Customer quote

From "can I get pricing?" to a signed PDF in fifteen minutes

The customer quote workflow is the most-run flow on the platform. Customers email or call asking for pricing on assemblies, often by their own internal SKU. The AI gathers context (who they are, what they typically buy, current cost basis, AR status), proposes line-by-line pricing, Mike reviews, the system renders an HTML artifact + PDF, and the email goes out.

The contract is stored in workflow_definitions WHERE workflow_type='draft_quote' — risk level 2 (low-med), expected duration 15 minutes, three HITL gates at steps 1, 2, and 4.

The defining property: nothing about pricing is invented. Every line traces to pricing_master, assembly_cost_rollup, or an explicit override that Mike typed into chat. The AI narrates; the database decides.

Trigger conditions

• Customer emails quotes@ or orders@ asking for pricing on an assembly.
• Sales call wraps up and Mike needs to send a follow-up quote.
• RFP comes in via the intake page (Dropbox or upload) and needs a structured response.
• Existing customer wants pricing on a new pack/format of an item they already buy.
• Mike wants to test a price point before formalizing it into the pricing_master.

Customer Z requests pricing on assembly A123

The five beats

1. 01

   Gather context (auto)

   The get_customer_360 chat tool pulls the customer record from the D1 mirror, the trailing-90-day order history, current AR aging buckets, and any flags from customer_health_scores. This becomes the prompt context for the line-builder.

   Reads customers, so_lines, customer_health_scores, ar_aging

   Time ~600ms
2. 02

   Build lines (HITL)

   propose_quote looks up each assembly's cost in assembly_cost_rollup, applies the customer's negotiated margin band (or the platform default if none), and writes draft rows to quote_lines with status='draft'. A proposed_action row lands in Mike's queue.

   Writes quotes, quote_lines (draft), proposed_actions

   HITL step 1
3. 03

   Approve lines (HITL)

   Mike taps Approve in /proposed-actions.html. The quote rows flip to status='approved' and the atomic claim (R560 pattern) prevents double-approval. Mike can edit individual lines before approving — those edits write back to the same quote_lines rows.

   Writes quote_lines.status='approved'

   Time Mike approves within 5–30 min typically
4. 04

   Render PDF (auto)

   render_quote_pdf generates the HTML artifact (Pages Function at /quote/) and converts to PDF via Browser Rendering. Both land in R2 under quotes//. The HTML link is what gets put in the email body.

   Writes R2 quotes/

   Time ~4–8s
5. 05

   Send to customer (HITL)

   A draft email row in outbound_email_log with status='pending_review'. Mike skims, taps Send, the email goes out via Email Routing. Event quote.sent is appended to the events ledger. The customer-facing tracked link is logged for open/click tracking.

   Writes outbound_email_log, events

   HITL step 4 (send)

What's different after the workflow runs

• A row in quotes with status sent and a link to the HTML artifact at /quote/.
• Per-line audit trail in quote_lines tying back to pricing_master and assembly_cost_rollup.
• An outbound email row with status='sent', message ID, and tracked open/click links.
• An event quote.sent in the events ledger so customer health + sales velocity watchers see it.
• Mike's chat history retains the full transcript for next-time recall.

What can go wrong and how to recover

Adjacent workflows + diagrams

Code paths + invariants

Dated trail · spot stale claims

Dated trail of when this doc was last touched, what changed, and what to look at if it feels stale.

The machine-readable spec

Canonical fields, table names, endpoint signatures. What code should match, what tests should assert. workflow_type · customer_quote · risk_level · 2.

Inputs (required + optional)

D1 tables written

Endpoints called

Events fired

It broke at 2am — what now

Different from "how do I use this." This is the page Mike pulls up when something is wrong: logs to check, recovery steps, who to escalate to.

Logs to check

• workflow_run_log · top-level run audit
• workflow_step_log · per-step trace
• workflow_verify_results · post-window verify outcomes
• cron_locks · stuck cron lock detection
• events · workflow.completed / workflow.failed event trail
• reflexion_log · per-run narrative (if reflexion_enabled)
• npx wrangler tail · live Worker logs

Kill switch · emergency stop

If this workflow is misbehaving in a high-impact way (creating bad proposed_actions in volume, pushing wrong things to NS), flip a kill switch:

• kill:ns_writes · stops every NS push platform-wide
• kill:proposed_apply · stops HITL approvals from executing fan-out
• kill:high_risk_ops · stops risk_level >= 4 fan-out

See kill-switches-state-machine.html for the full state machine + recovery procedure.

Escalation

Primary: Mike Levine (single-admin) · mikelevine@globalfoodsolutions.co. For prolonged outage during business hours, notify warehouse lead + accounting lead so they can defer dependent work.

What's not done · what's uncertain

What's not done, what's uncertain, what we punted. Captured so it survives context switches and doesn't die in someone's head.

• STUB
  Auto-rollover quotes (90-day expiry)

  Today quotes don't auto-expire; they just sit. Need a cron that flags quotes >90 days old and emails Mike to refresh or close.

• OPEN
  Should quote acceptance trigger a sales order automatically?

  Today acceptance is silent — Mike still creates the SO. Could be auto-staged as proposed_action with action_type='create_sales_order'.

• DECISION
  Per-customer quote templates

  Schools want different headers than commercial. Today one template; need a customer_quote_template field.

• STUB
  Quote analytics (acceptance rate, time-to-decision)

  Events fire but no dashboard reads them yet.