Guided Journeys

How to run it: open the presenter notes in a separate window and keep that window off the shared screen. Then start presenting here. The notes window follows every step and its Back / Next buttons drive this window; → and ← work here too. What the audience sees is a thin progress line and the page itself — no talk track.

1. 1Finance 360Act 1 · the problem — 112% combined ratio is an underwriting loss

   presenter notes

   DoPoint at the KPI strip: combined ratio 112%, PAT ₹93.86 Cr, loss ratio 79% — then the P&L walk below.SayGWP grew 11% to ₹5,548 Cr — and we lost money underwriting it. Combined ratio 112%. PAT is positive only because ₹598 Cr of investment income covers the hole. Growth is not the problem. Leakage inside claims is.

   then ↓
2. 2Health 360Act 1 · which book — Health at ~99% loss ratio

   presenter notes

   DoPoint at the segment split: Group Health at 110% ICR against Retail at 93%.SayNow we know which book. Where inside it? Claims. Let us open one.

   then ↓
3. 3Document IntelligenceAct 2 · one claim enters — read the bill

   presenter notes

   DoClick Bill — scan, then Extract & check. If the room asks for a real one, Bill — phone photo or Upload your own — a skewed photo can drop a digit, which is itself worth saying out loud.SayThat is an image of a bill, read by the vision model; the arithmetic is ours. Ten line items sum to ₹3,46,900 against a printed ₹3,56,600 — a ₹9,700 gap, and that is a fact, not an opinion. Consumables at 36% against a 15% norm, a single-private room billed against a twin-sharing entitlement, trocar and sutures unbundled from the package.

   then ↓
4. 4Coverage ReasoningAct 2 · is it covered? — rules decide, model explains

   presenter notes

   DoClick Health · PED inside waiting, then Assess coverage.SayREFER, citing CV-03. The rules engine made the call; the model explained it and listed what to collect before settling. It never got a vote.

   then ↓
5. 5Fraud & SIU 360Act 2 · fraud triage with a held-out backtest

   presenter notes

   DoClick Health · bill inflation → Run AI triage. Then click Re-score on any row of the case table. Finally run Motor · routine (clean).Say0.81, High, hold payment — with precedent HLT-FR-3301 from the SIU book. Re-score holds that case OUT of its own reference book and scores it blind. And the clean claim scores 0.05 — an engine that flags everything is worthless.

   then ↓
6. 6Hospital & Provider 360Act 3 · from the claim to the provider

   presenter notes

   DoPoint at the provider anomaly table — every factor is shown next to the score.SayThe hospital behind that bill bills above peers with an elevated denial rate. Computed, auditable, no model in the loop.

   then ↓
7. 7Fraud Network AnalysisAct 3 · from the provider to the network

   presenter notes

   DoClick SIU memo on the top hub.SayEighty-five repeat-claimant hubs — one customer, several policies, many claims. One claim is a number; the network is the story. And notice the empty co-occurrence panel: the engine reports a true negative rather than hiding it.

   then ↓
8. 8SimulationAct 4 · what is fixing it worth?

   presenter notes

   DoClick Combined ratio below 110%, then drag Leakage recovered yourself. Then click NATCAT.SayOne combined-ratio point is ₹37.5 Cr. This is arithmetic against the audited book — move a slider, it recomputes. NATCAT is the capital stress: ₹800 Cr gross, net after the treaty.

   then ↓
9. 9Decision StackAct 4 · the whole stack on one subject

   presenter notes

   DoClick Health loss ratio at ~99% → Run decision stack.SaySee, Predict, Reason, Simulate, Decide, Act — streamed in, with owners and systems. Predict and Simulate carry a Modeled badge. We say what they are.

   then ↓
10. 10AI Agentic WorkforceAct 5 · act — the agentic workforce

    presenter notes

    DoClaims Agent → fraud alert. Fraud Detection → network analysis. Provider → pre-auth.SayThirty named skills, each reasoning over its own engine's output. Agentic here means audited, not animated.

    then ↓
11. 11Activity & DigestAct 5 · the audit trail

    presenter notes

    DoPoint at the three agent runs that just landed.SayEvery run writes to the same decision log the Today queue reads. That is the first half of the feedback loop — the second half, models learning from outcomes, is marked planned.

    then ↓
12. 12Lapse, Retention & Sales AIAct 5 · close the loop on growth

    presenter notes

    DoClick Next best action on the top retention row.SayLapse and cross-sell scored in code; the model picks the move and writes the script. Retain, cross-sell, win-back — grounded on the features, never free-floating.

    then ↓
13. 13SciKIQ Fabric MapAct 6 · the architecture, badged honestly

    presenter notes

    DoShow the badge tally under the decision stack: 58 computed · 38 live AI · 6 planned.SayEvery bullet on the architecture opens the thing that implements it — and says what it is. Planned means not built. We say so.

    then ↓
14. 14Connectors & IntegrationsAct 6 · what is wired, what is not

    presenter notes

    DoScroll to Not connected.SaySix connected, nine not. SSO, TPA, GARV, Power BI need the real systems. Listing them as available would be the easiest claim in this room to disprove — so we do not.

    Outcome

1. 1Strategy & Goalsthe four pillars & targets the plan rests onthen ↓
2. 2Health 360fix the ~99% health loss ratio (pricing · network · fraud)then ↓
3. 3Motor 360the motor P&L: OD loss vs tariffed-TP profitthen ↓
4. 4Enterprise 360is the consolidated book on track?then ↓
5. 5Finance 360combined ratio, claims & the P&Lthen ↓
6. 6Products & Lines 360products & lines by loss ratiothen ↓
7. 7Turnaround & Growththe turnaround & growth to ₹10,000 cr GWPOutcome

1. 1AI Use Casesthe AI use-case portfolio: LEO · FG ConGo · I-Assist · i-ViSS · IRISthen ↓
2. 2Fraud & SIU 360fraud & SIU: 8-10% leakage, detection & recoverythen ↓
3. 3Claims & Renewals 360claims TAT, cashless & renewals the AI speeds upthen ↓
4. 4Programs & Digital 360the digital programs: TCS BaNCS, IRIS, LEO rolloutthen ↓
5. 5Products & Lines 360digital issuance & STP by product linethen ↓
6. 6Turnaround & Growthhow AI & fraud control bend the combined ratioOutcome

1. 1Finance 360combined ratio, claims & quality of earningsthen ↓
2. 2Cash 360premium collection, float & investment incomethen ↓
3. 3Products & Lines 360loss ratio & underwriting result by linethen ↓
4. 4Turnaround & Growthsolvency, capital & the path to ₹10,000 crOutcome

1. 1Ontology & Meshthe data mesh behind the metricsthen ↓
2. 2Company Hierarchythe three lenses (org · line · legal entity)then ↓
3. 3Org Roll-up 360footprint: 167 branches & zones, the data grainthen ↓
4. 4Value Creation Plantoday → the ₹10,000 cr value-creation leversthen ↓
5. 5Programs & Digital 360the turnaround & digital programsthen ↓
6. 6Customer Masterresolve the policyholder golden record (MDM)then ↓
7. 7Turnaround & Growthcombined-ratio turnaround, solvency & growth to ₹10,000 crOutcome

1. 1New Business & Underwriting 360new-business pipeline & underwriting funnelthen ↓
2. 2Channels & Customers 360channels: agency · brokers · bancassurance · directthen ↓
3. 3Quote & Proposal 360quote & proposal across Motor, Health & Commercialthen ↓
4. 4Claims & Renewals 360renewals & retention the channel carriesthen ↓
5. 5Branch & Distribution 360branch & bancassurance footprint (Central Bank 4,500+ branches)then ↓
6. 6Quote-to-Premium 360watch it convert quote → premiumOutcome

1. 1Control Towersense the cross-book signals & alertsthen ↓
2. 2AI Use Caseslet an AI use-case act on the alertthen ↓
3. 3Claims & Renewals 360claims, cashless & renewals healththen ↓
4. 4Workforce 360operations force & branch productivitythen ↓
5. 5Reinsurance & Network 360reinsurance, hospital & garage network riskOutcome