How People Use Claude in 2026: Usage Data, S-1 Clues and India

What do people actually use Claude for?

Cadences was published on 26 June 2026 and remains the latest edition of the Economic Index on the date of writing. Its chat and Cowork sample runs from 10 April to 10 June 2026. Anthropic's classifier found that 93% of conversations produce an artifact. This is the vendor describing its own product with its own classifier, so read it as a good map and an unaudited one. The practical point is the shape of the list. If you buy Claude only for developers, you are paying for the slice that represents about a sixth of what people do with it. Plan a rollout around written output first.

• •Work conversations lean harder towards paperwork: documents and reports 20%, explanations 9%, email drafts 7%.
• •So the most common work deliverable is a document, roughly three times as common as an email draft.

When do people use Claude, and why should an employer care?

Personal use rises from around 35% of chat and Cowork conversations on weekdays to just under 50% on weekends. Recipe requests are 2.3 times more frequent at 6 p.m. than the average, and on 14 April tax conversations were eight times as common as on an average day in May. The same people use the same tool for work and home. If you pay for Team or Enterprise seats, decide now whether personal use is allowed on them and write that down, because your staff have already formed a habit either way.

How does delegation differ between chat and Claude Code?

Cadences measures autonomy on a 1 to 5 scale, and Claude Code averages 0.37 points above chat and Cowork across all conversations, widening to 0.53 for scripts and code snippets. The sharper example is a blog post. The median chat or Cowork conversation producing one takes 13 rounds of back-and-forth, while the median Claude Code session producing one contains a single human prompt. The lesson travels well beyond code: a complete written brief replaces a long negotiation. Anthropic also surveyed about 9,700 linked respondents, drawn from Claude users and explicitly not representative of the general population.

• •Reported gains: speed 86%, scope 82%, quality 69%, savings on services they would otherwise buy 27%.
• •Our reading: cost savings is the weakest number on that list. Build your internal case on turnaround time and the range of work a small team can take on, and treat payroll savings as unproven.

How much of Claude's use comes from India?

Anthropic's India brief, published 16 February 2026 and based on conversations from 13 to 20 November 2025, says India accounts for 5.8% of total Claude.ai use, second only to the United States. That figure is a year old and we found no fresher India share. Per person the ranking is very different: 101st of 116 countries, adjusted for working-age population. Business Today, reporting February 2026 data in March, gave India 98th of 116 with a usage index of 0.28x against a global baseline of 1x. We could not find that later figure on an Anthropic page, so treat it as one outlet's reading of a newer edition. Cadences itself does not mention India.

• •Business Today's reading of February 2026 data has website and web app development as India's most common use, at 6.7% of usage.
• •The same report shows India over-indexing on web development (1.9x), resume and job application help (1.8x) and programming (1.6x).
• •Our reading: Indian developers are already heavy users, and the low per-person rank suggests the gap sits with everyone else in the company. Operations, sales and finance staff are where a small firm has the most room to gain.

What does the IPO filing reveal about how Claude is paid for?

Anthropic's S-1 is public, but we could not open it on sec.gov, so these figures are second-hand and come from INDmoney's reading of the filing and its first amendment. Nothing here is investment advice. For the nine months to 30 September 2026, consumption revenue was $26.306 billion and subscriptions $6.936 billion, which is about 79.1% and 20.9% of $33.242 billion. More than 11,900 organisations paid at least $100,000 over the trailing twelve months, about Rs 94 lakh each at the RBI reference rate of Rs 93.8579 per dollar on 9 November 2026, which we use throughout. That group produced $26.1 billion, or 74% of trailing-year revenue. Dividing one by the other gives an average of at most about $2.2 million per customer, roughly Rs 20.6 crore.

• •No public user count exists. None of the readings of the filing we opened reports monthly active users or subscribers, and Anthropic publishes neither. Third-party estimates vary too widely for us to repeat them as fact.
• •None of those readings gave an India revenue figure either, so 5.8% of Claude.ai use cannot be translated into rupees of revenue.
• •The decision this informs: Claude's money is made on metered work, so your bill will follow behaviour. Long sessions and many agents cost more, and a flat seat price tells you little about API or usage-credit spend. Our guide at /blog/claude-pro-max-ultra-usage-limits-november-2026-india covers the plan side.

Who in India has rolled Claude out at scale?

The named deployments are large and mostly belong to IT services firms. They are announced rollout sizes, and no company has published how many of those people use Claude each week. Read them as evidence of intent.

• •TCS "is rolling out Claude to 50,000 associates", and Infosys has a dedicated Anthropic Centre of Excellence, according to CRN Asia.
• •Cognizant "is deploying Claude to 350,000 employees globally", per Anthropic's February 2026 Bengaluru announcement. That number is worldwide and is not an India figure.
• •CRED reported 2x faster feature delivery and 10% better test coverage with Claude Code, and Air India uses Claude Code. Both appear in Anthropic's own post, so treat them as customer claims.
• •CRN Asia reports Claude Opus 5, Sonnet 5 and Haiku 4.5 on Amazon Bedrock's India endpoint, with processing kept in the Mumbai and Hyderabad regions. Reliance and CRED were named preview customers. If customer data is involved, read /blog/fable-5-2-30-day-retention-inference-hooks-dpdp-claude-code-security before you choose a route.

What are the newest ways people can use Claude?

Two launches in the past week widen what the next Economic Index will have to count. Anthropic has published no adoption numbers for either.

• •Personal Finance (beta, 5 November) is on Pro, Max and Ultra in the US only, so Indian users cannot use it. Accounts connect through Plaid, up to two years of transactions are visible, and Anthropic says access is read-only and Claude "doesn't ask for or see your bank password".
• •Claude Whiteboards (beta) is a shared board for brainstorming with Claude. Claude Sheets (beta) builds and edits spreadsheets and exports to .xlsx or CSV.
• •Claude Data, the renamed Claude Dashboards, is generally available and needs a data connector such as a warehouse or CRM.
• •All three work tools arrived on 10 November on all paid plans. Team plans have them on by default. On Enterprise, Whiteboards and Sheets are off by default and Claude Data turns on by default on 19 November 2026, so admins should check Organization settings before then.

What happens to a company that uses Claude most heavily?

Anthropic published data on its own staff on 16 October 2026, and it is the closest thing available to a view of the far end of adoption. In March 2026 the combined hours of autonomous and interactive agents passed human work hours for the first time, and by August interactive agent hours outnumbered employees six to one. The authors warn that many of their employment practices cannot be replicated across the economy, that tool adoption is not random, and that more code does not necessarily mean better code. They call the link between junior hiring and Claude use descriptive, although around a third of machine learning managers and a fifth of software managers said they were hiring less because of AI.

• •The bottleneck moved to people. Sessions sit waiting for human input around 80% of the time, and median meeting time is up 20% on a year ago.
• •Non-technical employees went from 2 coding-agent sessions a month in August 2025 to 90 in May 2026.
• •Outside Anthropic, a Founder Reports survey of 2,052 employed US adults using AI weekly measured the oversight burden known as botsitting. Among the 1,856 who had trained an employee, 60% said it takes at least as much effort as training a new hire, and 31.6% of all respondents spend three or more hours a week on it. The survey is self-reported and the publisher ran it itself.
• •In an Anthropic trial with 50 analysts using Claude Opus 4.8 in Claude Code, analysts matched the agent's lean 86% of the time against about 50% by chance, and only 15% credited the agent. The authors deliberately strengthened the agents' leans and studied one task in one field.

Which conclusions should you not draw from this data?

Each of these errors is easy to make from the headline numbers and leads to a poor buying decision.

What should a small Indian team copy?

Here is what we would do with a team of five to twenty people. Tech Arion builds with Claude Code and advises firms adopting it. This list is our judgement and has not been measured.

Frequently asked questions about how people use Claude

How Tech Arion can help

Tech Arion is a Hyderabad software and AI consultancy that builds with Claude Code. Our AI consulting team can help you choose the first two workflows, set a spend baseline and agree a review routine.