Lesson 16 of 32 · Part One, the principles

The Seed of a New Idea

Brainstorming

Do not ask for the perfect name. Ask for thirty rough ones. Volume and range are what the machine is genuinely better at than you - it will lay out more options in a minute than you would reach in a week - while the judgement stays where it belongs, with the person who knows the shop, the buyers and the budget. State a number, demand variety in named directions, forbid it from choosing a winner, and then do the choosing yourself.

Reach for it whenWhen you need a name, a theme, a headline or a way out of a stuck problem, and you can feel that your own first idea is also the most obvious one.

How to use The Seed of a New Idea in Perplexity

Be honest about the mismatch, because it is the whole of this chapter. Brainstorming needs divergence - thirty rough options, including bad ones - and an answer engine is built to converge: it searches, it narrows, it hands you the consensus with sources attached. Ask it for thirty names and you get perhaps a dozen cautious ones, each half-borrowed from a page it found. So route the generation to ChatGPT, Claude or Gemini, and use this for the two steps that bracket it. Before you brainstorm: what already exists. What the collections in your trade are called, what the shops near you already sell at that hour, what somebody else tried and reported. That scan stops you spending a week on an idea that has a name and a competitor already. After you brainstorm: whether your shortlist is taken, and what it would cost to do.

  • Generate elsewhere; use this before and after the brainstorm.
  • Before: what already exists, what is named, what others reported.
  • After: is the shortlist taken, and what will it cost?

From GenAI for Everyone: Prompt Engineering Handbook, First Edition 2025, by Raja Rajeshwari Seshadri and Shreeram Iyer.

Be honest about the mismatch, because it is the whole of this chapter. Brainstorming needs divergence - thirty rough options, including bad ones - and an answer engine is built to converge: it searches, it narrows, it hands you the consensus with sources attached. Ask it for thirty names and you get perhaps a dozen cautious ones, each half-borrowed from a page it found. So route the generation to ChatGPT, Claude or Gemini, and use this for the two steps that bracket it. Before you brainstorm: what already exists. What the collections in your trade are called, what the shops near you already sell at that hour, what somebody else tried and reported. That scan stops you spending a week on an idea that has a name and a competitor already. After you brainstorm: whether your shortlist is taken, and what it would cost to do.

Worth knowingThe convergence is not a setting you can turn off. Phrase it as a brainstorm and the answer still arrives in the shape of a summary of the field, with your wildest request quietly dropped, because nothing in the index asked to be strange. Nor will it hold a quota: told to produce five unconventional ideas it produces five ordinary ones labelled unconventional. Two things it does genuinely well in this chapter: an existing-landscape scan before you start, and a reality check after. Everything in between - the thirty seeds, the odd ones, the combining of 4 with 12 - belongs in another tool.
  • Generate elsewhere; use this before and after the brainstorm.
  • Before: what already exists, what is named, what others reported.
  • After: is the shortlist taken, and what will it cost?
  • Expect convergence - a quota of odd ideas will not be met.
  • Keep the landscape scan in a Space and reuse it each season.
The Gold Brick vs. the Ore

Instead of

Give me the perfect name for our saree collection.

Perplexity version

I will generate the names in another tool - asking a search engine for thirty names produces a dozen careful ones borrowed from pages it found. What I need here is the landscape I am naming into.

I am the secretary of a handloom weavers' cooperative in Pochampally, Telangana, launching an ikat cotton saree collection online for working women aged 25 to 45 in metros.

Cite everything, and give me the date on each page.

1. What are the handloom and ikat collections currently sold online actually called? Give me twenty real collection or brand names with the seller and a link, so I can see what ground is already crowded.
2. Which words recur in that naming - the Sanskrit and Telugu words, the loom words, the English ones? Tell me which are so common that a buyer will not register another one.
3. What does the geographical indication registration for Pochampally ikat permit or require in how we name and label the product? Cite the GI registry or the relevant authority, not a news article about it.
4. Where and how is the Indian trade mark register searched, and what should a cooperative do before printing a name on labels? Link the official route.
5. What do buyers in this segment actually say they are buying - the words that recur in reviews and in press coverage of handloom buyers? Quote them.

Then a one-page brief: the crowded words, the clear ground, the GI constraint, and ten words buyers use themselves. I will take it to another tool and generate thirty names against it.
Open Perplexity 1,500 characters

Then loop it

  1. Narrow to the crowded words. Which five are so heavily used across handloom brands that I should rule them out at the start? Show me the examples, not just the conclusion.
  2. I have a shortlist now: [paste three names]. For each, where does it already appear - a label, a shop, a company, a product in another trade - and in which languages does it mean something I should know about? Cite each.
  3. Save this in a Space called "Collection naming" and tell me which three checks I should repeat for every future collection.

Why it worksThe thirty seeds are grown elsewhere; what is sourced here is the ground they will be planted in - which words are exhausted, what the GI allows, and whether the shortlist is already somebody's label.

One Tired Theme vs. a Field of Twenty

Instead of

Suggest a theme for a 60th birthday party.

Perplexity version

I plan events in Jaipur. I will brainstorm the party themes in another tool, because this one converges and I need twenty directions rather than the consensus. What I want here is what is actually possible and what it costs.

The party: 60th birthday for a retired railway officer who loves old Hindi film songs and train journeys. ₹2.5 lakh, 120 guests aged 5 to 85, garden lawn, February evening, three weeks to build.

Cite everything.

1. Sunset time and typical evening temperature in Jaipur in February, from a meteorological source. I need to know exactly when the cold begins, because it decides the shape of the evening.
2. What licence is required to play recorded Hindi film music at a private event of this size in India, who issues it, and what it broadly involves. Cite the official position - the family will ask me and I must not guess.
3. What permissions a private lawn event of 120 people typically needs in Jaipur - noise rules and their timings, any local restriction on amplified sound. Official sources.
4. Prop, signage and memorabilia hire in or near Jaipur - who exists, what they list, with links. Say so plainly if the market is not visible online.
5. Anything reported about railway-themed or vintage-film-themed events in India, with photographs or coverage I can show the family as a reference.

Then a one-page brief: the time the cold arrives, the licences, the permissions, the vendors who exist. I will take it to another tool and generate twenty themes that respect all of it.
Open Perplexity 1,512 characters

Then loop it

  1. Narrow to the music licence. Who exactly issues it for a private function, what is the published basis of the fee, and is there any exemption for a family event? Quote the official page.
  2. I have two themes shortlisted: [paste them]. For each, what does it need that I would have to hire in Jaipur, and does a supplier for it visibly exist? Cite the listings.
  3. Save this in a Space called "Jaipur events" so the permissions and licence research is not repeated for every party, and tell me which facts go stale each year.

Why it worksThe twenty themes come from elsewhere, but the cold at eight o'clock, the music licence and whether a Jaipur vendor exists are facts - and a theme that ignores them dies on the lawn.

The Stubborn Problem vs. Twenty-Five Ways Out

Instead of

How can I increase sales in my bakery?

Perplexity version

I own a small bakery in Shillong with dead weekday afternoons, 2 to 5 pm. I will generate the twenty-five ideas in another tool, since this one narrows towards what others already do. What I want from you is what others already do - which is a genuinely useful thing to know before I invent anything.

Cite everything and tell me the date on each page.

1. What has been reported about bakeries, cafes or small food shops anywhere in India solving an afternoon lull - what they tried, and whether it worked? Real cases with coverage, not suggestions.
2. What is published about the colleges and the hospital on my road - class timings, examination periods, visiting hours, hostel meal timings? Anything official that tells me when people near me are free at three in the afternoon.
3. Shillong's tourist season by month, and whether weekday afternoon footfall in the town is reported as seasonal. Cite the tourism data.
4. What the food delivery platforms charge a small bakery in a town like mine - commission structure and onboarding - from their own published pages, since I have no rider of my own.
5. Is anything published about afternoon tea or snack habits in Meghalaya or the North East that I should know before I design a 3 pm product? Say so if there is nothing - an honest gap is better than a guess about my own customers.

Then a one-page brief of what exists, what worked elsewhere, and what the platforms cost. I will brainstorm against it in another tool.
Open Perplexity 1,472 characters

Then loop it

  1. Narrow to the reported cases. Give me the three clearest examples of a small bakery changing its afternoon trade, with what they did and what the coverage says it achieved. Quote them.
  2. I have three ideas shortlisted: [paste them]. For each, is there any reported precedent, and what would it cost me to try according to published prices? Cite what you can and say where nothing exists.
  3. Save this in a Space called "Bakery afternoons" and list the two things I should re-check next season, since the college calendar and the tourist months are the parts that move.

Why it worksTwenty-five ideas need divergence and come from another tool - what this one supplies is the scan of what has already been tried and what the neighbours' timetables actually are, so the invention starts from the real street.

Questions

What is The Seed of a New Idea in prompt engineering?

Do not ask for the perfect name. Ask for thirty rough ones. Volume and range are what the machine is genuinely better at than you - it will lay out more options in a minute than you would reach in a week - while the judgement stays where it belongs, with the person who knows the shop, the buyers and the budget. State a number, demand variety in named directions, forbid it from choosing a winner, and then do the choosing yourself.

When should I use this technique?

When you need a name, a theme, a headline or a way out of a stuck problem, and you can feel that your own first idea is also the most obvious one.

How do I apply it in Perplexity?

Be honest about the mismatch, because it is the whole of this chapter. Brainstorming needs divergence - thirty rough options, including bad ones - and an answer engine is built to converge: it searches, it narrows, it hands you the consensus with sources attached. Ask it for thirty names and you get perhaps a dozen cautious ones, each half-borrowed from a page it found. So route the generation to ChatGPT, Claude or Gemini, and use this for the two steps that bracket it. Before you brainstorm: what already exists. What the collections in your trade are called, what the shops near you already sell at that hour, what somebody else tried and reported. That scan stops you spending a week on an idea that has a name and a competitor already. After you brainstorm: whether your shortlist is taken, and what it would cost to do.

What are the limits of this technique in Perplexity?

The convergence is not a setting you can turn off. Phrase it as a brainstorm and the answer still arrives in the shape of a summary of the field, with your wildest request quietly dropped, because nothing in the index asked to be strange. Nor will it hold a quota: told to produce five unconventional ideas it produces five ordinary ones labelled unconventional. Two things it does genuinely well in this chapter: an existing-landscape scan before you start, and a reality check after. Everything in between - the thirty seeds, the odd ones, the combining of 4 with 12 - belongs in another tool.

How do I improve the answer after the first try?

When the first list arrives, do not hunt for a single winner yet. Mark the three or four ideas that spark something, even if they are rough. Then loop: "Give me ten more like numbers 4 and 9," or "Combine 2 and 17." If the list feels too safe, say "Make the next ten bolder." If it drifts away from your situation, repeat one key fact, such as your budget or your customers. Two or three rounds usually turn ore into gold. Save the prompt that worked so you can reuse it next season.

Loop it

When the first list arrives, do not hunt for a single winner yet. Mark the three or four ideas that spark something, even if they are rough. Then loop: "Give me ten more like numbers 4 and 9," or "Combine 2 and 17." If the list feels too safe, say "Make the next ten bolder." If it drifts away from your situation, repeat one key fact, such as your budget or your customers. Two or three rounds usually turn ore into gold. Save the prompt that worked so you can reuse it next season.

This lesson on the other tools

The technique is the same; the way you write it is not. ChatGPT, Claude, Gemini.