ShiftCurve Lab

Enterprise

Investment intelligence, built to your spec.

An AI-native data layer that unifies Charles River, Bloomberg, and APEX into a single live view. Replace PM spreadsheets with real-time position management, scenario testing, trusted performance attribution, and CIO-level risk visibility. Working prototype in 8 weeks.

8 weeks

to replace one PM's spreadsheet

16 weeks

to CIO dashboard with scenario testing

24 weeks

to full team rollout

The Problem

Your PMs are still running spreadsheets.

You have Charles River, Bloomberg PORT, and APEX. Yet portfolio managers maintain shadow spreadsheets because nothing unifies position data, risk analytics, and scenario testing into a single live view. The spreadsheet is the unofficial integration layer - and it is fragile.

Spreadsheet Risk

Portfolio managers maintain shadow spreadsheets because Charles River, Bloomberg PORT, and APEX don't unify into a single working view. Key person risk, fat finger risk, no resilience.

CIO Blind Spot

The head of investments has no aggregated view across all PM books. Getting a cross-portfolio risk picture requires manually collecting and reconciling individual spreadsheets.

No Forward or Backward View

PMs can't reconstruct portfolio state at any prior date, and scenario testing (rate changes, FX moves, duration impacts) requires rebuilding the spreadsheet from scratch.

Disconnected Systems

Charles River handles orders, Bloomberg PORT does analytics, APEX delivers daily NAV - but nothing stitches them together in real-time. The PM's spreadsheet is the unofficial integration layer.

Attribution Black Hole

FactSet couldn't solve it in 10 years. Bloomberg PORT couldn't in 2. The problem isn't the math - it's that nobody trusts the data going in. Without a verified data pipeline, attribution outputs are meaningless. PMs fall back to gut feel instead of quantified performance drivers.

The Comparison

What you have now vs. what we build.

Enterprise Platform
ShiftCurve Build
Time to replace one PM's spreadsheet
Never (not their problem to solve)
8 weeks
CIO cross-portfolio view
Manual aggregation from PM spreadsheets
Live, unified, real-time
Scenario testing
Bloomberg PORT (separate tool, no PM context)
Native, natural language, integrated
Charles River integration
Already have it (data stays siloed)
REST API, intraday positions, reconciled
APEX reconciliation
Daily PDF, manual break resolution
Automated three-way, exception reporting
Backward view
Not available without manual reconstruction
Any date, from transaction history
Performance attribution
FactSet/PORT attribution exists but PMs don't trust the outputs (data quality)
Data Trust Layer validates inputs, then transaction-based Brinson-Fachler + fixed income attribution with full audit trail
AI layer
Bloomberg AI Commentary (text only, Sep 2025)
Interactive NL queries over live data
Institutional knowledge
Lives in PM heads and spreadsheets. Walks out the door when someone leaves.
Compounding knowledge wiki - every query, scenario, and reconciliation builds permanent institutional intelligence
Data security
Vendor cloud (your data leaves the building)
On-premises AI for sensitive data. ISO 27001:2022 ISMS in progress. Your portfolio data never touches external APIs.

The Difference

A platform that gets smarter the longer you use it.

Most analytics platforms are stateless. Ask a question, get an answer, start from scratch next time. We build something fundamentally different: a persistent intelligence layer that compounds knowledge with every query, every reconciliation, and every decision your team makes.

Day 1

The platform connects to your data sources. Positions, pricing, and NAV flow in. You have a dashboard that replaces the spreadsheet.

Month 3

The platform has resolved 200 reconciliation breaks and knows which data sources to trust for which asset classes. Scenario queries reference prior analyses. Attribution outputs are trusted because the data inputs are verified.

Year 1

You don't just have a dashboard. You have a queryable institutional brain containing everything your team has learned about your portfolios. New analysts inherit that intelligence on day one.

Why this matters for attribution

Attribution has failed at fund managers for over a decade. Not because the math is hard - Brinson-Fachler is well understood. It fails because nobody trusts the data going in. Our platform builds that trust incrementally. Every reconciliation break resolved, every data quality issue flagged and fixed, every source reliability pattern detected - it all compounds into a Data Trust Layer that makes attribution outputs meaningful for the first time. After six months, the platform doesn't just calculate attribution - it can explain exactly why you should trust the numbers.

Platform Modules

Seven modules. One unified platform.

Seven modules. Every one designed to work independently or as part of the complete platform. Start with what hurts most, expand from there.

01

Unified Position View

Replace the spreadsheet with a live, reconciled book

Intraday positions from Charles River IMS via REST API
Daily reconciliation anchor from APEX fund admin (T+1 SFTP)
Automated three-way reconciliation with exception reporting
PM view (my book) and CIO view (all books aggregated) in one platform
02

Scenario Testing Engine

What-if analysis without rebuilding the spreadsheet

Parameterise key risk factors: rates, credit spreads, FX, inflation, duration
Natural language: 'What happens to duration if I increase TLT allocation by 5%'
Real-time VaR, stress testing, and portfolio-level impact modelling
Bloomberg rate curves (RBNZ, Fed) integrated for scenario calibration
03

Backward & Forward View

Reconstruct any portfolio state from transaction history

TimescaleDB stores full position history from day one
Reconstruct portfolio at any prior date from transaction log
Historical attribution: what drove returns last quarter, last year
Forward projections using scenario engine outputs
04

AI Intelligence Layer

The brain Bloomberg PORT doesn't have

Natural language queries over live portfolio data
Anomaly detection: flag unusual position changes or correlation shifts
Automated insight generation: concentration alerts, rebalance triggers
Reconciliation break analysis with suggested resolutions
05

Integration Architecture

Connects Charles River, Bloomberg, and APEX natively

Charles River REST API for intraday order and position data
Bloomberg BLPAPI/B-PIPE for market data, pricing, rate curves
APEX daily files via SFTP for NAV, cash, and position verification
Role-based access (PM sees own book + aggregates, CIO sees everything)
06

Performance Attribution Engine

The module FactSet and PORT couldn't deliver

Brinson-Fachler decomposition for equity portfolios: allocation, selection, interaction effects
Duration, curve, and credit attribution for fixed income books
Currency attribution via Karnosky-Singer framework for multi-currency portfolios
Transaction-based daily TWR - the only method that produces accurate attribution
Data Trust Layer: automated three-way reconciliation (Charles River vs Bloomberg vs APEX) before any attribution runs
Natural language: 'Why did we underperform by 30bps last month?' with decomposed, auditable answers
07

Institutional Knowledge Capture

Your team's expertise, permanently captured and compounding

Every reconciliation, scenario, and PM decision builds a persistent knowledge base that gets smarter over time
Data Trust Evolution: after 50 reconciliation breaks resolved, the platform KNOWS which sources to trust for which data types
Institutional memory: when a PM leaves, their knowledge stays - every question they asked, every scenario they ran, every insight they surfaced
Cross-PM intelligence: PM #2's questions benefit from every question PM #1 ever asked - no siloed knowledge
Proactive intelligence: the platform suggests questions you should be asking based on patterns it has detected across the knowledge base
New analyst onboarding in days, not months - the knowledge wiki contains everything the team has learned, structured and queryable

Engagement Model

Phased delivery. Value at every stage.

No big-bang implementation. Each phase delivers a working product you can evaluate before committing to the next.

Phase 1

Replace One PM's Spreadsheet

Weeks 1-8

Connect to Charles River and Bloomberg. Build a live position view that replaces one pilot PM's spreadsheet entirely. Daily reconciliation against APEX. If it isn't faster than the spreadsheet by week 8, we haven't succeeded.

Deliverables

Live position dashboard for one PM (multi-asset)
Charles River REST API integration (intraday positions)
Bloomberg BLPAPI connection (market data, pricing)
APEX daily reconciliation (T+1 position verification)
Basic scenario testing and risk metrics
Phase 2

CIO Dashboard & Scenario Engine

Weeks 9-16

Aggregate all PM books into a unified CIO risk view. Full scenario testing engine with parameterised risk factors. Natural language interface. Backward view from transaction history.

Deliverables

CIO aggregated view across all PM books
Scenario testing: rates, credit spreads, FX, inflation, duration
Natural language queries over portfolio data
Historical reconstruction (any prior date)
Automated daily reconciliation with exception reporting
Performance attribution prototype: Brinson-Fachler for pilot PM's equity book
Phase 3

Full Team Rollout & Automation

Weeks 17-24

Harden, automate, and extend to the full investment team. Integration with Charles River for order-level attribution. Role-based access, audit trails, and potential APEX live NAV integration.

Deliverables

All PMs onboarded with personalised book views
Full performance attribution engine: equity (Brinson-Fachler), fixed income (duration/credit), currency (Karnosky-Singer)
Data Trust Layer: automated pre-attribution reconciliation with exception reporting and audit trail
Role-based access (PM own book + aggregates, CIO everything)
Audit trail and version history
Potential APEX live NAV and cash integration

The Team

Lean team. Senior people. AI-multiplied.

A focused team with deep domain expertise, augmented by AI engineering tools that let a small group deliver what used to take a department. No bloat, no hand-offs, no learning on your dime.

Pranil Bilimoria — Founder & Technical Director

Seasoned delivery leader with over two decades in financial services, leading enterprise platform initiatives across ESG, regulatory, and investment operations. Has navigated the governance, vendor lock-in, and integration failures that define enterprise IT in fund management. Implemented Charles River IMS for a major NZ asset manager. Now building the tools that replace the spreadsheets, manual reconciliations, and broken data pipelines that the traditional approach leaves behind. Hands-on architecture, AI integration, real-time data, and full-stack delivery.

Embedded Domain Expert

Investment operations and fund management expertise from inside your organisation. Understands the workflow from trade execution through to client reporting. Bridges the gap between what portfolio managers need and what engineers build.

AI-Augmented Engineering

Senior developers working with cutting-edge AI tooling to ship at 3-4x the speed of traditional teams. Small, senior, fast. No bloated teams, no offshore hand-offs. The people who scope it are the people who build it.

Why Us

We've already built this. For ourselves.

We've implemented Charles River before

Our founder implemented Charles River IMS for a major NZ asset manager, built STP pipelines from Charles River to back-office systems, and designed margin lending platforms. We know where the data breaks happen because we've lived inside these systems.

AI-native, not AI-bolted

Bloomberg added AI Commentary to PORT in September 2025 - it writes about your portfolio. Our platform answers questions about it. Natural language scenario testing, anomaly detection, and insight generation built from the ground up, not retrofitted.

We solve the PM adoption problem

The prototype must be genuinely better than the spreadsheet by week 8, or PMs won't switch. We design for adoption first: the tool replaces the spreadsheet workflow, it doesn't add another system to check. If it isn't faster, we haven't succeeded.

NZ-built, you own the IP

No offshore development, no timezone gaps. FMA and FMCA reporting understood natively. Everything we build is yours - no licensing fees, no vendor lock-in, no annual renewals.

$60,000+ invested in AI infrastructure

We've processed over $60,000 in AI compute building and testing these systems. While others are running their first LLM experiment, we've already burned through the experimentation phase and emerged with production-grade infrastructure refined through millions of iterations. This isn't a prototype - it's battle-tested.

Data Security

Your data never leaves your building.

Fund managers handle sensitive position data, proprietary models, and investor information. We built our deployment model around that reality, not as an afterthought.

Fully Air-Gapped

Zero external connections

All AI inference runs locally on dedicated hardware deployed on your network. Open-source models process your portfolio data, position information, and proprietary analytics without a single byte leaving your infrastructure.

  • Local LLM on dedicated hardware
  • No API calls to external services
  • Fully auditable by your IT team
  • Works behind any firewall policy
Recommended

Enterprise AI Partnership

Maximum capability, zero data retention

Enterprise-grade AI with contractual zero-retention guarantees. Your data is processed but never stored, never used for training, and never accessible to anyone outside the API call. SOC 2 Type II certified infrastructure.

  • State-of-the-art reasoning capability
  • Zero data retention agreement
  • SOC 2 Type II certified
  • Data Processing Agreement included

Hybrid Deployment

Sensitive data local, everything else cloud

Portfolio data and position information processed by locally-hosted models. Code generation, documentation, and non-sensitive work uses enterprise cloud AI for maximum speed. Your compliance team defines the boundary.

  • Sensitive data stays on-premises
  • Cloud AI for speed on non-sensitive work
  • Compliance team sets the rules
  • Best balance of speed and security

ISO 27001:2022 Certification Planned

ShiftCurve is building its Information Security Management System to ISO 27001:2022, the international standard for information security. Our ISMS scope covers all enterprise engagement delivery, client data handling, and AI infrastructure operations. All 93 Annex A controls have been assessed, with a structured certification roadmap targeting formal accreditation in 2026. Our security architecture was designed for regulated financial services from day one - not retrofitted after the fact.

What goes where

Portfolio positions, NAV, holdingsOn-premises only
Investor and client dataOn-premises only
Proprietary models and parametersOn-premises only
Natural language queries about portfoliosOn-premises only
Dashboard code and UI componentsCloud AI (zero retention)
Data pipeline and ETL logicCloud AI (zero retention)
Documentation and reportsCloud AI (zero retention)

Ready to move faster than an RFP?

Start with a discovery conversation. No procurement hoops. No 50-page proposals. Just a real conversation about what you need and how fast we can build it.