Wow Moments Catalog¶
Twenty-four concrete UX moments across PropPie's three products. Each is designed to make a user stop and think "nobody else does this." Organised by the emotional category of the reaction, not by product.
For data attribute references, see ../20-data/data-attributes.md. For derived attribute math, see ../20-data/derived-attributes-spec.md.
Category A — "Show Your Work" Moments¶
Citation, source transparency, and radical auditability. These moments build trust by proving that every number comes from somewhere real.
A1. One-Click Source Trail¶
Product: All (Broker, Analytix, Fractional)
The moment: A user sees "median price in Hinjewadi: Rs 8,400/sqft" and taps the number. Instead of a tooltip, the screen splits to show the 87 individual IGR sale deeds that produced that median — each with deed number, registration date, parties, area, and price. They can sort, filter, and drill into any single deed. They realise this is not a broker's opinion; it is arithmetic over government records.
Data requirements: fin.sale_price, fin.sale_date, fin.sale_parties, area.carpet_sqft, loc.micromarket_id, plus per-attribute source_url and source_doc_id from the operational lineage columns.
Technical sketch: Every derived aggregate stores the list of contributing transaction IDs. The UI renders a drill-through panel using a pre-built query against the canonical store filtered by micromarket and date window. Source URLs deep-link to IGR Index-II pages.
Difficulty: Medium
Phase: 1
A2. Confidence Thermometer¶
Product: Broker, Fractional
The moment: The user asks Broker "What's the cap rate in Wakad for Grade A office?" The answer comes back with a visible confidence bar: "Cap rate 7.8% (confidence: high, based on 34 matched L&L + sale pairs from the last 12 months)." They ask the same question about a village near Alibaug and get: "Cap rate approximately 6.2% (confidence: low, based on 4 transactions — treat with caution)." The system doesn't pretend all answers are equal.
Data requirements: mkt.cap_rate_median, mkt.yield_benchmark, per-attribute confidence and confidence_method columns, transaction count (n) underlying each aggregate.
Technical sketch: Every derived attribute carries a confidence score (0-1) computed from sample size, source diversity, and recency. The UI maps this to a three-tier label (high/moderate/low) with the underlying n and date range always visible. Broker's LLM prompt includes the confidence tier so the narrative adapts language accordingly.
Difficulty: Easy
Phase: 1
A3. "Show Me the RERA Page"¶
Product: Broker, Fractional
The moment: A user is reviewing a project's claimed completion date. Broker says: "Promised completion: December 2026 (MahaRERA Form B, revision 3, filed 14 Mar 2025)." They tap the citation and see the actual MahaRERA filing page with the relevant field highlighted. Not a summary — the document itself. They feel like they have a lawyer's research assistant.
Data requirements: proj.promised_completion_dates (with version history), source_url, source_doc_page from operational lineage.
Technical sketch: The pipeline stores deep-link URLs to MahaRERA project pages and specific PDF page references for Form A/B filings. The UI opens an embedded PDF viewer scrolled to the cited page, with the relevant field highlighted using coordinate-based annotation stored at extraction time.
Difficulty: Medium
Phase: 1
A4. Developer Score Decomposition¶
Product: All
The moment: A user sees "Developer Trust Score: 62/100 (58th percentile vs. comparable Pune promoters)." They expand it. Five weighted bars appear — delivery delay (35%), complaints (25%), withdrawals (15%), cost overruns (15%), litigation (10%) — each with the raw input value, the normalised penalty, and a link to the underlying RERA filings. They see exactly why the score is what it is. No black box. They can disagree with the weights and still trust the inputs.
Data requirements: dev.trust_score and its five sub-components; proj.promised_completion_dates, proj.actual_completion_date, legal.complaints_maharera, fin.cost_overrun_pct, legal.litigation_records; per-component source_url.
Technical sketch: The trust score is pre-computed weekly per the formula in derived-attributes-spec.md. The decomposition view is a static rendering of the five penalty components with their raw values. Each component links to the underlying attribute's source documents. Defamation guardrails applied: percentile framing, no absolute labels, "request a review" link.
Difficulty: Easy
Phase: 1
A5. Assumption Editor on Projections¶
Product: Broker, Fractional
The moment: Broker shows a 5-year wealth trajectory chart with three probability bands. Below it: a row of editable assumption pills — "appreciation: 11%", "vacancy: 8%", "rent escalation: 5%", "holding period: 5 yr." The user drags appreciation down to 6% and the chart redraws in real time. They see the p50 outcome drop by 28%. They understand that the projection is a model, not a prophecy — and they can stress-test it themselves.
Data requirements: ai.wealth_trajectory_paths, ai.projected_irr_p50_p20_p80, mkt.price_appreciation_yoy_pct, lease.escalation_pct, frac.vacancy_pct, macro.repo_rate.
Technical sketch: The Monte Carlo simulation runs server-side with default parameters derived from micromarket history. The assumption editor sends modified parameters to a lightweight re-simulation endpoint (1,000 paths instead of 10,000 for interactive speed). Chart re-renders client-side from the returned percentile distribution. All default assumptions are cited to their source data.
Difficulty: Hard
Phase: 1
Category B — "I Didn't Know I Needed That" Moments¶
Hidden costs, buried risks, and non-obvious insights the user would never have found on their own.
B1. The True Cost Calculator¶
Product: Broker, Fractional
The moment: A user asks "What does it really cost to buy this 2BHK listed at Rs 85L?" Broker returns a full waterfall: stamp duty (Rs 5.1L at 6%), registration (Rs 25,500), GST if under-construction (Rs 4.25L at 5%), legal fees (Rs 15,000-25,000 typical), society transfer (Rs 25,000-50,000 if resale), MSEDCL deposit (Rs 8,000-12,000), maintenance advance (Rs 1.2L typical for this complex), and parking (Rs 3L if not included). Total true cost: Rs 99.3L — 17% above the "listed price." The user's jaw drops; a broker would have revealed these one at a time across months.
Data requirements: fin.sale_price, fin.stamp_duty_paid, fin.registration_fee_paid, ai.hidden_costs_breakdown; persona.residency_status (NRI stamp duty differs); loc.district (stamp duty rates vary); project-specific maintenance and deposit data from MahaRERA filings.
Technical sketch: Rule-based calculator with district-specific stamp duty tables, GST rules (under-construction vs. OC-received), and society/maintenance defaults calibrated per project type. Personalised to user's residency status and state. Outputs a structured waterfall with each line item sourced.
Difficulty: Medium
Phase: 1
B2. Carpet Area Reality Check¶
Product: Broker
The moment: A user pastes in a brochure claiming "1,180 sqft spacious 2BHK." Broker responds: "The brochure quotes super built-up area. Per the MahaRERA filing (Form B, schedule of apartments), carpet area is 718 sqft — that's 61% of the advertised number. At Rs 8,400/sqft carpet, the effective price is Rs 60.3L, not the Rs 49.6L implied by the brochure's per-sqft claim on super built-up. Here's how these area definitions work..." The user realises the brochure was designed to mislead.
Data requirements: area.carpet_sqft, area.built_up_sqft, area.super_built_up_sqft, fin.price_per_sqft_carpet, proj.maharera_number (to pull the schedule).
Technical sketch: Entity resolution matches the brochure project to MahaRERA data. The system compares user-supplied area figures against RERA-filed carpet area and flags discrepancies. LLM generates a plain-English explanation of area definitions with the specific numbers. All figures cited to MahaRERA schedule of apartments.
Difficulty: Easy
Phase: 1
B3. Title Chain Red Flag¶
Product: Broker, Analytix
The moment: A user asks Broker to check a property in Panvel. The response includes: "Title chain has a 14-year gap between 1998 and 2012 — no recorded transfers in that period. This is unusual for urban land and may indicate unregistered transactions, disputes, or adverse possession claims. The current chain has 6 recorded owners since 1985. Recommend: independent title search before proceeding." The user had no idea this was even something to look for.
Data requirements: legal.chain_of_title, legal.title_clarity_score, legal.encumbrances, legal.litigation_records, legal.land_class.
Technical sketch: The title chain is extracted from TSR documents and 7/12 extracts during pipeline ingestion. Gap detection is a rule: any period > 5 years without a recorded transfer in urban land triggers a flag. The title_clarity_score formula penalises chain gaps proportionally. LLM wraps the flag in context, always recommending independent verification.
Difficulty: Medium
Phase: 1
B4. The Stamp Duty Deadline Effect¶
Product: Analytix, Broker
The moment: An Analytix user studying Pune residential velocity notices a massive spike every March and September. The system annotates: "Transaction volume inflated 40-60% in weeks preceding stamp duty concession deadlines. The median price per sqft in these windows is 3-5% higher than the subsequent month. Consider adjusting seasonal comparisons." No other analytics product flags its own data's seasonal bias.
Data requirements: mkt.transaction_velocity_90d, fin.sale_date, fin.sale_price, mkt.median_price_per_sqft_90d; government stamp duty concession GR dates from policy.applicable_grs.
Technical sketch: Time-series anomaly detection calibrated against known policy dates (stamp duty concession GRs). The system stores a calendar of regulatory deadlines and automatically annotates velocity and price charts with event markers. Seasonality-adjusted metrics are computed as a variant of the raw velocity.
Difficulty: Medium
Phase: 2
B5. Escrow Health Indicator¶
Product: Broker, Fractional
The moment: A user evaluating an under-construction project sees: "Escrow account health: cautionary. 67% of units sold but only Rs 12 Cr of an estimated Rs 38 Cr project cost visible in escrow filings. This is below the 70% RERA-mandated escrow ratio. This may indicate delayed collections or fund diversion — verify with the developer's latest audited accounts." The user didn't even know RERA required escrow accounts.
Data requirements: fin.escrow_account_details, fin.escrow_balance_proxy, fin.estimated_cost_revised, unit.sold_count, unit.total_count, proj.estimated_cost_history.
Technical sketch: Proxy escrow balance is derived from sold units, average sale price, and the 70% rule. Comparison against declared project cost flags deviations. Confidence is marked as "approximate" since escrow balances are not publicly audited in real time. Narrative generated by LLM with explicit caveats.
Difficulty: Hard
Phase: 2
Category C — "Nobody Else Does This" Moments¶
Unique data fusion, unique analysis, or capabilities that simply do not exist elsewhere.
C1. Government Resolution Impact Feed¶
Product: Analytix, Broker
The moment: An Analytix user tracking a portfolio of 8 projects in MMR sees a notification: "New GR from Urban Development Department (dated 3 May 2026) increases permissible FSI for transit-oriented development zones within 500m of metro stations. 3 of your 8 projects fall within affected zones. Estimated impact: 12-18% increase in developable area." Each claim links to the actual GR PDF. No competitor even reads GRs systematically; PropPie classifies them by geography, impact direction, and affected projects — automatically.
Data requirements: policy.applicable_grs, policy.tailwind_flags, policy.headwind_flags, loc.lat, loc.lng, infra.metro_proximity_status, area.permissible_fsi.
Technical sketch: GR portal scraper feeds PDFs into an NLP pipeline: entity extraction (geography, policy type, effective date), impact classification (tailwind/headwind, magnitude), and geo-matching to micromarkets and projects. Results stored as structured events with source links. Push notifications for portfolio-matched GRs via Analytix dashboard or Broker alerts.
Difficulty: Hard
Phase: 1
C2. Cross-Source Contradiction Detector¶
Product: Broker, Analytix
The moment: A user asks about a specific project and Broker responds: "Note: the developer's website claims 85% sold. MahaRERA's latest quarterly filing (Q4 2025) reports 62% sold. The IGR registration count for this project over the same period shows 58 registered deeds against 140 total units (41% registered). These three numbers tell different stories. The RERA filing is the regulatory disclosure; the IGR count is the ground truth of completed registrations." The user sees contradiction-detection as a feature, not a bug.
Data requirements: unit.booked_count, unit.sold_count, unit.total_count (from MahaRERA); fin.sale_price, fin.sale_date (from IGR, matched to project); developer website claims (from web scraping or user-submitted).
Technical sketch: The pipeline resolves the same project across MahaRERA and IGR using proj.maharera_number and address matching. A reconciliation layer compares MahaRERA "booked" counts against IGR registered deed counts. Discrepancies exceeding a configurable threshold (default 15%) trigger a structured contradiction flag with all three figures cited.
Difficulty: Hard
Phase: 2
C3. Micromarket Lifecycle Map¶
Product: Analytix
The moment: An Analytix user opens the micromarket map of Pune. Each micromarket is colour-coded: green (Emerging), blue (Growing), amber (Mature), red (Cooling). They click Hinjewadi — Growing. Wagholi — Emerging. Koregaon Park — Mature. Each label links to the underlying time-series: velocity trend, price slope, new launch rate, sentiment. They can see, at a glance, where to look for growth and where to look for yield. No other Indian proptech classifies micromarkets by lifecycle stage using registered transaction data.
Data requirements: mkt.micromarket_lifecycle_stage, mkt.transaction_velocity_90d, mkt.price_appreciation_yoy_pct, mkt.sector_momentum_pct, mkt.absorption_rate, ai.sentiment_score, loc.micromarket_id.
Technical sketch: K-means clustering (k=4) on 24-month time-series features per derived-attributes-spec.md. Updated monthly. Rendered as a choropleth on a Mapbox/Google Maps layer with click-through to the feature dashboard per micromarket. Cluster labels anchored via manual review of 30 known-state micromarkets.
Difficulty: Medium
Phase: 1
C4. ASR Gap Scanner¶
Product: Broker, Analytix
The moment: A user evaluating a purchase in Bandra sees: "This property's sale price is 110% above the Annual Statement of Rates value. In this micromarket, the median ASR gap is 60-80%. A gap above 100% is in the top 15% — not necessarily overpriced, but worth benchmarking against comparables." Another user looking at a distressed sale in Karjat sees: "Sale price is 12% below ASR. This is unusual and may indicate a distress sale, a cash component, or a data-entry discrepancy. Only 3% of transactions in this area register below ASR." Nobody else systematically compares transaction prices against the government's own valuation.
Data requirements: fin.asr_gap_pct, fin.sale_price, fin.asr_value_per_sqft, area.carpet_sqft, loc.micromarket_id.
Technical sketch: ASR values are scraped from the IGR/ASR portal annually and mapped to survey numbers. Per-transaction gap is computed at ingestion. Micromarket-level gap distributions enable percentile framing. Outlier detection (>2 sigma from micromarket median) triggers narrative flags.
Difficulty: Medium
Phase: 1
C5. Delay Forensics Dashboard¶
Product: Analytix
The moment: A developer opens their portfolio view and sees: "Your 14 projects average 7.2-month delivery delay. Median peer delay for comparable Pune promoters with 10+ projects: 4.1 months. Worst offender: Project X (22 months, driven by 3 BPA revisions). Top peer practice correlating with shorter delays: pre-vetted building plans via approved consultants." The developer has never seen their own delay data benchmarked against peers with this specificity.
Data requirements: proj.promised_completion_dates, proj.actual_completion_date, dev.trust_score (delay sub-component), party.promoter_pan (for cross-project linkage), appr.building_plan_approval.
Technical sketch: Delivery delay is computed per project as the difference between the original promised completion date and actual completion (or latest revised date for ongoing projects). Developer-level aggregation uses PAN-based entity resolution. Peer benchmarking draws from the same city-tier and project-count cohort. Root-cause analysis correlates delay magnitude with approval timeline patterns.
Difficulty: Medium
Phase: 1
Category D — "This Feels Personal" Moments¶
Persona-aware, context-aware responses that make the user feel the system understands their specific situation — without crossing into advice.
D1. NRI Due Diligence Pack¶
Product: Broker
The moment: An NRI user in Singapore mentions they're evaluating a flat in Hinjewadi for retirement. Broker detects their persona (NRI, retirement goal, long horizon) and generates a structured due diligence pack: FEMA repatriation rules for this property type, NRI-specific stamp duty (if different), power-of-attorney requirements for remote registration, TDS implications (20% for NRI vs. 1% for resident), rental income taxation, and a list of what they can verify remotely vs. what requires a local representative. Every item is cited to the relevant FEMA circular or Income Tax section. The NRI realises they can do 80% of due diligence without flying to India.
Data requirements: persona.residency_status, persona.investment_goal, persona.horizon_years, loc.district, fin.sale_price, proj.type; FEMA/IT Act rule tables (static reference data).
Technical sketch: Persona detection triggers a template-based pack generator with NRI-specific modules. Each module is a parameterised rule (e.g., TDS rate lookup by residency status and property value). LLM stitches the modules into a coherent narrative with citations. The pack is exportable as PDF.
Difficulty: Medium
Phase: 2
D2. Portfolio Concentration Warning¶
Product: Fractional, Broker
The moment: A Fractional investor about to subscribe to a new warehousing asset sees a note: "If you invest in this asset, your PropPie portfolio will be 55% warehousing in MMR North. Your current exposure: 2 office assets (Pune), 1 warehouse (Bhiwandi). This is information, not advice — but here's how the sector breakdown compares to typical diversified commercial portfolios." The user has never seen any fractional platform contextualise a new investment against their existing holdings.
Data requirements: persona.existing_exposure, persona.budget_range, frac.spv_id (user's holdings), proj.type, loc.micromarket_id, sector-level portfolio weights (computed from user's holding history).
Technical sketch: User's fractional holdings are aggregated by sector, geography, and developer. New asset subscription triggers a real-time portfolio composition check. Concentration thresholds (e.g., >50% single sector, >60% single geography) generate informational flags. No recommendation; only factual portfolio composition shown with comparison to diversified benchmarks.
Difficulty: Easy
Phase: 1
D3. Tax-Bracket-Aware Yield Framing¶
Product: Broker, Fractional
The moment: A user in the 30% tax bracket asks about yields on a commercial property. Broker responds: "Gross rental yield: 7.8%. After 30% income tax on rental income (your stated bracket), maintenance, and property tax, the effective net yield is approximately 4.9%. For comparison: a 5-year FD at 7.1% pre-tax yields 5.0% post-tax in your bracket. The property's total return includes potential appreciation, which the FD does not — here's the probabilistic model for both over 5 years." The user has never seen rental yield compared to alternatives on a post-tax, apples-to-apples basis.
Data requirements: mkt.yield_benchmark, mkt.cap_rate_median, persona.tax_bracket, macro.home_loan_rate_median, ai.projected_irr_p50_p20_p80, ai.wealth_trajectory_paths.
Technical sketch: Post-tax yield calculator applies the user's declared tax bracket to gross rental income (Section 24 deduction applied where applicable). Alternative asset returns (FD, debt MF, REIT) computed at the same post-tax basis. Monte Carlo wealth path comparison across asset classes with user-specific tax treatment. All tax rules cited to relevant Income Tax Act sections.
Difficulty: Medium
Phase: 2
D4. Family Stage Context¶
Product: Broker
The moment: A user who indicated "young family" persona asks about a 2BHK in Wakad. Broker's comparison includes: "Within 3km: 4 CBSE schools (2 rated above 4.0 on Google), 2 hospitals with paediatric departments, 1 park larger than 2 acres. Commute to Hinjewadi Phase 1 (major IT hub): 18 minutes by car, 25 minutes by PMPML bus." An empty-nester asking about the same property gets different contextual information: maintenance-friendliness, proximity to healthcare, and rental yield if they're buying for income rather than occupancy.
Data requirements: persona.family_stage, infra.school_hospital_proximity, loc.commute_times, loc.infra_proximity_score, unit.type_breakdown, unit.amenities.
Technical sketch: Persona-tagged query routing: the LLM prompt includes the user's family stage, which activates different contextual modules. "Young family" activates school, paediatric care, and park enrichment via Google Places API. "Empty nester" activates healthcare, low-maintenance features, and rental yield. Contextual data is pre-enriched at the locality level and cached.
Difficulty: Medium
Phase: 2
D5. "Explain Like I'm 5" Toggle¶
Product: Broker, Fractional
The moment: A first-time buyer sees "FSI consumed: 2.8 against permissible 3.0" and has no idea what it means. They tap a toggle labelled "Explain this." Broker responds: "Floor Space Index (FSI) is the ratio of total built-up area to plot area. Think of it as how many floors of the plot's footprint the builder is allowed to construct. This project has used 93% of its allowed building volume — meaning almost no scope for additional construction on this plot. This is normal for completed projects." Every metric in the system has this toggle. The language adapts to the user's demonstrated expertise level over the conversation.
Data requirements: area.fsi_consumed, area.permissible_fsi; glossary definitions from ../90-memory/glossary.md; user interaction history (session-level expertise tracking).
Technical sketch: Each attribute ID maps to a glossary entry with three tiers of explanation (technical, plain, and simple). The ELI5 toggle sends the attribute ID and its current value to an LLM prompt with the "simple" tier and the specific numbers. Over a session, the system tracks how often the user uses the toggle to infer expertise level and proactively adjust language complexity.
Difficulty: Easy
Phase: 1
Category E — "Real-Time Intelligence" Moments¶
Live feeds, alerts, velocity signals, and the feeling that the platform is always watching on the user's behalf.
E1. Comparable Transaction Ping¶
Product: Analytix, Broker
The moment: An Analytix user tracking Hinjewadi commercial gets a notification at 10 AM: "New registration: 2,400 sqft Grade A office in Cerebrum IT Park sold for Rs 9,200/sqft (deed registered yesterday). This is 9% above the 90-day median for this micromarket. 3 comparable transactions in the last 30 days. View deed details." They click through to the deed summary with parties, area, and price — all from IGR. The platform is a persistent watchdog on their market.
Data requirements: fin.sale_price, fin.sale_date, fin.price_per_sqft_carpet, area.carpet_sqft, loc.micromarket_id, mkt.median_price_per_sqft_90d, proj.name; user's saved micromarket watchlist.
Technical sketch: Daily IGR ingestion triggers a matching engine against user watchlists (saved micromarket + segment filters). New transactions exceeding a configurable deviation from the rolling median are flagged as notable. Push notification via Analytix dashboard, email, or WhatsApp (channel per user preference). Each ping links to the full transaction record with source.
Difficulty: Medium
Phase: 1
E2. Policy Alert with Portfolio Impact¶
Product: Analytix
The moment: A developer with 20 active projects across MMR and Pune receives an alert: "Maharashtra Revenue Department GR (11 May 2026) revises Ready Reckoner rates for FY 2026-27. Average increase: 4.2% statewide. Impact on your portfolio: 6 of your 20 projects are in zones with above-average increases (6-9%). This may affect buyer stamp duty calculations and perceived pricing competitiveness. Affected projects listed below." The developer's sales team gets the information before their competitors even know the GR was published.
Data requirements: policy.applicable_grs, fin.asr_value_per_sqft (old vs. new), loc.micromarket_id, proj.maharera_number; user's project portfolio.
Technical sketch: GR scraper detects new ASR/Ready Reckoner publications. Automated diff computes zone-by-zone rate changes. Portfolio matching maps each project to its ASR zone via survey number or coordinates. Impact summary generated per user portfolio. Alert dispatched within 4 hours of GR publication.
Difficulty: Hard
Phase: 2
E3. Velocity Heatmap with Drill-Through¶
Product: Analytix
The moment: An institutional investor opens the Analytix dashboard and sees a heatmap of Pune's micromarkets. Colour intensity represents 90-day transaction velocity. Hinjewadi is glowing hot (117 transactions, up 34%). Wagholi is warm (62 transactions, up 12%). Hadapsar is cooling (41 transactions, down 18%). They click Hinjewadi. The cell expands to show the 117 transactions — price distribution, unit-type mix, top developers by volume, price trend line. Every cell in the heatmap is a portal into the underlying registered data. No other Indian proptech renders velocity at this granularity from government records.
Data requirements: mkt.transaction_velocity_90d, mkt.median_price_per_sqft_90d, mkt.price_appreciation_yoy_pct, loc.micromarket_id, loc.lat, loc.lng, fin.sale_price, fin.sale_date.
Technical sketch: Pre-aggregated velocity and price metrics per micromarket, refreshed daily. Rendered as a GeoJSON choropleth on a Mapbox layer. Drill-through queries the canonical store filtered by micromarket and date window, returning transaction-level records. Heatmap colour scale normalised to city-level distribution to maintain visual meaning across market cycles.
Difficulty: Medium
Phase: 1
E4. Sentiment Shift Alert¶
Product: Broker, Analytix
The moment: A user who saved a specific project in their watchlist gets a notification: "Sentiment for Lodha Palava, Dombivli has shifted from +0.31 (positive) to -0.08 (neutral-negative) over the last 14 days. Primary drivers: 4 news articles about delayed Phase 3 possession, 12 negative social media mentions about waterlogging, offset by 2 positive articles about metro connectivity. Sources listed below." The user sees early warning signals that would take weeks to surface through word-of-mouth.
Data requirements: ai.sentiment_score (with 14-day rolling window), proj.name, news and social mention corpus with per-mention source_url, sentiment polarity, and topic extraction.
Technical sketch: Sentiment score is computed daily per entity using the exponentially-weighted formula in derived-attributes-spec.md. A 14-day change exceeding 0.3 absolute triggers a shift alert. The alert includes the top contributing mentions (sorted by source weight and recency), each linked to the original article or post. Topic extraction via zero-shot classification labels each mention.
Difficulty: Hard
Phase: 2
E5. Construction Progress Tracker¶
Product: Broker, Fractional
The moment: A user tracking an under-construction project sees a timeline: "MahaRERA quarterly progress reports show: foundation complete (Q2 2024), structure to 5th floor (Q4 2024), structure to 12th floor (Q2 2025), plastering in progress (Q4 2025). The project is 2 quarters behind the original schedule per Form B. The developer's other active projects show a median delay of 4 months at this stage." The user can track construction progress from their phone using only public RERA data — no site visit required.
Data requirements: proj.promised_completion_dates, proj.actual_completion_date, MahaRERA quarterly progress report data (structured extraction of stage milestones), dev.trust_score (delay component), party.promoter_pan (cross-project comparison).
Technical sketch: MahaRERA quarterly progress reports are scraped and parsed for construction stage keywords (foundation, podium, structure floors, plastering, finishing, OC). Stages are normalised to a timeline. Comparison against the declared completion date produces a delay estimate. Developer-level aggregation provides peer benchmarking context.
Difficulty: Medium
Phase: 1
Summary Table¶
| # | Name | Product | Category | Difficulty | Phase |
|---|---|---|---|---|---|
| A1 | One-Click Source Trail | All | Show Your Work | Medium | 1 |
| A2 | Confidence Thermometer | Broker, Fractional | Show Your Work | Easy | 1 |
| A3 | Show Me the RERA Page | Broker, Fractional | Show Your Work | Medium | 1 |
| A4 | Developer Score Decomposition | All | Show Your Work | Easy | 1 |
| A5 | Assumption Editor on Projections | Broker, Fractional | Show Your Work | Hard | 1 |
| B1 | True Cost Calculator | Broker, Fractional | Didn't Know I Needed | Medium | 1 |
| B2 | Carpet Area Reality Check | Broker | Didn't Know I Needed | Easy | 1 |
| B3 | Title Chain Red Flag | Broker, Analytix | Didn't Know I Needed | Medium | 1 |
| B4 | Stamp Duty Deadline Effect | Analytix, Broker | Didn't Know I Needed | Medium | 2 |
| B5 | Escrow Health Indicator | Broker, Fractional | Didn't Know I Needed | Hard | 2 |
| C1 | GR Impact Feed | Analytix, Broker | Nobody Else Does This | Hard | 1 |
| C2 | Cross-Source Contradiction Detector | Broker, Analytix | Nobody Else Does This | Hard | 2 |
| C3 | Micromarket Lifecycle Map | Analytix | Nobody Else Does This | Medium | 1 |
| C4 | ASR Gap Scanner | Broker, Analytix | Nobody Else Does This | Medium | 1 |
| C5 | Delay Forensics Dashboard | Analytix | Nobody Else Does This | Medium | 1 |
| D1 | NRI Due Diligence Pack | Broker | This Feels Personal | Medium | 2 |
| D2 | Portfolio Concentration Warning | Fractional, Broker | This Feels Personal | Easy | 1 |
| D3 | Tax-Bracket-Aware Yield Framing | Broker, Fractional | This Feels Personal | Medium | 2 |
| D4 | Family Stage Context | Broker | This Feels Personal | Medium | 2 |
| D5 | ELI5 Toggle | Broker, Fractional | This Feels Personal | Easy | 1 |
| E1 | Comparable Transaction Ping | Analytix, Broker | Real-Time Intelligence | Medium | 1 |
| E2 | Policy Alert with Portfolio Impact | Analytix | Real-Time Intelligence | Hard | 2 |
| E3 | Velocity Heatmap with Drill-Through | Analytix | Real-Time Intelligence | Medium | 1 |
| E4 | Sentiment Shift Alert | Broker, Analytix | Real-Time Intelligence | Hard | 2 |
| E5 | Construction Progress Tracker | Broker, Fractional | Real-Time Intelligence | Medium | 1 |
Phase 1 (MVP): 16 moments — the core trust layer, essential personalisation, and foundational real-time feeds.
Phase 2 (Later): 8 moments — advanced data fusion, NRI-specific flows, and sophisticated alerting.
See also:
- ../20-data/data-attributes.md — full attribute catalogue
- ../20-data/derived-attributes-spec.md — math for derived attributes
- ../00-soul/SOUL.md — the honest broker posture that governs every moment above