How the configurator reaches an answer
Stock, Modified or Custom, plus Not in the tool yet. No dead ends. Every path ends with a person to talk to and a record of what the engineer was trying to build. When a request falls outside the series loaded so far, the tool says so and hands it to an engineer rather than guessing.
It educates before it qualifies
The tool teaches an engineer what is possible. Selling is what the follow-up conversation is for.
Nothing disqualifies
Order quantity never gates the experience, and an unsupported combination routes to an engineer rather than returning nothing.
Modified is a first-class answer
Much of the work sits between the catalog and a blank sheet. The middle outcome makes that visible to a prospect who assumed it was catalog only.
The product data stays editable
Series and part numbers are edited in the website, not in code. The few things that change rarely, like a rule, stay in code on purpose.
Pyvot working notes, shared deliberately
What this prototype is running on
Updated after Anderson Brass's notes on the Configurator Decision Workbook, and again after Ron and Mike's feedback on version 3. Where your notes or feedback told us what to change, it is marked as ABCO direction. Where it is still our draft, it says so.
Version 4: Ron and Mike's feedbackNew
One end connections field with a box for different ends, matched against the catalog. Manual handle with panel mount as an operation choice. Flare replaces SAE, and hose barb, hose bib and O.D. tube replace tube stub. Potable water is now Water, with a drinking water follow-up that preselects the NSF listings, including when the customer is not sure. After the October 20 working session, pressure and approval listings are notes for the engineer and no longer change the answer, and connection size adds under 1/8 inch, with a box for the exact size under 1/8 inch and over 2 inches. A connection size above 2 inches, or DN50, is still the only route to Custom.
Four series populated from your own material
The 3211BV, the 3222BV, the 352BV and the 360BV, with their real part lists. The 360BV is new in version 4: it is built FNPT x MNPT with standard panel-mount threads, so you can test different ends and panel mount. The 3211BV and 3222BV sit in the same 3200 family with different ends, so a three-way request with no ends specified fits both and the tool shows both. A request either matches a real part, or the tool hands it to an engineer.
Your workbook notes changed three rules and eight variables
Seat and seal are now separate. Stem packing folded into seal. Port size became connection size, pressure class became pressure rating, actuation became operation type, as you suggested. Listings are model specific, and pressure at temperature is deliberately out of scope.
The October 20 decisions are in the workbookNew
The Working Session Decisions tab records them, and the Open Questions tab lists what is still open. The Decision column is still empty on all 13 variables and on 4 of the 6 rules on the workbook's Rules tab, and that tab predates version 3, so the configuration rules below are the current list. Categories got the most formal answers: 8 approved, 1 disapproved, 7 marked as sub-groups.
How you described the product structure
Eight families, with sub-groups beneath some of them. Selector and fuel sit under both ball and plug valves. In the configurator these are labels on each series, so they appear on the result without changing how the questions work.
Gap register
| Gap | What it blocks | Owner | Priority |
|---|
What an engineer will expect that we cannot answer
Reviewed as a design engineer specifying a component. Some of these are deliberate scope decisions; the tool should say so rather than stay silent.
| Missing | Why an engineer notices |
|---|
Variable library
The variable library after your notes. Where a published spec sheet and the workbook disagree, the spec sheet wins.
| ID | Variable | Status | What changed |
|---|
Configuration rules
How to read the status column, here and in the variable library. V4 marks a row added or changed in version 4. Agreed: decided with Anderson Brass in the October 2026 working sessions. ABCO direction: from your workbook notes, not yet formally decided. Pyvot draft: our proposal, for your decision. Confirmed: settled in the September sessions. Open question: waiting on you.
| ID | Rule | Result path | Status |
|---|
Pyvot working notes, shared deliberately
How this gets built and who maintains it
A working answer to the question the on-site session left open: can ABCO keep this current without calling a developer every time a product changes. Series and part numbers, yes. A new option on a question, such as a seal compound, still needs a developer, and that is rare.
Three layers, two owners
The mistake to avoid is treating this as one system. It is three, and they change at very different rates.
ABCO owns
Series and part numbers
Each series' attributes and its real part list. This is the part that goes stale when products change, so it is the part ABCO edits directly.
Developer owns
Rules and option lists
A short set of fixed routing rules, one to Custom and the rest to Modified, plus the pick lists behind each question. They change rarely, so they live in code rather than in the editor. That keeps the editor small and removes a whole class of ways to break the tool.
Developer owns
The engine and the interface
The code that reads the series and parts, compares each request against them, and renders the result. ABCO should never touch this, and should never need to in order to change a product fact.
Recommendation: Webflow CMS holds the series and parts, baked into the page at publish
Model series and parts as two Webflow collections. On publish, render them into a JSON payload on the configurator page, and let the client side app read that payload directly.
This avoids runtime API calls, so there is no API token to protect, no rate limit to hit, and no loading spinner while the tool fetches its own rules. ABCO edits in the Webflow Editor, which they will already be trained on for the rest of the site, and hits publish. That is the entire update cycle, with no developer in it.
Attributes live on the series
The earlier draft needed a separate applicability collection joining every product to every variable. Storing each attribute directly on the series removes it. Families and sub-groups become labels, so a series under both ball valves and plug valves simply carries both. Part numbers are their own small collection, looked up rather than composed, so the tool can never hand an engineer a number Anderson Brass does not make.
Proposed collections
| Collection | What it holds | Est. items | Refs |
|---|---|---|---|
| Product Series | One row per series. Pattern, body, seat, inlet and outlet ends, panel mount and pressure rating as fields; family as a text label; one switch per approval listing it holds; description, image and spec sheet. | ~100 | 0 |
| Parts | The real part numbers for each series, with connection size. The tool only shows a part number that appears here. | ~420 | 1 |
Two collections and one reference. Rules, option lists and the question flow live in code.
What ABCO can change alone, and what they cannot
| Change | Who |
|---|---|
| Add a new series and its part numbers | ABCO, in the Editor |
| Change an attribute on a series, or record a listing it holds | ABCO, in the Editor |
| Add or retire an option, such as a new O-ring compound | Developer, rarely |
| Change a routing rule | Developer, rarely |
| A new outcome, a change to the question flow, or the result wording | Developer |
Kept deliberately small
An earlier draft put rules and option lists in the editor, with a rule-group field for combined conditions. Once attributes moved onto the series, most of those rules became plain comparisons with the series record, and the few that remain change rarely. Keeping them in code removes four collections and the risk of an editor change breaking the tool.
The alternative worth naming
Airtable or Google Sheets as the source of truth, synced into the site. It models relationships better than Webflow, and it is literally the workbook ABCO is already filling in, so there is no new editing surface to learn. The cost is a second system to pay for and a sync job to maintain, plus product data living outside the website that hosts it. Worth revisiting only if the catalog outgrows what the website editor handles comfortably. With two collections, that is a long way off. Start in Webflow.
Spec sheets already exist, and they change the plan
The current Wix site publishes a downloadable spec sheet for most series, and several series carry a second overview sheet as well. Those PDFs hold the part number tables, end connection combinations, pressure ratings, materials and dimensional drawings that the intake workbook treats as unknown.
Two consequences. The Product Series collection needs a file field for the sheet, and the configurator should link it on every outcome, which is the most useful thing to hand an engineer. And the migration needs a full inventory of these PDFs with a redirect map, because they sit on two different Wix URL patterns and engineers reach them from search results and saved links.
Update cycle
ABCO edits in the Webflow Editor, publishes, and the payload rebuilds with the site. Quarterly review at minimum, plus ad hoc when a series launches or a material option changes, which matches the cadence in the governance tab of the workbook. The thirty to sixty day internal beta agreed on site should gate the external launch, and during that window the rules panel in this prototype is the right review tool.