Andrew Tran

Logged from Claude Code: lead time computed across sixteen now-fields — median +4.2 months, three negative

→ Lab Effectivity · elected

'Research that ships: moving a lab from publish to push'

→ Lab outcomes visibility · elected

Replicating the experienced-developer RCT on enterprise monorepos

→ ROI · elected

SWaT-2026 — refreshed industrial telemetry anomaly set with slow-onset labels

→ Sensing Agents · elected

Week-eight retention across three assistant engagements: proactive and integrated wins 3.1×

→ Effective Assistants · elected

Three client clusters measured: 18%, 31%, 41% sustained utilisation over a quarter

→ On-Prem Inference · elected

TGA DAEN adverse-event export, linked to PBS dispensing counts

→ Drug Development

Logged from Claude Code: model-risk document drafted from the repo, accepted with nine edits

→ AI-DLCs · elected

World Models for Decision-Making: A Survey of Evaluation Practice

→ World Models

ABS Retail Trade + public scanner panel: the series behind the graveyard run

→ World Models

'We fine-tuned nightly for six months. Here is what broke.'

→ Non-Weight-Bound Continuous Learning · elected

Lab analytics production run dark since June: 3.8% exception rate, one human on exceptions

→ Dark Factory

Control Roles After Automation: Why Task Automation Rates Overstate Headcount Effects

→ Dark Factory

Delivery telemetry: tokens substitute for 20–30% of junior analyst hours on standard engagements

→ Post Economy · elected

Fusion milestone register: 9 pilot programmes, 0 net electricity, targets slipped a median 14 months in 2025

→ Fusion Impact on Pricing

AEMO ISP draft: data-centre load the largest new demand category; fusion absent from all scenarios

→ Fusion Impact on Pricing