechaLogics is the engineering company developing Design-OS, a traceable engineering workflow for Mechanical AI. We translate and analyze technical documents, existing CAD/PDF drawings, and mechanism videos. More than 30 years of mechanical-design experience is applied to turn submitted material and customer requirements into information that can be reviewed, traced, and used in practical work.
FROM REQUIREMENT TO DESIGN CRITERIA
01 — Interpret the intent
Separate purpose, required functions, operating conditions, constraints, priorities, and acceptance conditions. Make conflicts and missing assumptions visible before a conclusion is formed.
02 — Structure the design data
Organize requirements, functions, structural groups, components, interfaces, parameters, source evidence, revisions, and decision states as connected records.
03 — Establish the criteria
Select relevant criteria across geometry, materials, manufacturing, assembly, cost, schedule, and risk. State trade-offs and validation conditions.
04 — Control the decision
Keep candidate interpretations open until evidence is sufficient. Accepted, rejected, and HOLD items retain rationale, source, checking result, and unresolved conditions.
05 — Validate the generated output
Check generated translations, analyses, relationship maps, tables, and reports against source material, requirements, units and values, terminology, component and interface relationships, and applicable engineering criteria. Record source location, applied criterion, result, status, and open issue. Classify each conclusion as CONFIRMED, INFERRED, or HOLD.
Requirement
Function
Structural Group
Component / Geometry
→ Validate output: CONFIRMED · INFERRED · HOLD
OUR WORKING PRINCIPLES
Context before wording
Terminology, dimensions, components, interfaces, and operating sequences are interpreted together. No sentence, symbol, or value is treated in isolation.
Evidence before certainty
Confirmed facts, engineering inferences, and unresolved items remain separate. Where the material cannot support a conclusion, the result stays open for confirmation.
Relationships before fragments
Geometry, materials, manufacturing, assembly, cost, and risk influence one another. Reviews follow these relationships rather than disconnected observations.
Reviewable output
Findings include source locations, applied references, checking results, decision status, and open issues so the customer can review how each conclusion was reached.
DESIGN-OS R&D DIRECTION
CONNECTED DESIGN STATE
Customer intent is interpreted as requirements and functions, then linked to structural groups, interfaces, parameters, evidence, risks, decisions, and revision history. Geometry, drawings, and reports are outputs of that connected state.
AI SUPPORT WITH HUMAN CONTROL
AI assists with source decomposition, information structuring, comparison, contradiction and missing-evidence detection, option management, and next-check proposals. Engineering approval remains controlled by declared criteria, validators, and an experienced designer.
ENGINEERING KNOWLEDGE FOR AI
Standards, manuals, equations, design rules, constraints, and validated cases are divided into source-linked Knowledge Entries with units, applicability, exclusions, evidence grade, uncertainty, revision, and source. Only applicable, verified knowledge is reused.
DEVELOPMENT DIRECTION
Design-OS is being developed to connect customer intent, design state, applicable knowledge, AI-assisted analysis and generation, validation, approval, and change history in one controlled loop. When evidence is insufficient, the process stops for review instead of inventing a result.
Discuss a project