Anthropic just released another study, and the results are worth your attention. You can find the full research here.
In this study, Anthropic analyzed 100,000 real conversations from Claude.ai (Free, Pro & Max). The goal was to estimate how long typical knowledge-work tasks take with and without AI assistance.
It is a long read, but the conclusions are pretty remarkable.
What did Anthropic measure?
The researchers compared task duration with and without AI support. On average, the numbers look like this:
- Average task without AI: ~90 minutes
- Average task with AI: ~80% time savings
Time savings by task type
- Diagnostic-imagee checking: ~20% savings
- Compiling and summarizing reports: ~95% savings
Most tasks fall somewhere between 50–95% time reduction, with a clear peak around 80–90%.
Anthropic is transparent about the limitations of their research and the extrapolations involved. Still, their validation work is solid enough to take these findings seriously.
The nuance that really matters
One key takeaway should not be overlooked: AI only creates real value when certain conditions are met.
AI delivers productivity gains when:
- The job involves digital workflows, and
- People use AI correctly, consistently, and safely
This does not happen automatically just because an organization buys an AI tool.
It requires training, coaching, and new habits.
What does this mean for organizations?
Anthropic’s macro-model suggests that today’s AI could generate 1.8% annual labor-productivity growth across the entire economy.
At the organizational and individual level, the numbers are even more tangible.
Productivity gains by role
- Administrative & general office roles:
- 3–10% productivity gain
- ≈ 1–4 hours saved per week per employee
- AI-intensive knowledge workers
(developers, marketers, analysts, copywriters, consultants…)- 10–15%+ productivity gain
- ≈ 4–6+ hours saved per week
A properly trained administrative team can deliver 3–10% more output per year with the same headcount. Over time, that impact adds up significantly.
Bottom line
AI is no longer a “nice to have”.
It is a productivity lever — but only for teams that actually learn how to use it effectively.
How SirusAI helps organizations unlock this shift
At SirusAI, we support organizations in turning AI potential into measurable productivity gains by:
- Training teams to use AI correctly, safely, and confidently
- Identifying where the real-time savings are
- Optimizing workflows with AI and automation
- Building AI governance aligned with GDPR and the EU AI Act
We support teams end-to-end: strategy → adoption → working solutions.
If you are curious how much time your teams could save, or where to start, send me a message.
A 30-minute conversation can open your eyes to opportunities hiding in plain sight.
