What is the Manufacturing Intelligence Framework?

A systematic, industry-agnostic methodology for turning a custom manufacturing operation into a self-reinforcing, AI-augmented system. It splits into two interlocking frameworks: the Build (eight phases from raw product data to a working system) and the Run (the operating cycle that keeps it improving).

What are the eight phases?

Product Formulation, Operational Mapping, Relational Database, and Knowledge Base & SOPs form the sequential foundation; Project Intelligence, Configure → Price → Quote, and Sales & Client Management form the feedback triangle; and Content & Marketing closes the loop.

What is the feedback triangle?

The key architectural insight: phases 5–7 (project intelligence, quoting, and sales) are not a linear pipeline but a loop. Production truth, quote accuracy, and sales promises continuously correct each other, which is what turns a set of tools into an intelligence layer.

Why does it say every business sits on a product?

Because once a product is decomposed to its atoms (every material, labor second, and overhead dollar), pricing, quoting, scheduling, sales, and marketing become functions of that product data rather than manual effort.

The methodology

The Manufacturing Intelligence Framework

A systematic methodology for turning a custom manufacturing operation into a self-reinforcing, AI-augmented intelligence platform: two interlocking frameworks, eight phases, and the feedback triangle at their core. Documented as it’s built.

The core thesis

Every manufacturing business sits on a product. Understand that product at its most granular level (every material, every labor second, every overhead dollar) and you can automate everything above it.

Pricing, quoting, scheduling, sales, marketing, reporting: all of it becomes afunction of product data, not manual effort. That’s the foundation everything else builds on. Skip it, and nothing above it works. The framework is industry-agnostic: the case study is a custom cabinet manufacturer, but the architecture applies to any custom manufacturing operation: metalwork, print shops, textiles, electronics, ceramics, and more.

Interactive · the framework, appliedGenerate an operations blueprint for your shopAnswer six questions and get a tailored, foundation-first plan built on the phases below.

Watch · video overviewSmall Factories Don’t Need Enterprise Software AnymoreAn earlier talk walking through these ideas. The concepts still hold; the implementation has moved on a good deal since.

Two interlocking frameworks

The methodology splits into two complementary frameworks rather than one bloated checklist. Demonstrating systematic thinking about both implementation and sustainability is stronger than a single linear list.

Framework 1

The Build

8 phases, zero to system. Gets you the system.

Framework 2

The Run

Continuous refinement with a team. Makes the system smarter.

“The Build gets you the system. The Run makes it smarter.”

Each framework has three layers: the Phases (what to build), theNavigator (how to think before you build each one), and theTheory (the established principles each phase rests on). The thinking layer has its own page: the Navigator’s 40 questions.

The Build: the 8 phases

The framework flows through eight interconnected phases. Each builds on the last; each feeds the next. Skip one, and the floors above it crack.

The Build, end to end: Phases 1–4 are the sequential foundation; 5–7 form a feedback triangle that keeps pricing honest as production and sales move; Phase 8 turns the whole system’s output into content, and loops back to refine Phase 1.

01Built · in use
Product Formulation
“You can’t automate what you haven’t decomposed.”
WhatDeconstruct the product into atomic components (every material, labor step, time-per-operation, and overhead dollar) and define a Bill of Materials for each variation.
WhyThis is the DNA of the whole system. Every downstream function (quoting, scheduling, pricing) depends on this data being accurate. Without it you’re guessing; with it you’re engineering.
TheorySix Sigma DMAIC · Activity-Based Costing · Lean value-stream
In practice~300 catalog items decomposed into full BOMs across 7 workstations. The signature catch: one operation was estimated at 21.5 minutes but measured at 7.5, a 3× error the formulation surfaced.
02Built · in use
Operational Mapping
“Map the machine before you optimize it.”
WhatIdentify every resource needed to deliver (tools, people, skills, facilities, software) and link each one directly to the product formulation.
WhyOperations that aren’t mapped can’t be optimized. You map what actually happens, workarounds and tribal shortcuts included, not what should happen.
TheoryTheory of Constraints · Lean value-stream mapping
In practiceWorkflow mapped across the 7 workstations, each operation linked back to the Phase 1 formulation data, separating steps that need skill from steps that need time.
03Built · in use
Relational Database
“One source of truth.”
WhatBuild the relational core that connects products, variations, materials, labor, projects, and clients in one structure.
WhyA database isn’t a filing cabinet. It’s a set of relationships. Material → product → project → client → quote. Break one connection and you’re relying on memory to bridge the gap. One shared store is also what makes reporting trivial later: no ETL, no warehouse.
TheoryRelational modeling · single-source-of-truth architecture
In practiceA PostgreSQL core shared across quoting, inventory, production, and sales, so every report pulls from the same source as everything else.
04Built · in use
Knowledge Base & SOPs
“Capture the tribal knowledge before it walks out the door.”
WhatCapture explicit knowledge (procedures, specs, price lists) and convert tacit knowledge (the things people “just know”) into documented, retrievable form, with AI semantic retrieval on top.
WhyTraditional knowledge management is a filing cabinet: you have to know what you’re looking for. AI-powered retrieval is an expert colleague. It understands the question. Every “you just have to know that” is an undocumented SOP.
TheoryNonaka & Takeuchi: tacit→explicit conversion (SECI)
In practiceA knowledge portal with a Qdrant vector database for semantic search and an agent-memory layer for context, replacing tools that typically run $2K–$15K/yr.
05Built · in use
Project Intelligence
“Every completed project is a test of your system’s accuracy.”
WhatTrack real production data, cost variance, and outcomes on every project. The gap between estimated and actual is the most valuable data the business produces.
WhyStop treating closeout as the end. Treat it as the start of the next improvement. Every estimate-vs-reality gap is a pricing error, a process inefficiency, or a knowledge gap. Which one is it?
- 5AProduction Intelligence: cost variance, process improvement
- 5BCommercial Intelligence: pipeline analytics, pricing-pattern and change-order analysis, contract-value tracking
TheoryDMAIC Analyze & Improve · statistical process control
In practiceA sales dashboard with YTD stats, monthly trends, and rep performance; 60+ projects and 100+ inventory items tracked. Both tracks feed back into Phase 1 and Phase 6.
06Built · in use
Configure → Price → Quote
“A quote is not a creative act. It’s a calculation.”
WhatAutomated, contextual, real-cost quoting with self-correcting estimates.
WhyIf you can’t reduce quoting to a formula, even a complex one, you don’t yet understand your costs well enough. The quote should be the inevitable mathematical output of specs + labor + materials + margin. No guessing.
- 6ACatalog: dynamic BOM-priced items alongside locked custom items
- 6BQuote builder: multi-room, per-room overrides, quote locking, Excel export
- 6CEstimate container & versioning: AI-generated change summaries between versions
- 6DPricing strategy: one-off vs. bulk, customer tiers, optimization-aware costing
TheoryConfigure-Price-Quote discipline · cost-plus & margin modeling
In practiceA multi-room quote builder with bin-packing cut optimization (MaxRects, 5 strategies) that raised material utilization from ~31% to ~79%, which is what makes a unit meaningfully cheaper in an optimized batch than as a one-off.
Configure → Price → Quote · try it
A quote is a calculation, not a guess
Illustrative example: generic materials and made-up rates, not the company’s real pricing. The mechanism is real; the numbers are not.
Configure
Item
Width: 600 mm
Material
Hardware
Quantity: 1
The quote: every line, sourced
Material: Melamine$155
Labor: CNC · edgebanding · assembly$30
Hardware: None$0
Overhead (15%)$28
Margin (42%)$89
Unit price$302
Total (×1)$302
Material utilization31%
Raise the quantity to watch batch optimization lift utilization and cut the per-unit cost.
Change any input and the quote recomputes from material + labor + hardware + overhead + margin. No guessing. Order more and shared-sheet nesting lifts utilization from ~31% toward ~79%, so the same product is genuinely cheaper per unit in an optimized batch than as a one-off.
07Built · in use
Sales & Client Management
“Sales is not separate from operations.”
WhatA data-backed pipeline, contracts, change orders, and multi-factory project linking, not gut feeling.
WhyEvery quote is a promise about production capacity, material cost, and timeline. If the sales system isn’t connected to operational data, you’re making promises you can’t verify.
- 7AOutreach & lead generation: campaigns, touchpoint logging, follow-ups
- 7BPipeline management: stage tracking, conversion, stuck-lead detection
- 7CContracts: generation and document handling
- 7DCoordination: one sales project linking many factory projects
- 7EProfitability: tracing each project’s true margin
TheoryPipeline / funnel management · CRM discipline
In practiceMulti-factory project linking and a change-order system with auto-generated codes, remake-responsibility tracking, and database triggers that keep totals current.
08Emerging
Content & Marketing
“If Phases 1–7 are done right, marketing is just publishing what your system already knows.”
WhatTurn operations into growth: generate content from real project data, and feed the market’s response back into product and pricing decisions.
WhyMarketing isn’t a separate activity bolted on at the end. The project photos, the specs, the results are already there. You’re assembling, not creating. If you’re not proud enough to show the work publicly, that’s a product problem, not a marketing one.
TheoryContent flywheel · social proof · marketing automation
In practiceThis website is the first piece of Phase 8, published from what the system already knows. Planned next: auto-generated brochures from the BOM database and a portfolio chatbot for client validation.

The key architectural insight

The Feedback Triangle

Phases 5–7 are not a linear pipeline. They’re a feedback triangle with bidirectional data flow. A linear framework is a checklist; a system with feedback loops is anarchitecture. This is the difference between “I built some tools” and “I designed an intelligence layer.”

07 ⇄ 05A change order books a new factory project; its real production costs flow back as profitability.
05 ⇄ 06Actual costs correct the estimates, so the next quote lands closer to reality.
06 ⇄ 07Every quote is a promise sales makes about cost, capacity, and timeline.

The non-obvious part: Phases 5–7 aren’t a pipeline, they’re a loop. Production truth, quote accuracy, and sales promises continuously correct each other, which is what turns a set of tools into an intelligence layer. Phase 8 publishes what the loop learns and feeds the market’s response back to Phase 1.

Handoff integrity: the cross-phase layer

Most real-world failures happen not within a phase but at the seams betweenphases. At each transition, four questions keep the system honest: What data moves to the next phase? How do you verify it arrived correctly? What happens when the upstream data changes after the handoff? And who owns the data at each stage?

The Run: the operating cycle

The Build gets you the system once. The Run is what a team does with it every day afterward: every completed project feeds the feedback triangle, every estimate-vs-reality gap becomes a formulation correction or a new SOP, and every closeout sharpens the next quote. The system doesn’t just hold data. It gets measurably smarter the longer it runs. The Navigator’s readiness signals are how you know each loop is actually closing.

Enterprise equivalents

The framework replaces what large enterprises buy as separate, expensive systems. The terminology is useful for talking to outsiders, but internally it’s one integrated platform with interconnected modules, not separate systems stitched together.

Framework component	Enterprise equivalent	Typical cost
Phase 1: Product Formulation	PLM / PIM	—
Phase 4: Knowledge Layer	KMS (AI-powered)	$2K–$15K/yr
Phase 5: Project Intelligence	MES + BI	BI: $5K–$50K/yr
Phase 6: Configure → Price → Quote	Salesforce CPQ, Apttus	—
Phase 7: Sales	CRM / ERP sales modules	—
Phase 8: Content	DAM	—
The whole stack	“Composable ERP” / Smart Manufacturing	—

One important distinction: this is not a traditional PLM that starts from specs and pushes down to production. It’s a reverse-engineered formulation engine: it captures production data and builds the formulation up from actuals, so pricing is grounded in reality, not estimates.

The end state

When all eight phases are connected, you don’t just have a CRM. You have an autonomous operations system. Product changes cascade instantly through pricing, quotes, and marketing. Every project generates data that refines estimates and builds case studies. Agents query the knowledge base to generate quotes and flag issues. This isn’t software implementation. It’s operational transformation through systematic data architecture.

Where things stand

Phases 1–7 are built and in daily use inside a working operation: the foundation, the knowledge layer, and the full feedback triangle. Phase 8 is the emerging piece, andthis website is its first one: marketing as publishing what the system already knows. I document each phase as it meets reality, not as a plan that hasn’t yet.

Framework FAQ

What is the Manufacturing Intelligence Framework?: A systematic, industry-agnostic methodology for turning a custom manufacturing operation into a self-reinforcing, AI-augmented system. It splits into two interlocking frameworks: the Build (eight phases from raw product data to a working system) and the Run (the operating cycle that keeps it improving).
What are the eight phases?: Product Formulation, Operational Mapping, Relational Database, and Knowledge Base & SOPs form the sequential foundation; Project Intelligence, Configure → Price → Quote, and Sales & Client Management form the feedback triangle; and Content & Marketing closes the loop.
What is the feedback triangle?: The key architectural insight: phases 5–7 (project intelligence, quoting, and sales) are not a linear pipeline but a loop. Production truth, quote accuracy, and sales promises continuously correct each other, which is what turns a set of tools into an intelligence layer.
Why does it say every business sits on a product?: Because once a product is decomposed to its atoms (every material, labor second, and overhead dollar), pricing, quoting, scheduling, sales, and marketing become functions of that product data rather than manual effort.
Is the framework specific to cabinets?: No. It is industry-agnostic. The reference case is a custom cabinet manufacturer, but the architecture applies to any custom manufacturing operation: metalwork, print shops, textiles, electronics, ceramics, and more.