Production
- Dynamic
- Programmable
- Supports variation
- Governed by real operational constraints
- Changes continuously
The Specify thesis
Where commerce is going
Commerce is becoming a conversation again.
Commerce began with a customer explaining what they needed and a merchant determining what could be made.
This is the shift from mass production to mass specification.
- Flexible production
- Structured capability
- Interpretable customer intent
Mass specification
Industrialisation made commerce scalable by standardising the product. Catalogues made those standard products distributable. Ecommerce made the catalogue searchable and instantly purchasable, but it did not fundamentally change the model.
Now production is becoming programmable, and AI can translate informal customer intent into structured requirements.
The next commercial interface will not only show customers what has already been listed. It will help determine what can responsibly be produced, configured, sourced or combined for them.
Historical observation
Commerce began as a conversation
In early markets there was no complete catalogue to choose from. A customer explained what was needed. The merchant described what existed, what could be changed and what could be made.
What emerged was shaped by all of it at once:
- Customer intent
- Merchant knowledge
- Available materials
- Production capability
- Price constraints
- Time constraints
That model was genuinely responsive. It was also slow, expensive and bounded by the knowledge of whoever happened to be standing behind the counter. It could not be scaled, and the history that follows is the history of solving exactly that.
The product was not always selected before the conversation. It often emerged from it.
Historical observation
Scale required commerce to become one-way
Series production came first: reusing components, designs and methods across limited runs. It lowered cost while keeping some variation, and it was the first real compromise between adapting to a customer and producing at volume.
Industrialisation went further. Factories, machinery, standardised components and production lines made goods faster, cheaper and available to far more people than craft production ever reached. Efficiency depended on reducing variation, and it worked.
Commerce moved from
Tell us what you need, and we will determine what we can make.
to
Here is what we produce. Choose the option that fits you best.
The system became supply-driven. Businesses forecast demand, produced in advance, and then sold what had already been made. Producers decided what would exist before knowing precisely what any individual customer wanted. That was not a mistake. It was the price of making things affordable and available, and it is still why most of what we own exists at all.
Historical observation
The catalogue became the interface to production
A catalogue contains only what has already been defined, priced, photographed, described, approved and published. Everything in it is a decision that was made before the customer arrived.
Deliverable capability
- Materials
- Sizes
- Finishes
- Components
- Production methods
- Supplier options
Published catalogue
- Listed products
- Published options
- Photographed variants
Communication moved in one direction. The business said what it sold; the customer could select from that, but had limited means to say what they actually needed. Where a producer could support additional sizes, materials, finishes or configurations, those possibilities usually stayed invisible, because publishing them was work nobody had done.
The catalogue's limitations appear not only in what customers can select, but in what merchants, search systems and AI assistants can make visible.
A catalogue is not a business's capability. It is a published projection of that capability.
Historical observation
The internet digitised the catalogue
The internet made products globally searchable and instantly purchasable. What it largely did not do was change the shape of the thing being searched.
| Printed catalogue | Product page |
|---|---|
| Index | Navigation and search |
| Order form | Checkout |
The structure shows most clearly at the point where it runs out:
- No results
- This combination is unavailable
- Out of stock
- Contact us
Each of those is a storefront reaching the end of what was published and reporting it as an answer. In many cases the business could have helped. The interface made a commercial decision the business itself never made.
Ecommerce transformed distribution and convenience more than it transformed the relationship between customer demand and productive capability.
Current technological development
Production is becoming programmable
Computer-controlled machines, industrial robots, automated cutting, digital printing, modular assembly and additive manufacturing can change instructions between production runs, and sometimes between individual units.
The same infrastructure can increasingly absorb differences in:
- Dimensions
- Materials
- Colours
- Finishes
- Components
- Designs
- Feature combinations
As machinery becomes more programmable, connected and modular, the cost and operational difficulty of variation can fall. Production moves from pure economies of scale toward scale and scope together: volume and range through shared infrastructure.
None of that makes variation free, unlimited or universally practical. Machines have limits, materials have limits, and capacity has limits. Constraints do not disappear in this model. They become the thing that has to be represented properly.
Specify's thesis
Commerce has not caught up
Many businesses can already produce more variation than their website exposes. The production system supports different dimensions, materials and configurations; the storefront presents a small fixed catalogue. The two have drifted apart.
Commercial interface
- Fixed
- Pre-published
- Exposes listed products
- Governed by page availability
- Updated manually
A business may be capable of thousands of valid variations and still be unable to photograph, describe, price and publish each one as a separate product. At that point the catalogue is no longer a description of the business. It is the bottleneck.
The factory became programmable. The storefront remained a catalogue.
One request, five systems
The same requirement, put to commerce in five different centuries.
I need a table for twelve people, made for this room, delivered before September.
Craft commerce
Discussed directly
The maker asks about the room, proposes a timber, says what is possible before September and what is not. The answer is good. It does not scale past the people this one workshop can talk to.
Mass production
Nearest standard product
Three table sizes exist. One seats ten, one seats fourteen. The customer takes the fourteen and loses the room, or leaves.
Catalogue commerce
Browse predefined dimensions
A printed page lists the available lengths and finishes. Anything not printed is unknowable without a phone call, and the phone call depends on who answers it.
Ecommerce
Search and filter
The same options, now instantly. Filters narrow to zero results, which is presented as an answer rather than as the end of what was published.
Mass specification
Describe it and evaluate it
The requirement is stated once, structured, and compared against what the business can actually make, under its own constraints, with a person accountable for the answer. The result may still be no. It will be a real no, with a reason.
The requirement never changed. Only the system's ability to hear it did.
Current technological development
AI closes the interpretive gap, and only that gap
Customers can now describe what they want without first learning a merchant's catalogue structure, terminology and production system. That is the part that was genuinely missing, and it is the part a model is good at.
What interpretation can do
- Understand the request
- Identify requirements
- Ask what is missing
- Recognise alternatives
- Structure trade-offs
rests on
A language model cannot determine any of the second list from plausibility. It does not know what your machines can hold, which combinations your engineers ruled out, or who is allowed to commit to a date. Asked regardless, it will produce something that reads correctly and commits a business to work it may not be able to do.
So interpretation and authority have to be separate layers, and the authoritative one has to be real: structured capability, explicit constraints, approved price logic, actual lead times and accountable human decisions.
AI makes demand understandable. Structured capability makes the answer trustworthy.
Specify's thesis
From mass production to mass specification
Mass production made it possible to produce the same thing for many people. Flexible automation makes it increasingly possible to produce varied outcomes through shared infrastructure. Interpretation makes it possible to work out which valid outcome fits an individual request.
- Mass specification
- The scalable process of determining the right producible outcome for an individual request.
The right outcome is frequently not a custom one. It may be:
- An existing product
- A valid configuration
- A modified product
- A made-to-order specification
- A responsible alternative
- A solution involving several businesses
- Expert review
- A clear rejection
The customer can say
This is what I need.
The business can respond
This is what we can responsibly deliver.
The system can determine
This is the most appropriate way to connect the two.
This is not a return to pre-industrial production, and it is not a rejection of what industrialisation achieved. It is the responsiveness of the original conversation, carried by industrial efficiency, internet reach, programmable production, structured commercial knowledge and the interpretive ability of models.
The shift is not from standard products to custom products. It is from predefined selection to structured determination.
Specify's thesis
What changes, and for whom
For customers
Describe the outcome you need, rather than mastering a catalogue and an industry's vocabulary before you can ask for anything.
For merchants
Expose what can responsibly be delivered, not only what has been published, without having to publish every possible variant as a product.
For producers
Make production capability commercially discoverable and actionable rather than something a salesperson has to remember and relay.
For suppliers
Let real constraints and availability participate in the commercial decision earlier, while the answer can still change.
For AI assistants
Offer an authoritative commercial layer to consult, instead of forcing the assistant to invent a plausible answer on a business's behalf.
Specify's thesis
Principles for the next commercial system
-
Demand before navigation
Begin with what the customer is trying to achieve, not only with a category tree they have to translate their problem into.
-
Capability beyond the catalogue
Represent what can be produced, configured, modified and sourced, rather than only what has already been listed.
-
Constraints are part of the product
A trustworthy system has to know not only what is possible but under which conditions, and say so when the conditions are not met.
-
Authority must remain explicit
A plausible generated answer is not an approved commercial commitment, and no amount of fluency makes it one.
-
Every rejection is information
Demand that could not be served should become commercial memory rather than disappearing with the visitor.
-
Specialists should be connectable
Businesses should not have to become generalists in order to participate in a complete solution.
What Specify is building
Specify is building the infrastructure that makes this operational
- 01 Customer intent
- 02 Structured Opportunity
- 03 Capability and constraint evaluation
- 04 Responsible commercial outcome
- 05 Decision Record
- 06 Improved future journey
Customer intent becomes a Structured Opportunity. Capability and constraints determine what may responsibly be offered. Human authority stays explicit. The decision is recorded, including the demand that could not be served.
Long-term direction
Over time, connected capability may let several specialist businesses answer a request that no single catalogue could fulfil. That network does not exist yet, and this page is not claiming it does. What exists today is the first half: one business, one governed Capability Map, one honest answer.
The catalogue will remain. It will no longer be the boundary.
Standard products will continue to matter, and so will search, navigation and checkout. Most of what people buy should stay exactly as easy to buy as it is now.
What changes is that the published catalogue no longer has to represent the full limit of what commerce can answer. A customer can begin with the outcome they need. A business can respond with what it can genuinely deliver. Capability, constraints and authority can become part of the commercial interface rather than sitting behind it.
Commerce began as a conversation.
The next version may finally allow that conversation to scale.