01 / Activity
AI tool activity signal
Get a clear view of all AI tools being used across your organisation and individual subscriptions that may create added cost and risk.
read the signal
Your people are already using AI.
The question is who is governing it.
Get clear signals on which AI tools are in use across your organisation, what they’re costing, and how they’re impacting your work — so you can make decisions based on what is actually happening, not what you assume.
Do you have a GitHub Handle?
Enter it to personalise your report, or skip if you don't have one. We use public GitHub signals only, and only if you offer a handle. No private repository access required.
AI usage observability
Your organisation's AI signals
Unmanaged AI Spend
$24.1K/ month
↑ 15% vs last month
GitHub Copilot Fit Signal
78%governed AI
Strong fit
PR Throughput Opportunity
+245%
vs your current baseline
Tools In Use
6detected
4 outside procurement
Current AI spend vs. governed baseline
Current
Optimised
copilot · suggestion
1function auditAIUsage(org: Organisation) {
2 const tools = detectToolFootprint(org);
3 const spend = estimateUnmanagedSpend(org);
4 const stage = classifyAdoption(org); // agent-first
5 return arrow.signals({ tools, spend, stage, fit: 0.78 });
6}Tabaccept suggestion6 tools found · $18.7K / mo unmanaged
Before you start
The four stages of GitHub Copilot adoption.
Your report benchmarks your team against these four stages, using GitHub's own Copilot usage data across roughly 10,600 licensed developers (14 Jul to 10 Aug 2026).
PASSIVE
STAGE 0Licensed for GitHub Copilot, but not yet part of the regular development workflow.
CODE - FIRST
STAGE 1Licensed and using code completions or inline chat inside the IDE.
AGENT - FIRST
STAGE 2Working with one coding agent, such as the CLI, the coding agent, or code review.
MULTI - AGENT
STAGE 3Comfortable running two or more agents as part of everyday workflows.
Three signals, one minute
The signals every organization should see before AI becomes a board question.
02 / Cost
AI Spend Signal
Uncover token burn, inefficient workflows, and AI spend sitting outside your official budget. See the number your leadership team does not have visibility on.
read the signal
03 / Quality
Output Quality Signal
See where AI‑generated work from different tools is producing inconsistent patterns, increasing review burden, and creating long‑term maintenance exposure across your organisation.
read the signal
Why Organisations Choose Arrow + GitHub Copilot
Full visibility
See every AI tool running across your organisation, not only the ones you approved.
Spend control
Close the gap between your approved AI budget and what your people are actually spending.
Code quality
GitHub Code Quality surfaces checks and security alerts directly in the PR flow, keeping standards high without slowing teams down.
No lock‑in
GitHub Copilot works with Claude, ChatGPT, Gemini and more. Your teams keep the models they prefer.
How it works
From company profile to insight in sixty seconds.
Three steps, about a minute. Nothing is installed and nothing is scanned — the report is built from what you tell us, plus public GitHub signals if you choose to share a handle.
step 01input
Tell us who you are
Your work email and company. A GitHub handle is optional, and sharpens the signal if you give one.
> you@company.com ✓ company matched
step 02analyse
Answer three questions
Team size, the AI tools already in use, and the biggest pressure you are under right now.
team: 51-200 devs → team context added
step 03output
Read your signals
Spend as a share of payroll, how your PR throughput compares to GitHub's adoption benchmarks, and where GitHub Copilot fits.
signals.json spend_share: 4.2% copilot_fit: 0.78
FAQ
Answers for anyone responsible for what AI touches.
Everything leaders, managers, and operational owners ask before running their first signal report.
No. Nothing is installed and nothing is scanned. We work from your company profile, the three questions you answer, and public GitHub signals only if you choose to share a handle.
Under a minute. Three questions and an optional GitHub handle. There is no long form, nothing to install and nothing to download.
They are signals, not an audit. The report is built from what you tell us, your company profile and public GitHub signals if you share a handle, benchmarked against GitHub's own Copilot usage data across roughly 10,600 licensed developers (14 Jul to 10 Aug 2026). Read them as a direction to investigate, not as an invoice.
Four. Passive: licensed for GitHub Copilot, but not yet part of the regular development workflow. Code-first: using code completions or inline chat inside the IDE. Agent-first: working with one coding agent. Multi-agent: running two or more agents as part of everyday workflows. Your report places you against those four stages.
GitHub Copilot works with Claude, ChatGPT, Gemini and more, inside the environment you already govern, so your teams keep the models they prefer. Arrow turns what is running, what it costs and how it is affecting the work into one signal report you can act on.
Find out what AI is really doing in your organisation.
Get your AI signals in under a minute. No long form. No download. Just a quick AI usage and cost signal. We use public GitHub signals only, and only if you offer a handle. No private repository access required.