A weekly briefing for each Vertical Head
IIB Board AI Pack, page 20. First phase. Built here on synthetic data.
Today. Vertical Heads follow their numbers through monthly reports and dashboards. A sharp change can sit unnoticed until someone spots it in the figures.
What would change. Every week AI writes a one-page note for each Vertical Head from that week's numbers: what moved, the likely reasons and what needs a look. It includes the vertical's weekly alert, and an analyst checks it before it goes out.
This week across the four verticals
Each headline is the first line of that Vertical Head's note for the week 28 September to 4 October 2026. Choose one to read the note below. A note goes out only after the analyst approves it. This role sees state totals only, so the notes name no district.
The typical question
Written from a fixed template, because the AI draft had the wrong number of paragraphs. This is the Weekly briefing's own text.
Telangana two-wheeler claims up 68% against the usual level
In the 6 weeks to 4 October 2026 the demo sample had 2,469 motor claims. That is 2.2% above the usual level for this time of year. This change is small and may be noise. Policies issued, new and renewed, were 2.7% below the usual level in the 6 weeks to 27 September. The file checks held back 1,310 policy rows from these weeks, more than the shortfall, so this fall is likely missing data, not lost sales.
Telangana two-wheeler claims were 77 in the last 6 weeks, against a usual level of about 46 for this time of year. That is 68% above usual, which is very unusual. By claim type, theft claims were 19 against about 3, own damage claims were 52 against about 38 and third party claims were 6 against about 5. 13 insurers have more claims than usual. The biggest single rise is 5 claims, at Insurer J.
Also worth watching. Haryana two-wheeler policies issued were down 10%, which is unusual. Part of this drop is likely missing data, not lost sales. The file checks held back 53 rows of Insurer K's September policies file in Haryana. Jharkhand two-wheeler policies issued were down 14%, which is very unusual. Part of this drop is likely missing data, not lost sales. The file checks held back 24 rows of Insurer K's September policies file in Jharkhand.
Madhya Pradesh two-wheeler claims were up 20%. Madhya Pradesh private car claims were up 26%. These are within normal variation.
1 movement passed the alert level. Please record an outcome on each alert below.
No AI was used for this text. The template was given the 5 biggest changes in motor and the totals for the demo sample, from the Weekly briefing. Insurer code names only. No personal data. State totals only for this role.
Show the figures the note was written from
{
"briefing_week": "28 September 2026 to 4 October 2026",
"claims_period": "24 August 2026 to 4 October 2026",
"policies_period": "17 August 2026 to 27 September 2026",
"policy_figures_weeks_behind_claims": 1,
"sample_totals": {
"claims_last_6_weeks": 2469,
"claims_usual_level": 2416,
"claims_change_pct": 2.2,
"claims_direction": "up",
"claims_in_briefing_week": 424,
"policies_issued_last_6_weeks": 37820,
"policies_issued_usual_level": 38889,
"policies_issued_change_pct": 2.7,
"policies_issued_direction": "down",
"policies_issued_means": "new and renewed policies",
"claims_how_unusual": "normal variation",
"policies_issued_how_unusual": "very unusual",
"policy_rows_held_back_by_the_file_checks": 1310,
"comparison": "every change is against the usual level for this time of year, not the previous period"
},
"insurers_in_total": 25,
"movements": [
{
"rank": 1,
"state": "Telangana",
"vehicle_class": "two-wheeler",
"measure": "claims",
"direction": "up",
"last_6_weeks": 77,
"usual_level": 46,
"change_pct": 68,
"how_unusual": "very unusual",
"alert": true,
"period": "24 August 2026 to 4 October 2026",
"drivers": {
"claim_type": [
{
"claim_type": "theft",
"last_6_weeks": 19,
"usual_level": 3,
"above_usual": 16
},
{
"claim_type": "own damage",
"last_6_weeks": 52,
"usual_level": 38,
"above_usual": 14
},
{
"claim_type": "third party",
"last_6_weeks": 6,
"usual_level": 5,
"above_usual": 1
}
],
"insurer": [
{
"insurer": "Insurer J",
"last_6_weeks": 6,
"usual_level": 1,
"above_usual": 5
},
{
"insurer": "Insurer F",
"last_6_weeks": 8,
"usual_level": 3,
"above_usual": 5
},
{
"insurer": "Insurer S",
"last_6_weeks": 8,
"usual_level": 4,
"above_usual": 4
}
]
},
"insurers_with_a_rise": 13,
"biggest_single_insurer": {
"insurer": "Insurer J",
"above_usual": 5
},
"theft": {
"last_6_weeks": 19,
"usual_level": 3,
"times_usual_level": 6
}
},
{
"rank": 2,
"state": "Haryana",
"vehicle_class": "two-wheeler",
"measure": "policies issued",
"direction": "down",
"last_6_weeks": 786,
"usual_level": 869,
"change_pct": 10,
"how_unusual": "unusual",
"alert": false,
"period": "17 August 2026 to 27 September 2026",
"drivers": {
"cover_type": [
{
"cover_type": "comprehensive",
"last_6_weeks": 385,
"usual_level": 437,
"below_usual": 52
},
{
"cover_type": "third party only",
"last_6_weeks": 341,
"usual_level": 377,
"below_usual": 36
}
],
"insurer": [
{
"insurer": "Insurer K",
"last_6_weeks": 42,
"usual_level": 66,
"below_usual": 24
},
{
"insurer": "Insurer Y",
"last_6_weeks": 44,
"usual_level": 56,
"below_usual": 12
},
{
"insurer": "Insurer S",
"last_6_weeks": 47,
"usual_level": 56,
"below_usual": 9
}
]
},
"insurers_with_a_fall": 17,
"biggest_single_insurer": {
"insurer": "Insurer K",
"below_usual": 24
},
"held_back": [
{
"insurer": "Insurer K",
"file": "September policies file",
"rows_held_back_in_state": 53,
"up_to_this_many_in_this_movement": 36,
"insurer_shortfall": 24,
"meaning": "likely missing data, not lost sales"
}
]
},
{
"rank": 3,
"state": "Madhya Pradesh",
"vehicle_class": "two-wheeler",
"measure": "claims",
"direction": "up",
"last_6_weeks": 116,
"usual_level": 96,
"change_pct": 20,
"how_unusual": "normal variation",
"alert": false,
"period": "24 August 2026 to 4 October 2026",
"drivers": {
"claim_type": [
{
"claim_type": "own damage",
"last_6_weeks": 88,
"usual_level": 72,
"above_usual": 16
},
{
"claim_type": "third party",
"last_6_weeks": 14,
"usual_level": 9,
"above_usual": 5
}
],
"insurer": [
{
"insurer": "Insurer W",
"last_6_weeks": 12,
"usual_level": 6,
"above_usual": 6
},
{
"insurer": "Insurer S",
"last_6_weeks": 9,
"usual_level": 6,
"above_usual": 3
},
{
"insurer": "Insurer Q",
"last_6_weeks": 5,
"usual_level": 2,
"above_usual": 3
}
]
},
"insurers_with_a_rise": 11,
"biggest_single_insurer": {
"insurer": "Insurer W",
"above_usual": 6
}
},
{
"rank": 4,
"state": "Madhya Pradesh",
"vehicle_class": "private car",
"measure": "claims",
"direction": "up",
"last_6_weeks": 61,
"usual_level": 48,
"change_pct": 26,
"how_unusual": "normal variation",
"alert": false,
"period": "24 August 2026 to 4 October 2026",
"drivers": {
"claim_type": [
{
"claim_type": "own damage",
"last_6_weeks": 54,
"usual_level": 41,
"above_usual": 13
},
{
"claim_type": "third party",
"last_6_weeks": 7,
"usual_level": 6,
"above_usual": 1
}
],
"insurer": [
{
"insurer": "Insurer Y",
"last_6_weeks": 7,
"usual_level": 2,
"above_usual": 5
},
{
"insurer": "Insurer D",
"last_6_weeks": 4,
"usual_level": 2,
"above_usual": 2
},
{
"insurer": "Insurer M",
"last_6_weeks": 3,
"usual_level": 1,
"above_usual": 2
}
]
},
"insurers_with_a_rise": 9,
"biggest_single_insurer": {
"insurer": "Insurer Y",
"above_usual": 5
}
},
{
"rank": 5,
"state": "Jharkhand",
"vehicle_class": "two-wheeler",
"measure": "policies issued",
"direction": "down",
"last_6_weeks": 462,
"usual_level": 535,
"change_pct": 14,
"how_unusual": "very unusual",
"alert": false,
"period": "17 August 2026 to 27 September 2026",
"drivers": {
"cover_type": [
{
"cover_type": "third party only",
"last_6_weeks": 202,
"usual_level": 239,
"below_usual": 37
},
{
"cover_type": "comprehensive",
"last_6_weeks": 234,
"usual_level": 262,
"below_usual": 28
},
{
"cover_type": "own damage only",
"last_6_weeks": 26,
"usual_level": 34,
"below_usual": 8
}
],
"insurer": [
{
"insurer": "Insurer K",
"last_6_weeks": 16,
"usual_level": 43,
"below_usual": 27
},
{
"insurer": "Insurer Y",
"last_6_weeks": 23,
"usual_level": 37,
"below_usual": 14
},
{
"insurer": "Insurer P",
"last_6_weeks": 35,
"usual_level": 46,
"below_usual": 11
}
]
},
"insurers_with_a_fall": 15,
"biggest_single_insurer": {
"insurer": "Insurer K",
"below_usual": 27
},
"held_back": [
{
"insurer": "Insurer K",
"file": "September policies file",
"rows_held_back_in_state": 24,
"up_to_this_many_in_this_movement": 17,
"insurer_shortfall": 27,
"meaning": "likely missing data, not lost sales"
}
]
}
],
"alerts": 1,
"rto_detail": false
}The analyst's check
Draft Version 1.
Only the Analyst role approves a note or sends it back. A note goes out once it is approved.
What moved in Motor
| Movement | Last 6 weeks | Usual level | Change | How unusual | Alert |
|---|---|---|---|---|---|
| Telangana two-wheeler claims | 77 | about 46 | +68.0% | Very unusual | alert |
| Haryana two-wheeler policies issued | 786 | about 869 | -10.0% | Unusual | |
| Madhya Pradesh two-wheeler claims | 116 | about 96 | +20.0% | Normal variation | |
| Madhya Pradesh private car claims | 61 | about 48 | +26.0% | Normal variation | |
| Jharkhand two-wheeler policies issued | 462 | about 535 | -14.0% | Very unusual |
The last 6 weeks are 24 August to 4 October 2026. Each figure is compared with the usual level for this time of year: the level of the 26 weeks before, scaled by how the same weeks went last year. Motor's figures are the Weekly briefing's, one movement per state and vehicle class. Open the Weekly briefing for the drivers and the outcomes.
Motor claims, week by week
20 weeks shown, up to 4 October 2026. The grey line is the usual weekly level for the last 6 weeks. Every state and vehicle class together.
This week's alert: Telangana two-wheeler claims, up 68%
Very unusual for this time of year. 77 in the last 6 weeks against about 46.
From use 4, claims early warning, the Motor vertical's weekly alert. A flag is for review by a person. Open claims early warning to record an outcome on each item.
Written by Azure OpenAI from the figures below. Every number is checked against them.
Hospital admissions and infection cases remain within normal variation, but six hospitals flagged for high billing
Hospital admissions in the demo sample were 7,045 in the last 6 weeks. This is 3.5% above the usual level for this time of year, which is 6,810. The change is within normal variation and may be noise. In the most recent week, admissions were 1,109, close to the usual weekly level of 1,135. The rise sits mainly in Rajasthan, West Bengal and Uttar Pradesh. Rajasthan had 318 admissions, 62 above usual. West Bengal and Uttar Pradesh each had 60 above usual. Most of the increase was for infection treatments, which were 187 above usual. Insurer B had the biggest single rise, with 64 above usual.
Admissions for infections such as dengue, malaria and typhoid were 2,779 in the last 6 weeks. This is 6.8% above the usual level of 2,603. This change is also within normal variation. In the most recent week, there were 342 admissions, below the usual weekly level of 434. The rise sits mainly in Delhi, Telangana and West Bengal. Delhi had 58 above usual, Telangana 49 and West Bengal 35. Most of the increase was for dengue fever, which was 102 above usual, and typhoid fever, which was 59 above usual. Insurer B again had the biggest single rise, with 52 above usual. The data alone does not show the reason for these changes.
The weekly alert flagged six hospitals for review by a person. The alert found that these hospitals billed far above similar hospitals for the same procedures and kept patients longer. The first items flagged were Vaidyavan Super Speciality Hospital, Haryana, and Shubhprana Hospital and Research Centre, Rajasthan. Vaidyavan billed 2.3 times and Shubhprana 2.2 times similar hospitals, and both kept patients about 1.5 times as long. The total billed above similar hospitals was ₹4.6 cr. These flags are not findings but are for review.
The AI was given 2 headline figures for Health against the usual level for this time of year, where each change sits, and this week's hospital outliers alert with 6 items. Insurer code names only. No personal data. State totals only for this role.
Show the figures the AI was given
{
"vertical": "Health",
"note_for": "the Health Vertical Head",
"briefing_week": "28 September to 4 October 2026",
"last_6_weeks": "24 August to 4 October 2026",
"comparison": "every change is against the usual level for this time of year, not the previous period",
"figures_are": "a demo sample of synthetic insurers, not the market",
"detail": "state totals only",
"measures": [
{
"measure": "hospital admissions",
"short_name": "Hospital admissions",
"counted_in": "admissions",
"last_6_weeks": 7045,
"usual_level": 6810,
"change_pct": 3.5,
"direction": "up",
"how_unusual": "normal variation",
"in_briefing_week": 1109,
"usual_weekly_level": 1135,
"where_it_sits": {
"by_state": [
{
"state": "Rajasthan",
"last_6_weeks": 318,
"usual_level": 256,
"above_usual": 62,
"pct_of_the_change": 27
},
{
"state": "West Bengal",
"last_6_weeks": 342,
"usual_level": 282,
"above_usual": 60,
"pct_of_the_change": 26
},
{
"state": "Uttar Pradesh",
"last_6_weeks": 584,
"usual_level": 524,
"above_usual": 60,
"pct_of_the_change": 25
}
],
"by_treatment": [
{
"treatment": "infection",
"last_6_weeks": 2779,
"usual_level": 2592,
"above_usual": 187,
"pct_of_the_change": 79
}
]
},
"insurers_with_a_rise": 15,
"insurers_in_total": 25,
"biggest_single_insurer": {
"insurer": "Insurer B",
"above_usual": 64
}
},
{
"measure": "admissions for infections such as dengue, malaria and typhoid",
"short_name": "Admissions for infections",
"counted_in": "admissions",
"last_6_weeks": 2779,
"usual_level": 2603,
"change_pct": 6.8,
"direction": "up",
"how_unusual": "normal variation",
"in_briefing_week": 342,
"usual_weekly_level": 434,
"where_it_sits": {
"by_state": [
{
"state": "Delhi",
"last_6_weeks": 245,
"usual_level": 187,
"above_usual": 58,
"pct_of_the_change": 33
},
{
"state": "Telangana",
"last_6_weeks": 226,
"usual_level": 177,
"above_usual": 49,
"pct_of_the_change": 28
},
{
"state": "West Bengal",
"last_6_weeks": 154,
"usual_level": 119,
"above_usual": 35,
"pct_of_the_change": 20
}
],
"by_illness": [
{
"illness": "Dengue fever",
"last_6_weeks": 732,
"usual_level": 630,
"above_usual": 102,
"pct_of_the_change": 58
},
{
"illness": "Typhoid fever",
"last_6_weeks": 393,
"usual_level": 334,
"above_usual": 59,
"pct_of_the_change": 34
}
]
},
"insurers_with_a_rise": 17,
"insurers_in_total": 25,
"biggest_single_insurer": {
"insurer": "Insurer B",
"above_usual": 52
}
}
],
"alert": {
"from_use": "Hospital outliers",
"headline": "Six hospitals bill far above similar hospitals",
"flags_are": "for review by a person, not findings",
"items": [
{
"name": "Vaidyavan Super Speciality Hospital, Haryana",
"why": "Billed 2.3 times similar hospitals for the same procedures and kept patients 1.5 times as long.",
"bill_vs_similar_hospitals": 2.26,
"stay_vs_similar_hospitals": 1.49,
"discharged_claims_compared": 66,
"insurers": 21,
"flagged_since_week_ending": "28 June 2026",
"new_this_week": false
},
{
"name": "Shubhprana Hospital and Research Centre, Rajasthan",
"why": "Billed 2.2 times similar hospitals for the same procedures and kept patients 1.5 times as long.",
"bill_vs_similar_hospitals": 2.24,
"stay_vs_similar_hospitals": 1.54,
"discharged_claims_compared": 38,
"insurers": 15,
"flagged_since_week_ending": "28 June 2026",
"new_this_week": false
},
{
"name": "Ayushkiran Healthcare, Tamil Nadu",
"why": "Billed 2.2 times similar hospitals for the same procedures and kept patients 1.6 times as long.",
"bill_vs_similar_hospitals": 2.18,
"stay_vs_similar_hospitals": 1.64,
"discharged_claims_compared": 77,
"insurers": 21,
"flagged_since_week_ending": "28 June 2026",
"new_this_week": false
},
{
"name": "Pranavriksha Hospital and Research Centre, Chandigarh",
"why": "Billed 2.1 times similar hospitals for the same procedures and kept patients 1.4 times as long.",
"bill_vs_similar_hospitals": 2.12,
"stay_vs_similar_hospitals": 1.41,
"discharged_claims_compared": 54,
"insurers": 18,
"flagged_since_week_ending": "28 June 2026",
"new_this_week": false
},
{
"name": "Swasthkiran Super Speciality Hospital, Delhi",
"why": "Billed 2.4 times similar hospitals for the same procedures and kept patients 1.5 times as long.",
"bill_vs_similar_hospitals": 2.36,
"stay_vs_similar_hospitals": 1.52,
"discharged_claims_compared": 51,
"insurers": 18,
"flagged_since_week_ending": "5 July 2026",
"new_this_week": false
},
{
"name": "Sukhvardhan Super Speciality Hospital, Maharashtra",
"why": "Billed 2.2 times similar hospitals for the same procedures and kept patients 1.7 times as long.",
"bill_vs_similar_hospitals": 2.16,
"stay_vs_similar_hospitals": 1.68,
"discharged_claims_compared": 80,
"insurers": 21,
"flagged_since_week_ending": "5 July 2026",
"new_this_week": false
}
],
"hospitals_flagged": 6,
"new_this_week": 0,
"compared_over": "6 July to 4 October 2026",
"compared_with": "hospitals of the same type in the same city tier, for the same procedures",
"billed_above_similar_hospitals": "₹4.6 cr"
}
}The analyst's check
Draft Version 1.
Only the Analyst role approves a note or sends it back. A note goes out once it is approved.
What moved in Health
| Measure | Last 6 weeks | Usual level | Change | How unusual | This week |
|---|---|---|---|---|---|
| Hospital admissionsHealth claims, counted by the week of admission | 7,045 | about 6,810 | +3.5% | Normal variation | 1,109 |
| Admissions for infectionsDengue, malaria, typhoid and other infections | 2,779 | about 2,603 | +6.8% | Normal variation | 342 |
The last 6 weeks are 24 August to 4 October 2026. Each figure is compared with the usual level for this time of year: the level of the 26 weeks before, scaled by how the same weeks went last year. Choose a row to chart it.
Hospital admissions, week by week
20 weeks shown, up to 4 October 2026. The grey line is the usual weekly level for the last 6 weeks.
This week's alert: Six hospitals bill far above similar hospitals
Billed 2.3 times similar hospitals for the same procedures and kept patients 1.5 times as long.
Billed 2.2 times similar hospitals for the same procedures and kept patients 1.5 times as long.
Billed 2.2 times similar hospitals for the same procedures and kept patients 1.6 times as long.
Billed 2.1 times similar hospitals for the same procedures and kept patients 1.4 times as long.
Billed 2.4 times similar hospitals for the same procedures and kept patients 1.5 times as long.
Billed 2.2 times similar hospitals for the same procedures and kept patients 1.7 times as long.
From use 8, hospital outliers, the Health vertical's weekly alert. A flag is for review by a person. Open hospital outliers to record an outcome on each item.
Written by Azure OpenAI from the figures below. Every number is checked against them.
Early death claims and late reporting both very unusual, mainly in Bihar PoS sales
Early death claims in the demo sample were very unusual in the last 6 weeks. There were 31 early death claims, much higher than the usual level of 12. In the latest week, there were 5, compared to the usual weekly level of 2.0. Most of this rise sits in Bihar, with 28 claims against the usual 8.3. Nearly all of these were through the PoS sales channel, which had 28 claims against the usual 8.7. Six out of eight insurers saw a rise, with Insurer E alone accounting for about 8 more than usual.
Death claims told late, more than 30 days after the death, were also very unusual. There were 46 such claims in the last 6 weeks, compared to the usual 28. In the latest week, there were 9, against the usual 4.6. Most of the increase was in Bihar, with 23 claims against the usual 8.8, and mainly through PoS, which had 23 claims against the usual 8.5. Bancassurance also saw a smaller rise. Six insurers saw more late claims, with Insurer E again the largest single contributor.
Total death claims were up by 13% to 162, compared to the usual 143. This is within normal variation and may be noise. The increase was mainly in Bihar, Rajasthan and Andhra Pradesh, and through the PoS channel. Six insurers saw a rise, with Insurer E again the largest part.
The weekly alert flagged a cluster of early death claims in PoS sales in Bihar. In the last 26 weeks, there were 62 early death claims in Bihar PoS, across five insurers, compared to fewer than 1 expected. Most of these were on policies without a medical check, and most were told to the insurer late. The alert is for review by a person, not a finding. The amount claimed on these flagged claims is ₹10.0 cr.
The AI was given 3 headline figures for Life against the usual level for this time of year, where each change sits, and this week's early death claims alert with 1 item. Insurer code names only. No personal data. State totals only for this role.
Show the figures the AI was given
{
"vertical": "Life",
"note_for": "the Life Vertical Head",
"briefing_week": "28 September to 4 October 2026",
"last_6_weeks": "24 August to 4 October 2026",
"comparison": "every change is against the usual level for this time of year, not the previous period",
"figures_are": "a demo sample of synthetic insurers, not the market",
"detail": "state totals only",
"measures": [
{
"measure": "early death claims, where the death came within a year of the policy start",
"short_name": "Early death claims",
"counted_in": "claims",
"last_6_weeks": 31,
"usual_level": 12,
"change_pct": 163.9,
"direction": "up",
"how_unusual": "very unusual",
"in_briefing_week": 5,
"usual_weekly_level": 2.0,
"where_it_sits": {
"by_state": [
{
"state": "Bihar",
"last_6_weeks": 28,
"usual_level": 8.3,
"above_usual": 20,
"pct_of_the_change": 102
}
],
"by_sales_channel": [
{
"sales_channel": "PoS",
"last_6_weeks": 28,
"usual_level": 8.7,
"above_usual": 19,
"pct_of_the_change": 100
}
]
},
"insurers_with_a_rise": 6,
"insurers_in_total": 8,
"biggest_single_insurer": {
"insurer": "Insurer E",
"above_usual": 7.7
}
},
{
"measure": "death claims told late, more than 30 days after the death",
"short_name": "Death claims told late",
"counted_in": "claims",
"last_6_weeks": 46,
"usual_level": 28,
"change_pct": 66.4,
"direction": "up",
"how_unusual": "very unusual",
"in_briefing_week": 9,
"usual_weekly_level": 4.6,
"where_it_sits": {
"by_state": [
{
"state": "Bihar",
"last_6_weeks": 23,
"usual_level": 8.8,
"above_usual": 14,
"pct_of_the_change": 78
}
],
"by_sales_channel": [
{
"sales_channel": "PoS",
"last_6_weeks": 23,
"usual_level": 8.5,
"above_usual": 15,
"pct_of_the_change": 79
},
{
"sales_channel": "bancassurance",
"last_6_weeks": 12,
"usual_level": 7.8,
"above_usual": 4.2,
"pct_of_the_change": 23
}
]
},
"insurers_with_a_rise": 6,
"insurers_in_total": 8,
"biggest_single_insurer": {
"insurer": "Insurer E",
"above_usual": 7.2
}
},
{
"measure": "death claims",
"short_name": "Death claims",
"counted_in": "claims",
"last_6_weeks": 162,
"usual_level": 143,
"change_pct": 13.4,
"direction": "up",
"how_unusual": "normal variation",
"in_briefing_week": 28,
"usual_weekly_level": 24,
"where_it_sits": {
"by_state": [
{
"state": "Bihar",
"last_6_weeks": 32,
"usual_level": 14,
"above_usual": 18,
"pct_of_the_change": 94
},
{
"state": "Rajasthan",
"last_6_weeks": 12,
"usual_level": 6.9,
"above_usual": 5.1,
"pct_of_the_change": 26
},
{
"state": "Andhra Pradesh",
"last_6_weeks": 7,
"usual_level": 3.5,
"above_usual": 3.5,
"pct_of_the_change": 18
}
],
"by_sales_channel": [
{
"sales_channel": "PoS",
"last_6_weeks": 37,
"usual_level": 12,
"above_usual": 25,
"pct_of_the_change": 131
}
]
},
"insurers_with_a_rise": 6,
"insurers_in_total": 8,
"biggest_single_insurer": {
"insurer": "Insurer E",
"above_usual": 11
}
}
],
"alert": {
"from_use": "Early death claims",
"headline": "Early death claims cluster in PoS sales in Bihar",
"flags_are": "for review by a person, not findings",
"items": [
{
"name": "Bihar · PoS",
"why": "62 early death claims in the last 26 weeks against about 0.07 expected, across five insurers. 95% were on policies issued without a medical check and 90% were told to the insurer more than 30 days after the death.",
"channel": "PoS",
"counted_over": "the last 26 weeks, 6 April to 4 October 2026",
"early_death_claims": 62,
"expected_by_chance": 0.07,
"times_expected": "over 100",
"early_death_claims_this_week": 4,
"insurers": 5,
"insurer_labels": [
"Insurer A",
"Insurer E",
"Insurer H",
"Insurer I",
"Insurer O"
],
"pct_on_policies_without_a_medical_check": 95,
"pct_told_late": 90,
"median_days_from_start_to_death": 90,
"how_unusual": "far beyond chance"
}
],
"amount_claimed_on_flagged_claims": "₹10.0 cr",
"insurers_spanned": 5,
"early_means": "a death within 365 days of the policy start",
"told_late_means": "told to the insurer more than 30 days after the death"
}
}The analyst's check
Draft Version 1.
Only the Analyst role approves a note or sends it back. A note goes out once it is approved.
What moved in Life
| Measure | Last 6 weeks | Usual level | Change | How unusual | This week |
|---|---|---|---|---|---|
| Early death claimsThe death came within a year of the policy start | 31 | about 12 | +163.9% | Very unusual | 5 |
| Death claims told lateTold to the insurer more than 30 days after the death | 46 | about 28 | +66.4% | Very unusual | 9 |
| Death claimsCounted by the week the insurer was told | 162 | about 143 | +13.4% | Normal variation | 28 |
The last 6 weeks are 24 August to 4 October 2026. Each figure is compared with the usual level for this time of year: the level of the 26 weeks before, scaled by how the same weeks went last year. Choose a row to chart it.
Early death claims, week by week
20 weeks shown, up to 4 October 2026. The grey line is the usual weekly level for the last 6 weeks.
Where each change sits, by place, group and insurer
Early death claims, where the death came within a year of the policy start
By state
| State | Last 6 weeks | Usual level | Difference |
|---|---|---|---|
| Bihar | 28 | 8.3 | 20 |
| Rajasthan | 2 | 0.0 | 2.0 |
| Punjab | 1 | 0.2 | 0.8 |
By sales channel
| Sales channel | Last 6 weeks | Usual level | Difference |
|---|---|---|---|
| PoS | 28 | 8.7 | 19 |
| direct | 1 | 0.0 | 1.0 |
| bancassurance | 2 | 1.4 | 0.6 |
By insurer
| Insurer | Last 6 weeks | Usual level | Difference |
|---|---|---|---|
| Insurer E | 9 | 1.3 | 7.7 |
| Insurer I | 8 | 1.7 | 6.3 |
| Insurer A | 5 | 2.0 | 3.0 |
| Insurer H | 6 | 3.5 | 2.5 |
| Insurer O | 2 | 1.5 | 0.5 |
6 of 8 insurers moved the same way.
Death claims told late, more than 30 days after the death
By state
| State | Last 6 weeks | Usual level | Difference |
|---|---|---|---|
| Bihar | 23 | 8.8 | 14 |
| Delhi | 3 | 0.9 | 2.1 |
| Andhra Pradesh | 2 | 0.5 | 1.5 |
By sales channel
| Sales channel | Last 6 weeks | Usual level | Difference |
|---|---|---|---|
| PoS | 23 | 8.5 | 14 |
| bancassurance | 12 | 7.8 | 4.2 |
| online | 2 | 0.7 | 1.3 |
By insurer
| Insurer | Last 6 weeks | Usual level | Difference |
|---|---|---|---|
| Insurer E | 10 | 2.8 | 7.2 |
| Insurer I | 6 | 2.1 | 3.9 |
| Insurer H | 11 | 7.6 | 3.4 |
| Insurer C | 4 | 1.0 | 3.0 |
| Insurer A | 6 | 3.8 | 2.2 |
6 of 8 insurers moved the same way.
Each row's usual level is its own share of the whole's usual level, with its own season where it has enough claims to show one. Difference is how far it sits above or below that. Insurers are shown by code name.
This week's alert: Early death claims cluster in PoS sales in Bihar
62 early death claims in the last 26 weeks against about 0.07 expected, across five insurers. 95% were on policies issued without a medical check and 90% were told to the insurer more than 30 days after the death.
Insurers: Insurer A, Insurer E, Insurer H, Insurer I, Insurer O.
From use 11, early death claims, the Life vertical's weekly alert. A flag is for review by a person. Open early death claims to record an outcome on each item.
Written by Azure OpenAI from the figures below. Every number is checked against them.
Property claims and amounts claimed are far above the usual level, mainly in Assam and due to flood
Property claims reported in the demo sample were very unusual in the last 6 weeks. There were 213 claims, which is more than double the usual level of 104 for this time of year. In the latest week, 98 claims were reported, compared to the usual weekly level of 17. Most of this rise sits in Assam, which had 107 claims against a usual level of 4. Flood was the main cause, with 143 claims compared to the usual 23. The data alone does not show why this happened.
The amount claimed on property claims was also very unusual. In the last 6 weeks, the demo sample saw claims of ₹68.9 cr, much higher than the usual level of ₹12.6 cr. In the latest week, ₹48.2 cr was claimed, while the usual weekly level is ₹2.1 cr. Assam accounted for ₹57.8 cr of the claims, compared to its usual level of ₹53.7 lakh. Flood was again the main cause, with ₹62.7 cr claimed against a usual level of ₹1.7 cr. The largest single insurer rise was Insurer E, with ₹13.8 cr above the usual level.
New property business, measured by sum insured of new property policies, was also very unusual. The demo sample wrote ₹4,362 cr in the last 6 weeks, compared to the usual level of ₹3,176 cr. In the latest week, ₹803 cr was written, above the usual weekly level of ₹529 cr. Most of the rise was in Maharashtra, with ₹2,251 cr written against a usual level of ₹541 cr. Warehouses made up most of this, with ₹2,321 cr written compared to the usual ₹619 cr. The largest single insurer rise was Insurer R, with ₹216 cr above the usual level.
The weekly alert from Exposure watch needs a look. It found that insured warehouse value is building up fast in one district of Maharashtra. Across the state, insured warehouse value in force is ₹6,935 cr, which is 1.36 times its 13-week average of ₹5,107 cr. There were 61 new warehouse policies in the last 4 weeks, written by 24 insurers. This flag is for review by a person, not a finding.
The AI was given 3 headline figures for Other Lines against the usual level for this time of year, where each change sits, and this week's exposure watch alert with 1 item. Insurer code names only. No personal data. State totals only for this role.
Show the figures the AI was given
{
"vertical": "Other Lines",
"note_for": "the Other Lines Vertical Head",
"briefing_week": "28 September to 4 October 2026",
"last_6_weeks": "24 August to 4 October 2026",
"comparison": "every change is against the usual level for this time of year, not the previous period",
"figures_are": "a demo sample of synthetic insurers, not the market",
"detail": "state totals only",
"measures": [
{
"measure": "property claims reported",
"short_name": "Property claims",
"counted_in": "claims",
"last_6_weeks": 213,
"usual_level": 104,
"change_pct": 105.5,
"direction": "up",
"how_unusual": "very unusual",
"in_briefing_week": 98,
"usual_weekly_level": 17,
"where_it_sits": {
"by_state": [
{
"state": "Assam",
"last_6_weeks": 107,
"usual_level": 4.0,
"above_usual": 103,
"pct_of_the_change": 94
}
],
"by_cause": [
{
"cause": "flood",
"last_6_weeks": 143,
"usual_level": 23,
"above_usual": 120,
"pct_of_the_change": 110
}
]
},
"insurers_with_a_rise": 23,
"insurers_in_total": 25,
"biggest_single_insurer": {
"insurer": "Insurer L",
"above_usual": 12
}
},
{
"measure": "amount claimed on property claims",
"short_name": "The amount claimed on property claims",
"counted_in": "rupees",
"last_6_weeks": "₹68.9 cr",
"usual_level": "₹12.6 cr",
"change_pct": 446.0,
"direction": "up",
"how_unusual": "very unusual",
"in_briefing_week": "₹48.2 cr",
"usual_weekly_level": "₹2.1 cr",
"where_it_sits": {
"by_state": [
{
"state": "Assam",
"last_6_weeks": "₹57.8 cr",
"usual_level": "₹53.7 lakh",
"above_usual": "₹57.3 cr",
"pct_of_the_change": 102
}
],
"by_cause": [
{
"cause": "flood",
"last_6_weeks": "₹62.7 cr",
"usual_level": "₹1.7 cr",
"above_usual": "₹61.0 cr",
"pct_of_the_change": 108
}
]
},
"insurers_with_a_rise": 21,
"insurers_in_total": 25,
"biggest_single_insurer": {
"insurer": "Insurer E",
"above_usual": "₹13.8 cr"
}
},
{
"measure": "sum insured of new property policies",
"short_name": "New property business",
"counted_in": "rupees",
"last_6_weeks": "₹4,362 cr",
"usual_level": "₹3,176 cr",
"change_pct": 37.4,
"direction": "up",
"how_unusual": "very unusual",
"in_briefing_week": "₹803 cr",
"usual_weekly_level": "₹529 cr",
"where_it_sits": {
"by_state": [
{
"state": "Maharashtra",
"last_6_weeks": "₹2,251 cr",
"usual_level": "₹541 cr",
"above_usual": "₹1,710 cr",
"pct_of_the_change": 144
}
],
"by_property_type": [
{
"property_type": "warehouses",
"last_6_weeks": "₹2,321 cr",
"usual_level": "₹619 cr",
"above_usual": "₹1,702 cr",
"pct_of_the_change": 143
}
]
},
"insurers_with_a_rise": 18,
"insurers_in_total": 25,
"biggest_single_insurer": {
"insurer": "Insurer R",
"above_usual": "₹216 cr"
}
}
],
"alert": {
"from_use": "Exposure watch",
"headline": "Insured warehouse value is building up fast in one district of Maharashtra",
"flags_are": "for review by a person, not findings",
"items": [
{
"name": "Maharashtra, warehouses",
"why": "Insured warehouse value is building up fast in one district of Maharashtra. Across the state it is up 35.8% on its 13-week average.",
"insured_value_in_force": "₹6,935 cr",
"average_of_the_13_weeks_before": "₹5,107 cr",
"times_its_average": 1.36,
"new_policies_in_the_last_4_weeks": 61,
"insurers": 24,
"districts_flagged_in_the_state": 1
}
],
"false_flags_over_weeks_replayed": {
"weeks_replayed": 91,
"false_flags": 0
}
}
}The analyst's check
Draft Version 1.
Only the Analyst role approves a note or sends it back. A note goes out once it is approved.
What moved in Other Lines
| Measure | Last 6 weeks | Usual level | Change | How unusual | This week |
|---|---|---|---|---|---|
| Property claimsClaims reported, for every cause | 213 | about 104 | +105.5% | Very unusual | 98 |
| The amount claimed on property claimsThe amount asked for on the claims reported | ₹68.9 cr | about ₹12.6 cr | +446.0% | Very unusual | ₹48.2 cr |
| New property businessSum insured of new property policies | ₹4,362 cr | about ₹3,176 cr | +37.4% | Very unusual | ₹803 cr |
The last 6 weeks are 24 August to 4 October 2026. Each figure is compared with the usual level for this time of year: the level of the 26 weeks before, scaled by how the same weeks went last year. Choose a row to chart it.
Property claims, week by week
20 weeks shown, up to 4 October 2026. The grey line is the usual weekly level for the last 6 weeks.
Where each change sits, by place, group and insurer
Property claims reported
By state
| State | Last 6 weeks | Usual level | Difference |
|---|---|---|---|
| Assam | 107 | 4.0 | 103 |
| Uttarakhand | 5 | 0.9 | 4.1 |
| Maharashtra | 17 | 14 | 2.8 |
By cause
| Cause | Last 6 weeks | Usual level | Difference |
|---|---|---|---|
| flood | 143 | 23 | 120 |
| burglary | 21 | 18 | 2.8 |
| earthquake | 2 | 1.5 | 0.5 |
By insurer
| Insurer | Last 6 weeks | Usual level | Difference |
|---|---|---|---|
| Insurer L | 20 | 7.6 | 12 |
| Insurer C | 12 | 2.4 | 9.6 |
| Insurer Q | 16 | 7.5 | 8.5 |
| Insurer E | 14 | 6.2 | 7.8 |
| Insurer B | 12 | 4.6 | 7.4 |
23 of 25 insurers moved the same way.
Amount claimed on property claims
By state
| State | Last 6 weeks | Usual level | Difference |
|---|---|---|---|
| Assam | ₹57.8 cr | ₹53.7 lakh | ₹57.3 cr |
| Andhra Pradesh | ₹2.7 cr | ₹38.7 lakh | ₹2.3 cr |
| Kerala | ₹1.2 cr | ₹13.5 lakh | ₹1.0 cr |
By cause
| Cause | Last 6 weeks | Usual level | Difference |
|---|---|---|---|
| flood | ₹62.7 cr | ₹1.7 cr | ₹61.0 cr |
| burglary | ₹2.0 cr | ₹1.3 cr | ₹71.1 lakh |
| earthquake | ₹13.3 lakh | ₹4.5 lakh | ₹8.8 lakh |
By insurer
| Insurer | Last 6 weeks | Usual level | Difference |
|---|---|---|---|
| Insurer E | ₹14.3 cr | ₹49.1 lakh | ₹13.8 cr |
| Insurer K | ₹8.6 cr | ₹94.3 lakh | ₹7.6 cr |
| Insurer P | ₹6.2 cr | ₹8.4 lakh | ₹6.1 cr |
| Insurer I | ₹6.0 cr | ₹1.0 cr | ₹5.0 cr |
| Insurer N | ₹4.5 cr | ₹31.6 lakh | ₹4.1 cr |
21 of 25 insurers moved the same way.
Sum insured of new property policies
By state
| State | Last 6 weeks | Usual level | Difference |
|---|---|---|---|
| Maharashtra | ₹2,251 cr | ₹541 cr | ₹1,710 cr |
| Andhra Pradesh | ₹209 cr | ₹106 cr | ₹103 cr |
| Goa | ₹64.7 cr | ₹2.7 cr | ₹62.0 cr |
By property type
| Property type | Last 6 weeks | Usual level | Difference |
|---|---|---|---|
| warehouses | ₹2,321 cr | ₹619 cr | ₹1,702 cr |
| houses | ₹121 cr | ₹110 cr | ₹11.5 cr |
| shops | ₹57.0 cr | ₹49.6 cr | ₹7.4 cr |
By insurer
| Insurer | Last 6 weeks | Usual level | Difference |
|---|---|---|---|
| Insurer R | ₹338 cr | ₹122 cr | ₹216 cr |
| Insurer E | ₹342 cr | ₹186 cr | ₹156 cr |
| Insurer J | ₹221 cr | ₹86.8 cr | ₹134 cr |
| Insurer C | ₹229 cr | ₹99.4 cr | ₹129 cr |
| Insurer Y | ₹263 cr | ₹139 cr | ₹124 cr |
18 of 25 insurers moved the same way.
Each row's usual level is its own share of the whole's usual level, with its own season where it has enough claims to show one. Difference is how far it sits above or below that. Insurers are shown by code name.
This week's alert: Insured warehouse value is building up fast in one district of Maharashtra
Insured warehouse value is building up fast in one district of Maharashtra. Across the state it is up 35.8% on its 13-week average.
From use 14, exposure watch, the Other Lines vertical's weekly alert. A flag is for review by a person. Open exposure watch to record an outcome on each item.
How we would measure it
Each note against its planted problem
| Vertical | Planted problem | In the alert | Named in the note | Note written by |
|---|---|---|---|---|
| Motor | Two-wheeler claims and thefts rising in Telangana | Yes Ranked 1 of 5 movements, an alert. | Yes | The template |
| Health | 6 hospitals billing about 2.1 times similar hospitals | Yes 6 of 6 planted hospitals flagged, and no other. | Yes | Azure OpenAI |
| Life | Early death claims on PoS policies in 2 districts of Bihar | Yes The Bihar PoS cluster is flagged at state level. | Yes | Azure OpenAI |
| Other Lines | Insured warehouse value building up in one district of Maharashtra | Yes Maharashtra warehouses flagged. | Yes | Azure OpenAI |
Named means the note's own words name the planted place: Telangana two-wheelers for Motor, one of the planted hospitals for Health, Bihar and PoS for Life, and Maharashtra warehouses for Other Lines. This is a check on words, not a judgement of how good the note is. That judgement is the analyst's.
How it works
Every week the page gathers each vertical's figures: its headline numbers for the last 6 weeks against the usual level for this time of year, where each change sits, and the vertical's weekly alert. Azure OpenAI writes a short note from those figures only. Every number in the note is checked against them, and so is the house style. A draft that fails is rewritten once, then a fixed template is used, and the page says which wrote each note. Motor's note is the Weekly briefing, written the same way. The analyst then approves each note or sends it back with a comment.
The method in full, and the technical terms
The usual level for the last 6 weeks is the weekly level of the 26 weeks before them, scaled by a seasonal index: how the same 6 weeks last year compared with the 26 weeks before those. Early death claims and death claims told late borrow the index of all death claims, and the amount claimed on property borrows the index of property claims, because the smaller series are too thin to show a season of their own.
How unusual a change is comes from a z-score: the gap from the usual level divided by its expected spread. The spread allows for chance in the counts, for week to week swings seen in the 26 weeks (measured robustly, so one storm week does not set it), and for the uncertainty in last year's index. Very unusual is 2.5 or more, unusual is 2 or more, the same thresholds as the Weekly briefing.
Where a change sits uses the same rule as the Weekly briefing: each state, district, group or insurer's figure less its own usual level, the levels adding up to the whole's. Each keeps its own season where it has the volume to show one, pulled towards the whole's season when it is small. The note only names those that carry a sixth of the change or more.
Notes are kept against a fingerprint of their figures (a hash), so the same week's figures are written once. Notes the AI wrote are kept on disk, so an approved note stays the same after a restart. Reviews are kept in the app database, one row per action, in the table briefing_review.
No personal data is read for this page: the figures come from the prepared weekly views, which hold none. The AI reads insurer code names only. The AI drafts and an analyst decides what goes out. This use needs no outside data, so nothing on it is mock.