The Discipline of Precise Limitation

This is the chapter Perplexity fits worst. It answers questions with sources rather than writing to a word count, and a banned-words list or a strict table format will not hold. The honest adaptation flips the task: instead of asking it to obey your constraints, ask it whether your constraints are right - whether 150 words is a real standard, whether your analogy hides something, whether a no-impact exclusion is what the evidence actually supports.

• Ask whether your constraint is justified, not for obedience.
• Expect word limits and strict formats to be ignored.
• Academic focus for evidence behind a safety constraint.
• Ask what your chosen analogy is hiding.
• Skip it entirely when no sourcing is needed.

Constraint-following is where Perplexity is weakest - it writes to answer a question, with citations, not to a 150-word limit with a banned-words list. Do not expect the word count to hold. What it can do that the others cannot is tell you whether the constraint itself is a good idea.

I am about to write a 150-word explanation of what an API is for a non-technical project manager, built on the analogy of a waiter taking an order to the kitchen.

1. How do widely-read explainers aimed at non-technical readers actually explain APIs? Cite the sources and quote the framing each uses.
2. How common is the waiter or restaurant analogy in these explanations? If it is near-universal, say so - my audience may already have met it, which would waste the one analogy I get to use.
3. Which analogies for APIs have been publicly criticised as misleading, and on what grounds? Cite the criticism.
4. Where does the waiter analogy actually break down against how APIs work - authentication, rate limits, versioning, asynchronous responses? Cite technical sources. This is what I most need: I want to know what my simplification is hiding before a developer in the room points it out.
5. Is there published guidance on explanation length for business audiences - does 150 words have anything behind it or is it a number I invented? Cite whatever exists, and say plainly if nothing does.

Cite everything. Then I will write the constrained 150 words in a tool that will respect a word count.

Skip Perplexity for this one. A celebratory internal email to your own sales team needs no sources, no live web search and no citations - it needs a tone you control, and this tool will cite its way through a congratulations note. Write it in ChatGPT, Claude or Gemini.

There is a narrow version that is worth asking here, and only if your constraint set is the thing you are unsure about rather than the email:

1. What does published research on workplace recognition say about how team achievements are best announced - public versus private, team-level versus individual, immediate versus delayed? Use Academic focus and cite the studies.
2. Is there evidence on whether naming individuals in a team announcement helps or harms morale? I have constrained my email to exclude individual names and I want to know whether that constraint is right. Cite both sides if the evidence is mixed.
3. What is documented about jargon in internal communication - does avoiding it measurably improve how messages land, or is that just a style preference? Cite what exists and say so if the evidence is thin.
4. Are there Indian workplace-specific findings on recognition and celebration norms, or is the literature mostly Western? Say which, plainly.

That gives you a defensible constraint set. The email itself takes ninety seconds in a tool built for writing - and the answer to "should I single out my top performer?" is worth more than a draft.

Perplexity will not give you a well-formatted workout table and will not respect a negative-constraint list reliably. But a sensitive-knee exclusion is exactly the kind of constraint you want checked against evidence rather than against a model's confidence, and that is a sourced-research question.

I am building a 3-day-per-week beginner strength plan using only bodyweight exercises and dumbbells. Hard exclusions: no barbell, no machines, no jumping or running, no bench or pull-up bar. The no-impact constraint is because of sensitive knees.

1. What does published guidance say about strength training with knee pain or patellofemoral issues - which movement patterns are generally recommended and which are advised against? Use Academic focus. Cite the sources and give dates.
2. Which commonly-prescribed beginner exercises are specifically flagged in that literature as problematic for sensitive knees? Deep squats, lunges, step-ups - what does the evidence actually say, and where do sources disagree with each other?
3. What is the evidence on dumbbell-and-bodyweight-only programmes for building functional strength in beginners? Is anything meaningfully lost by excluding barbells at this level? Cite it.
4. What do recognised guidelines say about frequency and volume for a beginner - is three days a week supported? Cite the guideline bodies.
5. Where does published guidance say someone with knee pain should be assessed by a professional before following a general plan rather than self-prescribing? Quote it.

Cite everything, with dates, and flag anything you can only source to one place or to a commercial fitness site rather than clinical or research material.

Then take the evidence into another tool to build the actual table. Note what answer 5 says: if the sources point to an assessment, that is the finding, not an obstacle to route around.