ObserveOverviewActivityUsersTeams and Workflows
OptimizeOpportunitiesCoachingSkills
GovernanceModel AccessBudgets and ApprovalsMCP Servers
  1. Users
  2. Jules Huang
J

Jules Huang

jules.huang@northline.aiSenior Software Engineer · Engineering
Tools used
ClaudePerplexity
AI spend (3 mo)$3,746.72
Savings$1,627.50

Jules's Token Efficiency Over Time

Score out of 100 · cooled off after a peak

Jules's Monthly AI Spend

Spend up 57% as usage expanded

What improved Jules's usage

  • Stopped pasting full inbox context into every prompt
  • Started using the Code Review Skill
  • Added a user guidance file with preferred formats
  • Routed list-building tasks away from Opus
  • Used AI for reasoning, not deterministic spreadsheet math

Jules's top workflow

Test Generation

Unit test stubs, edge-case lists, and flaky test triage

J
H
I
J
$1,29024 convos

Jules vs Engineering

Efficiency vs. cost compared with teammates
Efficiency vs. Cost
EfficiencyToken Spend

Jules's user.md

user.mdVery Efficient
14 lines · under the ~40-line always-on budget
# user.md

You're an experienced Senior Software Engineer working on product features in a TypeScript monorepo. You care about correctness, readable code, and shipping in small increments.

## Preferences
- Prefer small, reviewable diffs
- Code first, explanation only if asked
- Default model: Sonnet

## Project
- Package manager: pnpm
- Test: pnpm test · Lint before commit

## Do not
- Paste the full repo by default
- Rewrite unrelated files

## Load on demand
- @review.md · @debug.md · @tests.md