Sort a big Indian group's businesses into core bets and side bets

A group that owns many businesses cannot treat them all as equally central. The real question is which businesses carry the group's identity and strengths, which ones support them, and which look like experiments, judged by a test you set yourself and apply the same way to every business.

• What is your test for core: closeness to the group's best-known strengths, how the group talks about it, or something else?
• Which eight businesses will you pick so they genuinely differ from each other, not eight versions of one thing?
• What would make a business supporting rather than core, in one sentence you can reuse for every row?
• Which two marks are you least sure about, and what fact would settle each?

A one-page table of eight or more businesses from the Tata group's public company list, each marked core, supporting or side bet with a one-line reason that uses your stated test, two rows flagged as uncertain, and one sharp question for the group's management.

Marking businesses by how famous they are to you. A business you have heard of is not core by default; your test must decide, and the reason column must show the test at work.

• The Tata group's public page that lists its companies, saved or screenshotted on the day you read it
• Each chosen company's own public 'about' page, for one line on what it does
• Your written test for core, supporting and side bet, decided before you mark anything

AI must not state sizes, profits, dates or history for any Tata company, and must not guess what management thinks. Every mark rests on your test and the group's own public pages, nothing else.

1. 3.1 Write your test before you markCLAUDE CHAT

   Why this tool: A short back-and-forth sharpens a vague idea of 'core' into a test you can apply row by row.

   Describe your first idea of what makes a business core. Let Claude question it until you can write the test in two sentences, then save those sentences at the top of your table.

   Prompt to copy · fill in the [brackets]Copy

   I am sorting a business group's companies into core, supporting and side bets. My first idea of 'core' is: [your idea in one or two lines]. Ask me four questions that expose where this test is vague or would give odd results. Then help me rewrite it as a two-sentence test I can apply the same way to every business. Do not mark any businesses yourself.

2. 3.2 Build the table from the public listGOOGLE SHEETS

   Why this tool: A simple sheet keeps the business, what it does, your mark and your reason in columns you can sort later.

   Pick eight businesses from the public list that look very different. For each, write what it does in your own words from its public page, then your mark and a one-line reason that uses your test.

3. 3.3 Get a second split and compareCLAUDE CHAT

   Why this tool: Claude's split, made blind to yours, shows where your test is clear and where it is a judgement call.

   Paste only the business names and your one-line descriptions, plus your test. Compare its marks with yours and note every disagreement in a new column.

   Prompt to copy · fill in the [brackets]Copy

   Here is my test for core, supporting and side bet: [your two-sentence test]. Here are eight businesses with one line on what each does:
   [paste names and descriptions]
   Apply my test to each and give a mark with a one-line reason that quotes the test. Where the test does not decide clearly, say 'unclear' and explain why instead of guessing. Do not add facts about the companies.

4. 3.4 Sharpen your question for managementCLAUDE CHAT

   Why this tool: Critique turns a general question into one that management could not answer with a slogan.

   Write your question for the group's management, then ask Claude to find how it could be dodged and rewrite it yourself.

   Prompt to copy · fill in the [brackets]Copy

   My question for a business group's management is: [your question]. How could a manager answer it without really answering? Suggest what would make it harder to dodge, and tell me which of my two uncertain rows, [row one] and [row two], the question would help settle.

• Every business in the table appears on the group's public company list you saved
• Each reason uses the words of your test, not a general impression
• Where you and Claude disagreed, your final mark has a reason you can defend
• No row states a size, profit or date that you did not see on a public page
• Every cell in the table has a source you opened yourself
• Like is compared with like (same period, same unit)

AI tends to call the oldest or best-known businesses core and fills gaps with facts it half-remembers about the group. Strip any fact you cannot point to on a public page.

• Your test written at the top in two sentences
• Eight or more rows with mark and one-line reason
• Two rows flagged as least sure, with what you would need to know
• One question for management that is hard to dodge

One page: your test in a box at the top, the table in the middle sorted core first, and the management question at the bottom.

Explain your test for core and one business where your mark differed from Claude's, and why you kept your own.