ShiftCurve Lab

Active R&D

What we build. Where it stands.

Honest positioning. Everything here is categorised by actual readiness - not marketing optimism. In Production means real users and real output. In Build means active development with weekly iterations. In Research means we understand the problem deeply and have working prototypes.

In Production

Live systems with real users, real data, and real output.

AEE Report Generator

AI-powered Assessment of Environmental Effects reports for resource consent applications. Compiles intelligence from GIS databases, District Plan rules, and property data into polished council-ready documents. Not templates - each report is generated from compiled intelligence specific to the site and application.

36+
Reports delivered
4
Council GIS integrations
5 min
Automated pre-assessment
Claude AIGIS APIsDocument GenerationSupabase

Digital Presence Platform

End-to-end website, SEO, and content delivery for professional services firms. AI reinforcement loop optimises Google Ads daily - auto-pruning wasted spend, blocking bad clicks, and learning what converts. Live clients generating leads from day one.

95+
Lighthouse scores
5 days
Delivery timeline
4.19%
CTR (vs 3% avg)
Next.js 15Google Ads AIGA4Vercel

Content Intelligence Pipeline

Automated content marketing engine. Overnight scouts gather industry intel, competitor activity, and market signals. Weekly factory selects topics, generates copy, and renders branded video posts. Self-optimising feedback loop learns which content performs and adjusts strategy automatically.

100+
Intel items gathered
5
Scout sources
4/week
Video posts produced
Remotion VideoLLM StrategistMulti-Tenant

In Build

Actively under development with real users testing iteratively.

BC CRM / Workspace

Purpose-built town planning workspace. Contact management, council status tracking with RMA milestones, GIS property mapping across 5 councils with Esri satellite imagery, LINZ parcel boundaries, zone overlays, flood zones, and 3 Waters infrastructure. Fee proposals with PDF generation and cadence engine for follow-ups. Phase 3 of 5 complete - real users, real data, iterating weekly.

Phase 3 of 560%
14+
Database tables
37
NZ councils mapped
5
GIS integrations
Next.js 15Supabase ProEsriLINZPDF Generation

Compound

Multi-tenant marketing automation platform. Three coordinated AI loops handle paid acquisition, content production, and outbound engagement in one substrate - learning across tenants while keeping each brand voice isolated. Day 1 schema live, identity foundation locked, executor sandbox in design. First production tenant onboarding mid-2026.

Phase 3 of 525%
Multi-Tenant SaaSAI OrchestrationSandboxed ExecutorsSupabase

In Research

Working prototypes and deep domain understanding. Production-grade delivery is the engagement itself.

Operations Intelligence Platform

Multi-tenant operations platform replacing legacy field-services tooling. Job dispatching, on-site reporting, customer portal integration, and an AI layer that surfaces patterns across jobs, sites, and teams. Working prototype validated with first deployment partner. Production rollout scheduled mid-2026.

Multi-Tenant SaaSWorkflow AutomationField Operations

Dendra

Multi-tenant intelligence platform for expert-report writers. Email a project brief, get a pre-assessment back within an hour, confirm and receive a full report draft generated from tenant calibration plus a base corpus that compounds with every correction. Beachhead vertical is planning AEE in New Zealand. Built report-type-agnostic from day one - geotech, acoustic, and traffic are next. Working engine in production internally; productised tenant onboarding pre-launch.

Multi-Tenant SaaSEmail-Driven IntakeFederated IntelligenceDocument Generation

Why we categorise this way.

Enterprise buyers detect vapourware instantly. Calling something "In Build" when the hard problems are not solved erodes trust the moment someone asks a technical question. Calling it "In Research" with a working prototype and deep domain understanding is stronger - it says we know exactly how hard this is, and we will not pretend otherwise.

The CRM is the proof point. Phase 3 complete, real users, real data, iterating weekly. That is what delivery looks like.