◇ Zen Insurance · Nova AI

← queue

LEAD-2026-0429

95% confidence ? · rec: refer▶ View workflow run
Runbook play: Electrical Systems— Nova recommends Review & approve
judgment: electrical
Swimming PoolsRoof ClassElectrical SystemsPlumbingFire Protection & Defensible Space
broker tier-3judgment: electrical25 provided · 28 service-filled · 5 to ask
playbook:
🔒 Approval locked42 items still to resolve — 14 PlanDesign fields · 28 open items. Fill each with a source + reasoning (Nova learns from every fix) to unlock approval.
Property Address
120 Orchard Ave
City
Fresno
State
CA
Coverage A (Dwelling)
360000
Year Built
1978
Nova AI underwriting copilotProvisions — decide, and teach me as you go

Every provision is a decision I learn from. Approve, change the value, waive, or request it from the customer — with the source and your reasoning. I capture each one to my vector memory attributed to you (a full audit trail), and I use it to make a sharper per-provision recommendation next time.

5 clean · folded
Decision-tree outcomesthe DB rule engine’s per-point verdicts (clear / refer / decline)5 points · 2 to review

UW Decisions · 3 clear · 2 need attention · from decision_rules (DB)

◐ Refer to UWElectrical SystemsConstruction
No deterministic rule matched for Electrical Systems — refer to underwriter.
⚠ Triage — missing infoPlumbingPlumbing
A rule that would decide Plumbing is blocked by missing input(s): Water Heater Type, Water Heater Age (years), Water Heater Location.
✉ email producer: Water Heater Type✉ email producer: Water Heater Age (years)✉ email producer: Water Heater Location
Item-level approvals & request-info28 open · 0 approved · 0 unselected — bulk approve or compose the applicant ask28 items · 7 to ask
28 open · 0 approved · 0 unselected
Broker Tier97%rule-safe
UW AI Agent
broker_tier = Tier 3
Source text
“prefill (internal)”
— autofill
Insured has a primary policy with Zen Insurance97%rule-safe
UW AI Agent
has_primary_policy_with_carrier = No
Source text
“prefill (internal)”
— autofill
KYC Score70%reasoned
UW AI Agent
kyc_score = n/a
Source text
“prefill (kyc)”
— autofill
Replacement Cost (Dwelling)97%reasoned
UW AI Agent
replacement_cost = 487000
Source text
“prefill (assessor)”
— autofill
Purchase Year97%rule-safe
UW AI Agent
property_purchase_date = 2012
Source text
“prefill (assessor)”
— autofill
Lot Acreage92%rule-safe
UW AI Agent
acreage = 0.72
Source text
“prefill (mls)”
— autofill
Siding Material92%vision-input
UW AI Agent
siding_material = Wood
Source text
“prefill (aerial)”
— autofill
Foundation Type97%rule-safe
UW AI Agent
foundation_type = Post & Pier
Source text
“prefill (assessor)”
— autofill
Total Living Square Footage97%rule-safe
UW AI Agent
square_feet = 2730
Source text
“prefill (assessor)”
— autofill
Number of Stories97%rule-safe
UW AI Agent
num_stories = 3
Source text
“prefill (assessor)”
— autofill
Year Roof Replaced92%evidence-verify
UW AI Agent
roof_replacement_year = 2022
Source text
“prefill (permits)”
— autofill
Roof Surface Material92%vision-input
UW AI Agent
roof_material = Concrete Tile
Source text
“prefill (aerial)”
— autofill
Electrical Panel Size (amps)92%evidence-verify
UW AI Agent
electrical_panel_size_amps = 200
Source text
“prefill (permits)”
— autofill
General Plumbing Age (years)92%evidence-verify
UW AI Agent
plumbing_age_years = 22
Source text
“prefill (permits)”
— autofill
Miles to Fire Department97%rule-safe
UW AI Agent
distance_to_fire_department = 2.1
Source text
“prefill (hazard)”
— autofill
Fire Department Type / Staffing97%rule-safe
UW AI Agent
fire_department_type = Volunteer
Source text
“prefill (hazard)”
— autofill
Distance to Nearest Fire Hydrant (Feet)97%rule-safe
UW AI Agent
dist_to_nearest_fire_hydrant = 420
Source text
“prefill (hazard)”
— autofill
Road Access to Home97%rule-safe
UW AI Agent
road_access = Paved two-way
Source text
“prefill (hazard)”
— autofill
ISO Protection Class (PPC)97%rule-safe
UW AI Agent
protection_class = 6
Source text
“prefill (hazard)”
— autofill
Min Distance to Neighboring Home (ft)92%vision-input
UW AI Agent
min_distance_to_neighbor_ft = 220
Source text
“prefill (aerial)”
— autofill
Chapter 7A Compliant (WUI)97%reasoned
UW AI Agent
is_7a_compliant = No
Source text
“prefill (fhsz)”
— autofill
Slope Angle (degrees)97%rule-safe
UW AI Agent
slope_angle_deg = 13
Source text
“prefill (elevation)”
— autofill
Roof Fire Classification97%reasoned
UW AI Agent
roof_classification = Class A
Source text
“prefill (catmodel)”
— autofill
P(F) - Probability of Failure (fire simulation)97%reasoned
UW AI Agent
p_f = 0.700
Source text
“prefill (catmodel)”
— autofill
Siding Fire Classification97%reasoned
UW AI Agent
siding_classification = Combustible
Source text
“prefill (catmodel)”
— autofill
Vegetation Clearance / Defensible Space92%vision-input
UW AI Agent
vegetation_clearance = 100+ ft
Source text
“prefill (aerial)”
— autofill
Property currently listed for sale92%rule-safe
UW AI Agent
listed_for_sale = No
Source text
“prefill (mls)”
— autofill
Fire Department Response Time97%rule-safe
UW AI Agent
fire_dept_response_time = 15
Source text
“prefill (hazard)”
— autofill
Act on this lead

Nova pre-wrote a question for each missing item. For each ask, tell Nova why you need it — it learns what gets requested (and where some of it could be pulled instead of emailing the customer).

7 selected · a reason is required for each — Nova learns every ask
Capture reasoning on unclear provisionsfree-text UW rationale → Nova’s vector memory1 to explain

Unclear provisions — capture underwriter reasoning

These provisions were flagged missing and asked in the follow-up email; the reply came back as free text. Map it to the exact provision option and decide — your reasoning + source is remembered (vector memory) and auto-inferred on similar future cases, trending toward a trusted, auto-approvable pattern.
Plumbingunclear field: Water Heater Type
flagged missing · asked in the follow-up email
✦ Nova: paste the customer's actual words above — I learn from the raw reply, mapped by your judgment.

No vector memory yet for this provision — your first capture seeds it (384-dim MiniLM → vec_precedent).

Auto-filled PlanDesign formthe full binding form Nova pre-populated — item-approved fields are already confirmed; edit only to correct (edits still teach Nova)9 sections · 14 missing
Auto-filled PlanDesignThe customer submitted this instance; you are editing it. Each section folds — open one, edit a field, and every change is captured with a source + your reasoning, attributed to you and versioned (audit).
6 sections shown · 14 fields missing overall
Insured6 fields · 6 missing
Is the property owned by a trust?missing
Insured Phonemissing
Date of Birthmissing
Insured Occupationmissing
How many people live in the home?missing
KYC Scoresystem-filledverifyn/a
Property1 field · 1 missing
In gated community or multi-acre propertymissing
Construction1 field · 1 missing
Foundation TypeverifyPost & Pier
Plumbing1 field · 1 missing
Water Heater Typemissing
Protection3 fields · 3 missing
Fire Protective Devicesmissing
Opening Protectionmissing
Road Access to Homesystem-filledverifyPaved two-way
Wildfire2 fields · 2 missing
Siding Fire Classificationsystem-filledverifyCombustible
Vegetation Clearance / Defensible Spacesystem-filledverify100+ ft
Risk tree — first iterationour initial risk-appetite design, kept to show the iteration — superseded by provision-level approvals aligned with the runbook (▦)9 categories · 3 flagged
Risk hierarchy
Locationclear
Accountclear
Insuredopen
Is the property owned by a trust?
Date of Birth
How many people live in the home?
Primary Coveragesclear
Propertyclear
Constructionclear
Plumbingopen
Water Heater Type
Protectionopen
Fire Protective Devices
Wildfireclear
Terminal state

Follow-up — one consolidated email

A guardrail capping the agent at N asks (default 3) before handing to the UW — no spamming the customer/broker.

Nova pre-wrote 7 questions for the ONE consolidated email — review, trim, and send. Sending moves this lead to Awaiting reply and the email lands in Outbound.

inbound classification: New customer