Artificial Idea | AI careers · practical prompts · no hype Thursday, October 2, 2025 · Issue #18 · Prompt Tutorial
The research advantage
The research prompt stack: how consultants use AI to do 4 hours of research in 20 minutes
The bottleneck in most knowledge work is not thinking. It is the time required to gather enough information to think well. That bottleneck is gone.
Issue #17 made the case that the middle management roles surviving the current restructuring wave are those where value is concentrated in judgment rather than coordination. Judgment, however, does not operate in a vacuum. It operates on information. And the quality of the judgment a professional produces is directly limited by the quality, depth, and relevance of the information they are working with.
This creates a specific problem that AI solves better than almost any other application in the knowledge work context. Gathering sufficient information to make a good decision or produce a credible analysis has always been time-consuming, often consuming more of a professional's week than the actual thinking the information is meant to support. A consultant spending three hours building a competitive landscape before they can begin the strategic analysis. An analyst spending two hours synthesising industry reports before writing a single line of their own assessment. A business development professional spending ninety minutes researching a prospect before a thirty-minute call.
The research prompt stack below compresses that gathering phase without compressing the thinking phase. The thinking remains yours. The information assembly becomes a fraction of what it was.
How to use this stack
These prompts are designed to be run in sequence for any significant research task. Each one builds on the output of the previous one, moving from broad landscape to specific insight to actionable synthesis. Run them as a connected dialogue in a single session rather than as isolated queries, because the context accumulated across the session improves the quality of each subsequent output.
One important caveat before the prompts. AI language models have knowledge cutoffs and do not have real-time access to current data unless explicitly connected to search tools. For research tasks where recency matters, use these prompts in a tool with web search capability enabled, such as ChatGPT with browsing, Claude with search, or Perplexity. For research tasks where the underlying dynamics are relatively stable, such as competitive positioning, industry structure, or strategic frameworks, the knowledge cutoff is rarely a significant limitation.
Prompt 1: The landscape builder
The problem it solves: getting a rapid, structured orientation to an unfamiliar market, industry, or competitive context before going deeper.
You are a senior research analyst at a top-tier
consulting firm. I need a rapid orientation
to the following topic before conducting
deeper analysis.
Topic: [describe the market, industry, company,
or competitive context you are researching]
Purpose of my research: [what decision or
output this research is feeding into]
What I already know: [brief summary of your
existing knowledge to avoid covering ground
you do not need covered]
Please provide:
1. A structured overview of the landscape
in four to five paragraphs, covering the
key players, dynamics, recent developments,
and the two or three most important tensions
or trends shaping the space right now
2. The five most important things someone
needs to understand about this space
that are not obvious from a surface reading
3. The conventional wisdom about this space
that is worth questioning, with your reasoning
4. The three most useful sources or data sets
I should consult to go deeper on the
most important dynamics
Flag clearly where your information may be
outdated and where I should verify with
current sources before acting on it.
Point three, the conventional wisdom worth questioning, is the output that most research processes never produce because they are designed to confirm what is already believed rather than challenge it. It consistently surfaces the most valuable insight in the landscape building phase and takes the analysis somewhere more interesting than a standard competitive overview would go.
Prompt 2: The competitive intelligence builder
The problem it solves: developing a structured, comparative understanding of how specific competitors or players are positioned, without spending hours on individual company research.
You are a competitive intelligence analyst
preparing a briefing on the following
set of competitors or players.
Context from previous research: [paste
relevant output from Prompt 1]
Players to analyse: [list the specific
companies, products, or players]
For each player, please provide:
1. Their current strategic positioning
in one paragraph: what they are
optimising for and who they are
primarily serving
2. Their most significant competitive
advantage and how durable it is
likely to be over the next two to
three years
3. Their most significant vulnerability
or the assumption underlying their
strategy that is most likely to
prove wrong
4. How they are likely to respond
to a new entrant or significant
market change
After covering individual players, please:
5. Identify the competitive dynamic
that the entire field is organising
around, whether explicitly or implicitly
6. Identify any positioning territory
none of the players are currently
claiming that represents a genuine
strategic opportunity
Flag where your analysis is based on
limited information and where
additional research would materially
change the picture.
The instruction to identify unclaimed positioning territory transforms a competitive analysis from a descriptive exercise into a strategic one. Most competitive research tells you what exists. This prompt is specifically designed to tell you what does not exist but could, which is where the most valuable strategic insights consistently live.
Prompt 3: The stakeholder mapper
The problem it solves: understanding the human landscape of a market or organisation, including who holds influence, what they care about, and how decisions actually get made rather than how they are supposed to get made.
You are an organisational and market analyst
helping me understand the stakeholder dynamics
in the following context.
Context: [describe the market, organisation,
or decision-making environment you are
mapping, drawing on outputs from
previous prompts where relevant]
Please map:
1. The key stakeholders or decision-makers
in this context, categorised by their
formal authority and their actual influence
where those two things differ
2. What each key stakeholder primarily
cares about, both professionally
and in the context of this specific
situation
3. The alliances, tensions, or conflicts
of interest between stakeholders that
are most likely to affect how decisions
get made
4. The stakeholder whose support is most
critical to a successful outcome and
the stakeholder whose opposition is
most likely to derail one
5. The information or framing most likely
to be persuasive to the most critical
stakeholder, based on what you have
identified about their priorities
This analysis should reflect how organisations
actually work, not how their formal structures
suggest they should work.
The explicit instruction to reflect how organisations actually work rather than how their structures suggest they should is the constraint that makes this prompt useful rather than generic. Formal stakeholder maps that follow org chart logic consistently miss the people and dynamics that actually determine outcomes. This prompt is designed to surface the informal power structure alongside the formal one.
Prompt 4: The insight synthesiser
The problem it solves: turning the accumulated output of the previous three prompts into a concise, actionable synthesis that can be used directly in a presentation, report, or decision-making conversation.
You are a senior partner preparing to
present research findings to a senior
client or leadership team. You have
twenty minutes and you need to communicate
the most important insights, not
a comprehensive summary.
Here is the research I have conducted:
[paste the most important outputs
from Prompts 1 through 3]
Purpose of this synthesis: [what decision
or action this research is meant to inform]
Audience: [who will receive this synthesis
and what they care about most]
Please produce:
1. The single most important insight
from this research in one sentence,
the thing that should change how
the audience thinks about this situation
2. Three supporting findings that substantiate
the headline insight, each in two
to three sentences with the most
relevant evidence
3. The most significant uncertainty
in this analysis and what would
need to be true for it to materially
change the conclusion
4. The recommended next action, stated
as specifically as possible given
what the research supports
Constraints: No findings without evidence.
No recommendations that the research
does not support. If the research is
insufficient to support a recommendation,
say so and specify what additional
research would be required.
The constraint that recommendations must be supported by the research is the one most likely to produce useful pushback from the model. When it flags insufficient evidence for a conclusion, that flag is valuable. It tells you where your research has gaps before you present findings that a well-prepared audience will identify as unsupported.
Prompt 5: The question generator
The problem it solves: identifying what you still do not know after completing the research, and formulating the specific questions that would most improve the quality of your analysis if answered.
You are a rigorous research director
reviewing the work of an analyst who
has completed an initial research phase.
Here is the research conducted so far:
[paste key outputs from previous prompts]
Here is what the research is meant to
inform: [decision or output]
Please identify:
1. The three most important questions
this research has not answered,
ranked by their potential impact
on the conclusions
2. For each unanswered question,
the most efficient way to get
a reliable answer: a specific
source, a type of expert to consult,
or a data set to obtain
3. The assumption in the current
analysis that is doing the most
work and has the least evidence
behind it
4. Whether the research conducted
so far is sufficient to support
the decision or output it is
meant to inform, and if not,
what the minimum additional
research requirement is
Be direct about gaps. An analysis
that presents itself as more complete
than it is causes worse decisions
than one that is honest about
its limitations.
Running this prompt at the end of a research session consistently produces two things. First, a clearer sense of confidence level: how much you actually know versus how much you think you know. Second, a prioritised list of what to do next if time permits, rather than a vague sense that more research would be better.
The combination of those two outputs is more valuable than either alone.
The 20-minute schedule
Prompt 1: six minutes, including reading and annotating the output. Prompt 2: five minutes for two to three competitors at the level of depth this prompt produces. Prompt 3: four minutes. Prompt 4: three minutes, plus two minutes editing the synthesis for your specific audience.
Total: twenty minutes for a research foundation that would previously have taken a junior analyst half a day to assemble to a comparable standard of structure and insight.
The twenty minutes does not include verification of specific facts, current data gathering for time-sensitive research, or the deeper domain expertise that experienced professionals bring to interpreting what the research means. Those elements remain human responsibilities. What the stack eliminates is the assembly time: the hours of reading, tabbing between sources, and organising notes that precede the actual analysis.
Assembly is not analysis. These prompts handle the assembly. The analysis is still yours.
Monday we are addressing something that has been building as a theme across the past several issues and that deserves direct treatment: the specific professional behaviour that most reliably signals to an organisation that someone is a person of genuine strategic value rather than operational competence. It is not what most people are currently working on developing.
The gap between the two is where careers stall, and it is a gap that AI has made simultaneously more visible and more closeable than at any previous point.
— The Artificial Idea team
