Chapter 02

The audit process: we learn your operation

3.5 weeks. The CTO sponsors the work, then platform, security, DX, infrastructure, product engineering, support engineering, and enablement owners bring in managers and ICs from the teams doing the work.

Audit Inputs
Interviews
60 sessions / 49 stakeholders

We interview leadership, process owners, managers, senior ICs, new hires, platform teams, and reviewers across representative teams.

  • CE
    CTO/VP Eng
    1x60min · strategic priorities
  • EM
    Engineering managers and platform owners
    10x45min · team friction + process
  • SI
    Senior ICs and reviewers
    14x30min · daily workflow + code review
  • DI
    DX, infra, security, and IT
    7x45min · tooling, access, and governance
  • NH
    New hires and enablement owners
    4x30min · onboarding and adoption blockers
  • SE
    Staff engineers and tech leads
    8x45min · architecture review + standards
  • PA
    Platform and DevEx engineers
    6x30min · CI, repo context, and tooling gaps
  • OO
    On-call owners and SREs
    4x45min · incident flow + runbook quality
  • RH
    Recent hires and mentors
    6x30min · first PR path + tribal knowledge
All sessions transcribed and tagged
Documentation
116 sources ingested

Runbooks, engineering playbooks, incident reviews, onboarding docs, AI policy, PR samples, architecture docs, and tool telemetry are indexed.

  • Tooling inventory
    Current state
  • Engineering playbook / on-call runbook
    Runbook
  • Code review checklist
    Quality policy
  • Onboarding curriculum
    Training
  • AI usage policy
    Governance
  • Incident postmortems
    Operational data
  • PR samples + review comments
    Workflow data
  • Architecture decision records
    Reference
  • CI failure history
    Operational data
Indexed and queryable by agents
Software
12 systems mapped

Source, tickets, docs, observability, CI, incident, security, feature-flag, and AI tooling are mapped to daily developer workflows.

GI
GitHub
Source
LI
Linear
Tickets
Slack
Communication
NO
Notion
Docs
DA
Datadog
Observability
PA
PagerDuty
Incident
CU
Cursor
IDE
CC
Claude Code
AI coding agent
BU
Buildkite
CI
SN
Snyk
Security
JI
Jira
Ticketing
LA
LaunchDarkly
Feature flags
All systems connected
240
Engineers represented
24
Workflows analyzed
22%
AI tool adoption baseline
380K
Hours of friction / yr
Automation map
Of 380,000 manual hours / yr
How much we take off your plate
Total identified manual hours, split between what we automate and what stays human
74%
Automatable · 280,440 hrs
99,560 hrs human
Of the 280,440 automatable hours, here’s what each agent contributes
Each bar floats at the running total. Built in priority order, summing to total automated.
0K70K140K210K280K+75KP1AI IDE$9.2M+55KP2Code Search$4.8M+74KP3Code Review$4.2M+36KP4Onboarding$3.6M+40KP5MCP$2.2M280Khrs / yrTOTALAutomated$46M / yr
P1AI IDE Standardization
90% automatable · $9.2M
P2Codebase Search (Greptile)
80% automatable · $4.8M
P3AI Code Review
70% automatable · $4.2M
P4Onboarding Curriculum
60% automatable · $3.6M
P5Internal MCP Servers
65% automatable · $2.2M
Stays human · architecture judgment, mentorship, code ownership99,560 hrs / yr remain manual
Implementation timeline19.5 weeks · then ongoing
Phase / Workflow
W1
W2
W3
W4
W5
W6
W7
W8
W9
W10
W11
W12
W13
W14
W15
W16
W17
W18
W19
Audit
3.5 weeks
CTO kickoff
Team interviews
IC shadowing
Tool map
Adoption model
Rollout sign-off
Build · workflows shipped in priority order
P1AI IDE Standardization
8w
P2Codebase Search (Greptile)
9w
P3AI Code Review
10w
P4Onboarding Curriculum
8w
P5Internal MCP Servers
6w
Operate
ongoing →
Next: Build Process