You are running AI
everywhere. What is the ROI?
Agents on Microsoft. Automations on Google. Something a team built on AWS last quarter. We put all of it on one screen, so you can see what it returns.
Example organisation, illustrative figures.
Ask which of your AI is paying for itself.
Ask in plain English. The assistant answers from your connected systems and licence data, and shows where each figure came from.
Which AI products have the most idle seats?
Microsoft 365 Copilot: 168 of its 600 seats weren’t used this quarter. Next are Microsoft Foundry Agent, with 130 of 500, and the Copilot Studio HR bot, with 103 of 270. Across everything, 542 of 2,120 seats are idle.
What’s our net AI ROI this quarter?
45% in Q3. AI is creating £82,822 of value a month against £57,000 in total cost of ownership, a net gain of £25,822 a month.
How does win rate compare for reps who use AI heavily?
The six reps who use AI heavily win 28.4% of their deals, against 21.9% for the six who don’t, over the last six months. Their average deal is also larger: £43,083 against £17,960.
Show me a table of net ROI by AI product.
Amazon Bedrock (214%) and GitHub Copilot (184%) return the most. Microsoft 365 Copilot and the HR bot cost more than they return.
Chart ticket resolution time, AI vs manual.
AI-assisted tickets close faster every month. Over six months they average 41.0 hours against 51.6 for manual tickets, 20% faster.
Answers from Lumela’s demo data.
- No made-up numbers
Every figure comes from your own systems.
- Read-only
It can’t change records, settings or integrations.
- Access by team
Admins choose which systems each team can ask about.
- Full audit trail
Every question and lookup is logged.
No current system shows you everything.
Every platform sells you agents, and every team builds automations on top of them. Each one reports on itself and nothing else. We connect to all of them and give you a single view: how many are live, what they cost, and what the return on investment is.
- Agents
- 147
- Automations
- 2,310
- Monthly cost
- £41,800
Figures for an example organisation.
You are paying for 240 agent seats. Ninety-one are in use.
Seats get bought in bulk at the start of a rollout and nobody checks them again. We show you which ones are live, which ones have never been touched, and exactly what the difference is worth.
- Agent licences paid for
- 240
- Licensed across five platforms
- Used in the last 30 days
- 91
- Everything else sits idle
- Spend returning nothing
- £53,640/yr
- Renews automatically next quarter
When is AI profitable?
We add up what your AI costs to build and run, then set it against what it measurably brings back. The line crosses zero in the month it starts paying for itself.
- Set-up
- £186,000
- Implementation, consulting, training
- Running
- £41,800/mo
- Licences, infrastructure, admin
- Breaks even
- 7 months
- From go-live
- Net ROI
- 45%
- Over 36 months
About Lumela
Lumela is an AI adoption and ROI tracker. It reads the systems your organisation already uses and reports what each agent and automation costs you, and what it brings back.
The before-AI baseline comes from the same place. Email threads and support tickets carry timestamps from long before anyone switched an agent on, so we read that history instead of asking you for numbers nobody wrote down.
We’re a small team in the UK, looking for our first pilot organisations. If that could be you, email info@lumela.co.uk.