AI family protection as a distribution engine
Households renew every year. Most insurers only find out afterwards.
A family safety app households open every day, with an insurance engine underneath it that captures every policy they hold, tells you thirty days before each one renews, and converts the opportunity into a sale your agent closes.
renewal events surfaced in one year. Without the vault, an insurer sees only the events on the policies it already writes: 6 of them.
The mechanic, in four moves
Household engages, vault fills, renewal event fires, sale completes. Scroll to watch the app carry each move.
-
Move 1 of 4
The household engages
Safety is used weekly, not annually. Location circles, the Bubble, driving reports and an assistance button keep the platform present when circumstances change, at a cost per household price-led acquisition cannot reach.
Your circle
Youssef left the BubbleSchool zone, 16:42. Arrived home 17:05.82Driving score4Safe zones -
Move 2 of 4
The vault fills
Documents are uploaded and read. The partner sees the household’s entire insurance picture — motor, travel, home and health — every premium it pays elsewhere, and every renewal date in it.
Policy vault
Motor, Dacia Logan14 MarWritten by you. 4 180 MADHome, Casablanca02 JunAnother insurer. 2 640 MADTravel, family annual18 JulAnother insurer. 1 150 MADHealth, top-up01 NovAnother insurer. 6 300 MAD14 270MAD household spend per year648Protection Score -
Move 3 of 4
The renewal event fires
Thirty days out, the concierge opens the conversation with a reason-coded assessment: renew, switch, improve, add, consolidate, or do nothing. Every recommendation carries a visible reason.
Concierge
Your home cover renews in thirty days at 2 640 MAD. It excludes water damage, which caused two of the three claims in your building. I can quote a policy that includes it.
Reason codeIMPROVEExtraction confidence0.96Interpretation confidence0.71Interpretation below threshold, so the item was reviewed by a human before it reached the household.
-
Move 4 of 4
The sale completes
The customer buys on the digital rail, or stalls and is caught by the omnichannel contact centre and handed to a live agent with the household record, the vault and the quote already open.
Home, twelve months
Premium2 880Water damage included. Excess 1 500 MAD.Karim, agency Maarif. Household record, vault, and the quote already open on his screen.
Policy issued02 JunCommission to partner346 MADNext event in vault18 Jul, travelLines held with you: 2 of 4
A photographed document becomes a structured policy.
Select a field to see the clause it came from. Fictional policy wording, used for illustration. Amanati never publishes a real insurer’s document.
Atlas Assurance S.A., Casablanca. Policy 4471-MT-0093, private motor, twelve months from 14 March.
Section 2. Cover is provided for third party liability, fire, theft and accidental damage to the insured vehicle described in the schedule.
Section 2.4. The maximum amount payable for damage to the insured vehicle is 180 000 MAD, being the agreed value at inception.
Section 4.1. Each and every claim for accidental damage is subject to an excess of 3 500 MAD, borne by the insured.
Section 6.2. No indemnity is payable where the vehicle is driven by a person under 23 years of age, or by any person not named in the schedule.
Schedule. This policy expires at 23:59 on 13 March and is offered for renewal from 14 March.
Extraction confidence answers whether the system read the document correctly. Interpretation confidence answers whether the conclusion drawn from that reading is sound. The vault reports them separately, and low interpretation routes the item to human review before it can influence a recommendation.
One line is a moment. Four lines are a diary.
Switch the lines on to see what one household produces across a year. Motor brings the household in. The other three are where revenue per household compounds.
One line, one annual moment. The relationship is dormant for the other eleven months.
Premiums are illustrative Moroccan market figures for a single modelled household, not an average. Motor is in pilot, travel and home in build, health planned. Lines of business
Your case
One platform, four different commercial arguments.
Renewal control across four lines.
New business from app-generated leads, every household policy visible including the ones you do not write, and an agent force working a live opportunity diary instead of a call list. The primary KPI is the share of the household renewal events you control.
- Household renewal events controlledShare of all household events you see and workPrimary KPI, model assumption
- Policies per householdLines held with you once the vault is completeModel assumption
- Agent productivityQualified opportunities per agent per monthPartially evidenced
What we can evidence, and what the pilot will measure.
Amanati is pre-pilot on verified outcome data, so this section carries the evidence chain rather than results. Every link is classified. Nothing here is a measured Amanati figure. Proof and results
The claimSafety utility earns weekly household attention
The mechanismLocation, safe zones, driving reports and an assistance button are used against events in family life, not at renewal
StatusPartially evidenced
What the pilot measuresComparable family safety apps report weekly active use. Amanati’s pilot will measure weekly active households and session frequency per household.
The claimA complete vault reveals policies the partner does not write
The mechanismMulti-document extraction against a single household record, normalised across insurers
StatusEvidenced
What the pilot measuresExtraction is live on motor documents. The pilot will measure vault completion rate and fields per policy at accepted confidence.
The claimRenewal dates become a scheduled diary
The mechanismEvery extracted policy carries a renewal date, so the concierge can open thirty days out with a reason-coded assessment
StatusEvidenced
What the pilot measuresThe mechanism is built. The pilot will measure events surfaced, T minus 30 engagement rate and quote rate per event.
The claimReason-coded conversations convert better than price-led outreach
The mechanismA named gap, a known date and a stated premium to beat, in place of a cold approach
StatusAssumed
What the pilot measuresConversion on reason-coded events against the partner’s own renewal baseline.
The claimVerified behaviour improves motor loss ratio
The mechanismMonthly household reporting changes driving, and scored data sharpens selection and pricing
StatusAssumed
What the pilot measuresLoss ratio on the scored cohort against a matched control, over a minimum of four quarters.
Partner ROI model
Your book, your assumptions.
Every input is yours to change. The defaults match the full ROI model: a position we can defend, not one that flatters the output. The working is shown underneath.
1 788 273
MAD annual commission retained by the partner from app-generated sales, after the Amanati share.
- Active app households
- Renewal events surfaced per yearThe number your agents can work
- Quotes generated
- Policies sold, renewal and cross-sell
- Gross written premium, MAD
- Total commission pool, MAD
- Amanati share, MADIllustrative, set in the agreement
- Renewal events controlledPrimary KPI, vault completion times engagement
- Scored motor householdsCohort carrying the loss ratio test, assumed 1.5 to 4.0 points on a matched control
Active households equal book times adoption. Renewal events equal active households times policies per household times vault completion. Quotes equal events times engagement times quote rate. Sales equal quotes times conversion, plus cross-sell attach on vaulted households. The model excludes retention uplift, loss ratio improvement and premium under management, all three material and none yet measured. Treat the output as a scale indication, not a forecast.
Three deployment routes, running side by side.
Across all three, Amanati earns a percentage of the commission on every sale the app generates, covering lead delivery, the digital conversion tool, and the handoff to a live agent.
Recommended for an insurer
Insurer direct
You run the broker engine as your own digital channel, under your own brand, converting through your own contact centre.
How insurer direct worksRecommended for an insurer
Agent distribution
You fund a branded app and give it to your tied agents. They receive the leads, the cross-sell and the upsell, and keep the relationship.
How agent distribution worksRecommended for an insurer
Independent broker
A large broker deploys it across its existing customer base as its own retention and cross-sell engine.
How the broker route works
The controls a regulator asks about first.
Anti-stalking controls
Consent is explicit, visible to the person being located, and revocable by them. Covert tracking is not a configuration option.
Data sovereignty
Household data is held in market, under the partner’s regulatory perimeter, with the processing boundary named in the agreement.
Evidence on every field
Extracted fields carry a source link to the clause. Low interpretation confidence routes to human review before it reaches a household.
Score boundaries
The Protection Score is a coverage measure. It is not an underwriting score, not a credit proxy, and must not be used to decline or price a risk.
Start with the numbers for your own book.
Bring your book size, average premium and current cross-sell rate. We will return the renewal events the platform would surface on your portfolio and what they are worth, with every assumption named.
- 24hA reply with two dates and the name of who will be on the call.
- 48hYour numbers run through the model and returned as a one-page business case.
- 45mThe briefing itself: the mechanic, a live walk of the app and vault, and the pilot design.
Request received
Thank you. The right person will reply to your work email with the material prepared against your book.