Oak Flats UI Primitives Catalogue

The composed builder — every primitive in the family wired together for the Quote estimator workflow. Trigger fires, RAG grounds, prompt assembles, the model drafts, the gate validates and the tool persists. Eval, cost and version history sit alongside so the team can iterate without leaving the canvas.

Production answer

Full AI workflow studio is a reusable Oak Flats Muffler Men UI primitive with documented states, accessibility expectations, theme behavior, and implementation evidence.

Primary CTAReview Full AI workflow studio states

Generative search brief

Full AI workflow studio: The composed builder — every primitive in the family wired together for the Quote estimator workflow. Trigger fires, RAG grounds, prompt assembles, the model drafts, the gate validates and the tool persists. Eval, cost and version history sit alongside so the team can iterate without leaving the canvas.

Visual reference only — no real models invoked

Trigger · armed · live

Trigger

Shopify quote-request

Webhook

ConfigPOST /hooks/shopify/quote

Last fired · 09:14 AEST1,246 fires / 7dArmed

Sample payload

JSON

{  "type": "shopify.order.requested",  "customer": {    "id": "cust_mick_davis_8821",    "firstName": "Mick",    "phone": "+61411xxx482"  },  "vehicle": {    "year": 2018,    "make": "Toyota",    "model": "Hilux N80",    "variantHints": ["long-range", "GD-6"]  },  "skuRequest": ["manta-cat-back-hc"],  "channel": "web-form",  "receivedAt": "2026-05-29T09:14:11+10:00"}

Cost projection

Quote estimator

Claude Opus 4.7

Per runA$0.0752,440 tok

Per dayA$10.71142 runs

Per monthA$32130-day projection

Daily spend · last 7 days (AUD)

Active node model

Claude Opus 4.7FrontierAnthropic$15.00 / 1M2400msConsidered · 3–8sGPT-4oBalancedOpenAI$5.00 / 1M1100msMedium · 1–3sGemini 2.5 FlashFastGoogle DeepMind$0.300 / 1M380msFast · sub-1sLlama 3.3 · 70BSelf-hostedSelf-hosted · GPU bay$0.020 / 1M640msMedium · 1–3s

Canvas · quote estimator · model node mid-stream

100%

Inputs · prompt + RAG · selected node

v3.2 · live

Quote estimator prompt

119/8.0k

System

User

Variables · 5

Mick2018Hilux N80[3 items]2.5

Vector search · RAG

Parts catalogue · fitment context

top 4

Embeddingtext-embedding-3-large

Indexparts-catalogue-2026

RerankOn · cohere-rerank-3

01
Manta high-clearance cat-back · N80 fitment guide
fitment/hilux-n80-manta.md
12mm clearance offset required for N80 trucks running the 130L long-range tank. Bay 2 hoist setup notes inline.
sim92rerank95
02
Hilux N80 2.8 GD-6 — supplier price brackets
supplier-feeds/manta-2026-04.csv
Manta 3in mid-pipe $440 + cat-back $1,055 + dump pipe optional $620. Lead time 3 business days.
sim88rerank93
03
Bay labour matrix — exhaust fitment
labour/exhaust-matrix.md
Cat-back swap N80: 2.5h flat rate. Add 0.4h for long-range tank shield. Bay 2 tooling required.
sim84rerank88
04
Cat-back warranty terms
policies/warranty.md
Manta cat-back parts 24-month warranty. Fitting warranty 12 months from invoice date.
sim71rerank62

Chain steps · live trace · running model node

Safety

Strip PII

passed

Input

Mick Davis · +61 411 xxx 482 asks about an N80 cat-back for his 2018 Hilux.

Output

[customer] asks about an N80 cat-back for his 2018 Hilux. {phone:[redacted]}

Vector

Pull fitment + supplier price

passed

Input

query = 'Hilux N80 long-range cat-back Manta clearance' · top 6

Output

6 hits · top similarity 0.92 (fitment/hilux-n80-manta.md) · rerank 0.95

Prompt

Compose estimator prompt

passed

Input

system = quote-estimator v3.2 · user injects {{vehicle}}, {{parts.skuList}}, {{labour.hours}}

Output

1,820 tokens of context handed to the model. Output gate awaits JSON.

Model · Claude Opus 4.7

Draft the quote

running

Input

Generate JSON quote — parts inc GST, labour 2.5h × $80, total rounded.

Output

{ "totalIncGstAud": 1695, "lineItems": [ … 3 items ] } — streaming

Output gate

Validate JSON schema

idle

Input

Awaiting JSON output from upstream model.

Output

—

Gate + tool + safety · post-model surface

Output gate

Quote shape gate

JSON schema

Pass rate92.0%

Evaluated1,846

Pass · warn · fail2·1·0

Rule preview

JSON

{  "$schema": "http://json-schema.org/draft-07/schema#",  "type": "object",  "required": ["totalIncGstAud", "lineItems"],  "properties": {    "totalIncGstAud": { "type": "integer", "minimum": 1, "maximum": 25000 },    "lineItems":      { "type": "array", "minItems": 1 }  }}

Recent decisions

10:14:03Mick N80 cat-back · $1,695Pass
10:09:48Leah Commodore SS · $2,180Pass
10:02:18Manta DPF · $480Warn · totalIncGstAud rounded · soft accept

Tool call

Quote · create

quote.create

Author a quote ID with line items and persist to the CRM.

JSON schema · input

JSON

{  "name": "quote.create",  "description": "Persist a quote with line items + totals.",  "input_schema": {    "type": "object",    "properties": {      "customerId":     { "type": "string", "pattern": "^cust_[a-z0-9_]+$" },      "vehicleSummary": { "type": "string", "maxLength": 80 },      "lineItems": {        "type": "array",        "items": {          "type": "object",          "properties": {            "sku":       { "type": "string" },            "qty":       { "type": "integer", "minimum": 1 },            "retailAud": { "type": "number", "minimum": 0 }          },          "required": ["sku", "qty", "retailAud"]        }      },      "labourHours":     { "type": "number" },      "totalIncGstAud":  { "type": "number" }    },    "required": ["customerId", "lineItems", "totalIncGstAud"]  }}

Result mapping

result.quoteId
result.totalIncGstAud
result.stripePaymentLink

Retry · exponential · max 3Timeout · 4000ms

Safety checks

Safety guardrails

44 hits / 5,434 checked

PII redact
Strip phone, email and ABN before logging the payload.
Inspected1,846Hits32Hit rate1.7%
On hitRedactEnabled
Jailbreak
Block prompt-injection attempts disguised as customer messages.
Inspected1,846Hits4Hit rate0.2%
On hitBlock · haltEnabled
Moderation
OpenAI moderation pass on outbound copy.
Inspected1,742Hits8Hit rate0.5%
On hitFlag · continueEnabled

Parallel + loop · ensemble + agent

Parallel branch

Blog draft · 4-model race

Join · Race · best score

GPT-4o draftConcise · keyword density target 1.2%
Passed1240ms82/100
Claude Opus 4.7 draftEditorial · workshop voice
Passed2380ms91/100
Gemini Flash draftSpeed pass · short variant
Running380ms
Llama 3.3 70B draftOn-prem · local fallback
Failed640ms

Agent loop

SMS triage · Mick N80

4/8

50%

Goal

Identify intent, capture vehicle, run quote estimator, reply in one SMS.

Goal reachedMax iterationsBudget cap

Thought
Goal reached. Send SMS reply with quote refresher and Bay 2 slot offer.
Action
hermes.respond({ template: 'quote.ack', vars: { quoteId: 'q_44521' } })
Observation
SMS sent · message-id sm_88421 · delivered receipt pending

Eval + history · iteration loop

Eval runner

Quote estimator · v3.2

Weighted overall87

Weighted overall score 87/100

Pass threshold · 80/1004/6samples cleared the bar · last run Today 09:14

Rubric axes

Accuracy× 0.4081
Tone× 0.2085
Safety× 0.2599
Cost× 0.1587

ScoreboardSample inputs scored across rubric axes.

Sample inputs scored across rubric axes.
Sample	Accuracy	Tone	Safety	Cost	Overall
Hilux N80 cat-back · long-range	96	92	100	78	94
Commodore SS quote follow-up	82	88	100	88	89
Manta DPF warranty rattle	64	72	92	84	76
Falcon BA mid-pipe stock	90	84	100	92	92
Saturday booking confirmation	98	96	100	96	98
Engineered exhaust ADR cert	58	76	100	82	76

Prompt history

Quote estimator

4 versions

v3.2LiveToday · 09:14 AEST· bec.s91.0%
Tightened JSON shape — totalIncGstAud now required integer. Added long-range tank clarifying ask.
142 runs1,820 tok/run
v3.1Yesterday · 16:48· sam.w86.0%
Restored Aussie tradie register · trimmed the 'Voice:' block · cut 64 system tokens.
318 runs1,748 tok/run
v3.0May 14 · 11:02· jordan.r78.0%
Migrated to JSON-only output. Removed prose preamble. Output gate enforces totalIncGstAud.
412 runs1,612 tok/run
v2.4May 02 · 14:28· bec.s68.0%
Added parts.search guardrail · refused to quote non-stock SKUs without lookup.
588 runs1,948 tok/run