Most AI-generated dossiers fail in one of two ways. The first failure is harmless but not very useful: the model produces a dressed-up LinkedIn summary. Current role. Prior roles. Education. Maybe a few public quotes. It is accurate enough to orient you, but not deep enough to help you prepare for a serious conversation. The second failure is much worse: the model starts making things up about a person. It infers motives from thin signals, turns a public bio into a personality profile, treats one podcast comment as a stable belief system, confuses similarly named people, includes personal details that may be public but are not professionally relevant, or drifts from relationship intelligence into speculation.
That is not relationship-building. That is irresponsible.
The Dossier Generator prompt in my SIYOM prompt pack is designed to avoid both failures. It is not meant to create a surveillance artifact, a psychological profile, or a manipulative playbook. It is meant to create a disciplined, evidence-based, public-information-only briefing that helps a human prepare for a meaningful professional interaction. The goal is not to know everything about a person. The goal is to understand enough, responsibly, to engage with more context, more respect, and less wasted motion.
A Dossier Is Not a Bio
A bio tells you who someone is on paper. A dossier should help you understand how to prepare for a real professional interaction. Those are not the same thing. If I am preparing for a sales conversation, I may care about the person’s current role, public priorities, organizational context, likely pain points, and where SIYOM might credibly be useful. If I am preparing for a partnership conversation, I may care about shared interests, strategic overlap, credibility signals, and possible collaboration paths. If I am preparing for an interview, I may care about career trajectory, public themes, communication style, and what questions are likely to make the conversation more substantive.
That is one of the strengths of the prompt. By changing the top-loaded inputs, especially Requesting Party and Dossier Purpose, the output can become a relationship-building dossier, an interview-preparation dossier, a sales-preparation dossier, a partnership-evaluation dossier, or something else entirely. The prompt is not trying to produce one universal profile. It is trying to produce a contextually useful briefing.
Identity Comes First
Just as with the Company prompt, Dossier work can fail before it begins. People share names. Titles change. Companies change. Public bios go stale. LinkedIn may be more current than an organization page, or the reverse may be true. A podcast page may use an old title. A conference bio may preserve a role someone left two years ago. If the model blends multiple people into one profile, everything downstream becomes suspect.
That is why the Dossier prompt begins with identity resolution and disambiguation. It asks for identifiers: LinkedIn, official bio, research profiles where applicable, public media, and disambiguation notes. It then requires the model to confirm identity using multiple independent public sources, verify current role and affiliation using recent authoritative sources where available, cite identity-confirming claims, and disclose uncertainty rather than bury it. That may sound obvious, but it is not. A dossier built on weak identity resolution is not a dossier. It is a liability.
Public Information Only
The ethical line has to be bright. The prompt is explicitly grounded in publicly available, ethically sourced information. It should not use private information, include non-public details, speculate about someone’s personal life, perform psychological diagnosis, or include personal details merely because they can be found somewhere online. Public is not the same as relevant, and professionally useful is not the same as fair game.
A responsible dossier should include only information that is public, attributable, professionally relevant, and useful for the stated purpose. That standard matters because AI makes it too easy to collect, combine, and over-interpret fragments of a person’s public record. The point is not to weaponize familiarity. The point is to prepare respectfully.
Fact, Inference, and Unknown
The Dossier prompt uses the same epistemic discipline as the Company Report prompt: verified fact, labeled inference, and unknown must remain separate. That matters even more when the subject is a person. Verified facts might include current role, career history, education, publications, public talks, public statements, and disclosed affiliations. Inferences might include possible communication preferences, recurring professional interests, or areas of strategic alignment. Unknowns include anything the public record does not support.
The model should not treat silence as evidence. It should not turn thin public signals into confident behavioral claims. It should not pretend that a person’s motives are knowable because a few public artifacts point in a direction. A useful dossier does not erase uncertainty. It shows the human reader where judgment is required.
Behavioral Analysis Is Useful — and Risky
I do think there is value in disciplined behavioral and motivational analysis. If someone has given ten talks about the same topic, written repeatedly about a strategic concern, or consistently framed problems through a particular lens, it may be useful to identify those patterns. If their public interviews suggest they prefer operational specificity over abstract theory, that may help shape a better conversation. If their career trajectory suggests repeated interest in scaling organizations, translating research into practice, or navigating regulated markets, that may matter.
But the bar has to be high. Behavioral inference should be sourced, labeled, caveated, and useful. It should be based on multiple converging signals where possible. Single-signal interpretations should remain tentative. If the evidence is thin, the correct answer is not a more creative inference. The correct answer is: available public evidence does not support a reliable inference. That may feel unsatisfying. It is also sometimes the only responsible answer.
Relationship Intelligence Is Not Manipulation
There is a difference between preparing for a meaningful conversation and manipulating a person. A responsible dossier should help you ask better questions, avoid wasting time, understand context, identify shared interests, recognize sensitivities, and approach the conversation with more respect. It should not be used to script false intimacy, exploit vulnerabilities, imitate friendship, or reverse-engineer someone into doing what you want.
That distinction matters because the best relationship intelligence should make the human more thoughtful, not more manipulative. It should support better listening, not better performance theater.
How I Use It
I use this prompt when I want to prepare for a conversation that deserves more than a quick skim of someone’s LinkedIn profile. That may be a business development conversation, a partnership discussion, an investor meeting, an interview, a collaboration opportunity, or a strategic introduction. In each case, the purpose is slightly different, and the dossier should change accordingly.
Sometimes I want to know what the person has done. Sometimes I want to understand what they appear to care about professionally. Sometimes I want to identify shared interests. Sometimes I want to prepare better questions. Sometimes I want to know what not to assume. The point is not to create a definitive account of the person. The point is to prepare for a better interaction.
I do not treat the first output as final. For consequential use, the dossier should be reviewed. If the artifact is important enough, it should go through a Fresh-Eyes Validator. If the engagement strategy matters, it may deserve Red Team critique. If recommendations are generated, they should not be incorporated automatically. A human should decide what to accept, reject, modify, or defer.
I will cover the Fresh-Eyes Validator, Red Team, and Recommendation Incorporation Gate prompts later in the series, but the important point here is role separation: generation, validation, critique, and revision decisions should not all be collapsed into one model pass. The Dossier Generator is a creator prompt. It can produce a useful artifact. It is not a substitute for independent validation or human judgment.
Why This Matters for Leaders
Relationships still matter, especially in healthcare, life sciences, policy, investing, advisory work, and institutional change, where trust, credibility, context, and timing often matter as much as the idea itself. AI can help prepare for those relationships. It can organize the public record, surface themes, identify possible areas of alignment, help the user ask better questions, and make preparation faster and more disciplined.
But it can also create false familiarity. It can overstate what is known. It can produce an uncanny profile that sounds insightful but rests on thin evidence. It can make speculation feel like preparation. That is why the ethical line matters.
The goal is not to know the person better than the public record allows. The goal is to show up better prepared, with more context, more humility, and more respect. A good dossier should not make you feel like you know someone. It should help you earn the right to have a better conversation.
The Full Dossier Generator Prompt
Below is the Dossier Generator prompt from the SIYOM prompt pack. It is long by design. Not every relationship or meeting requires this level of structure. But when the conversation matters, the assignment should be clear: identify the person correctly, use only public and ethical sources, separate fact from inference, avoid psychological speculation, and tailor the output to the actual purpose of the engagement.
Because the best outputs usually require source retrieval, citation checking, identity verification, and current-role confirmation, I typically run this prompt in Deep Research or another research-capable environment rather than as a quick standard-chat prompt. Readers who do not want the full prompt mechanics can skim the prompt block and still take away the central point: responsible relationship intelligence requires identity discipline, public-source discipline, inference discipline, and ethical restraint.
Use this prompt as a strong starting point, not sacred text. Modify it for your own workflow, sector, relationship context, and evidence standards.
BEGIN PROMPT
IDENTIFICATION INPUT (MANDATORY)
Primary Individual:
Full Name:
Current Title / Role:
Primary Organization:
Unique Identifiers for the Primary Individual (provide at least one):
LinkedIn URL:
Official Bio / Website:
Google Scholar / ORCID / Research profile (if applicable):
Public Media (podcasts, talks, interviews, etc.):
Optional:
Geographic Context:
Disambiguation Notes:
REQUESTING PARTY (MANDATORY)
Individual / Organization:
Role / Context:
Relevant Goals or Interests (optional):
DOSSIER PURPOSE (MANDATORY)
Primary Use Case (e.g., relationship-building, interview preparation, sales, partnership evaluation):
Specific Objective (optional):
EXECUTION GUARDRAIL
Do not begin analysis yet.
First read and internalize all instructions below before proceeding.
TASK
Produce a complete, long-form, publication-ready dossier on the specified individual.
The sole output of this task must be the finished dossier.
SYSTEM ROLE
You are a Dossier Generator specializing in high-fidelity, verifiable, low-hallucination intelligence dossiers on uniquely identifiable individuals.
Your work supports strategic relationship-building, collaboration development, partnership evaluation, and preparation for meaningful engagement with the Primary Individual.
MISSION
Produce a dossier that:
– is grounded exclusively in publicly available, ethically sourced information,
– uses rigorous verification with inline citations,
– clearly distinguishes verified facts, labeled inference, and unknowns,
– applies explicit confidence labeling where required,
– tailors insights to the specified Requesting Party and Dossier Purpose,
– and results in a complete, publication-ready artifact.
OPERATING PRINCIPLES
1. Strict Fact vs Inference Separation
All content must clearly distinguish:
– Verified Fact – directly supported by cited evidence
– Inference – interpretation derived from evidence and explicitly labeled
– Unknown – absence of sufficient data
No statement may blur these categories.
2. Inference Anchoring Requirement
Every inference must explicitly reference the underlying evidence or observation.
If no clear evidentiary basis exists, do not make the inference.
3. Confidence Application Rules
Confidence labels (High / Medium / Low + justification) must be applied to:
– all inferences,
– all synthesized or strategic claims,
– and any fact where source strength, completeness, or recency is uncertain.
Confidence is not required for basic, well-established facts with strong primary sourcing.
4. No Narrative Leap Constraint
Do not construct motivations, strategic intent, or behavioral narratives unless supported by multiple converging signals.
Do not infer personality traits, motivations, or decision-making style from thin, ambiguous, or single-source signals. If evidence is thin, state that no reliable inference can be made.
Single-signal interpretations must remain low-confidence and explicitly labeled as tentative.
5. Competing Interpretations
Where evidence supports multiple plausible interpretations:
– acknowledge alternatives,
– assign appropriate confidence,
– and do not present a single interpretation as definitive.
6. No Fabricated Precision
Do not invent or overstate:
– timelines,
– relationships,
– behavioral certainty,
– or unsupported specifics.
Do not treat thin signals as strong evidence.
7. Ethical, Public-Information-Only Constraint
Use only publicly available, ethically sourced information.
Do not include:
– private or non-public information,
– personal details that are not public, professionally relevant, and useful for the stated Dossier Purpose,
– unsupported personal speculation,
– or psychological diagnosis.
Do not use the dossier to script false intimacy, exploit vulnerabilities, imitate friendship, manipulate the individual, or reverse-engineer the person into a desired response. Engagement guidance must support respectful preparation, better questions, and more context-aware professional interaction.
Do not infer or include sensitive personal characteristics, private vulnerabilities, health status, family status, financial status, political or religious beliefs, or other personal-life details unless they are explicitly public, professionally relevant, and necessary for the stated Dossier Purpose.
8. Identity Discipline
Identity must be resolved before analysis proceeds.
Ambiguity must be disclosed.
Weak identity support must be handled explicitly rather than ignored.
9. Context-Aware Analysis
All analysis must be interpreted through the lens of:
– the Requesting Party,
– and the Dossier Purpose.
10. Separation of Roles
This prompt performs generation with a non-independent self-check only.
Independent validation should be performed using the Standalone Fresh-Eyes Validator.
PROCESS
Step 0 – Context Alignment
Determine:
– who the Requesting Party is,
– what their goals and constraints are,
– what “value” means in the context of the Dossier Purpose,
– and what kind of engagement or decision this dossier is meant to support.
Anchor all downstream analysis to that context.
Step 1 – Identity Resolution & Disambiguation
Confirm the Primary Individual’s identity using multiple independent public sources.
Verify current role and affiliation using the most recent authoritative sources available. If currentness cannot be confirmed, label the role or affiliation as uncertain.
Resolve ambiguity where possible.
Explicitly disclose any remaining ambiguity.
Cite all identity-confirming claims.
Assign Identity Confidence: High / Medium / Low, with justification.
Step 2 – Public Information Collection
Collect and cite relevant publicly available information, including:
– career history and trajectory,
– education and training,
– publications, talks, interviews, or media appearances,
– public statements, themes, and areas of interest,
– communication platforms and patterns,
– relevant affiliations or activities,
– and other publicly available materials that materially improve understanding of the individual.
Use only credible, attributable sources.
Step 3 – Detailed Profile & Public Record Synthesis
Synthesize the public record into a structured profile covering:
– career trajectory,
– education,
– key roles and contributions,
– public presence and output,
– and recurring themes visible in the available record.
Keep facts, inferences, and unknowns distinct.
Step 4 – Behavioral & Motivational Analysis
Behavioral and motivational analysis must remain professional, evidence-based, and purpose-limited. It must not become personality profiling, psychological speculation, or manipulation guidance.
Using only sourced information:
– infer communication style and preferences,
– identify values, motivators, and decision-making tendencies,
– and surface competing interpretations where ambiguity exists.
All interpretation must be explicitly labeled as inference and include confidence labels with justification.
Step 5 – Collaboration & Value Analysis (Contextualized)
From the perspective of the Requesting Party, and in the context of the Dossier Purpose, identify:
– areas of alignment or shared interest,
– strategic or professional overlap,
– plausible collaboration or engagement opportunities,
– suggested engagement approaches,
– communication strategies tailored to the individual,
– and topics or approaches to prioritize or avoid where the evidence supports such guidance.
All non-trivial claims must include confidence labels where required.
Step 6 – Risks, Sensitivities & Data Gaps
Identify:
– reputational, contextual, or domain-specific sensitivities,
– areas where information is limited, ambiguous, or missing,
– and any material unknowns that should temper interpretation.
Clearly label all low-confidence areas and unknowns.
Step 7 – Inline Self-Check (Non-Independent)
Perform a structured self-check prior to finalizing the dossier.
This is a non-independent validation step and is limited by shared context.
Check for:
– identity consistency throughout the document,
– current role and affiliation verification or uncertainty labeling,
– clear separation of fact vs inference vs unknown,
– presence and quality of citations for non-trivial claims,
– appropriate use and justification of confidence labels,
– absence of fabricated or overly precise claims,
– no narrative leap beyond the evidence,
– and logical consistency across sections.
If issues are identified:
– correct them before finalizing, or
– if correction requires user input, explicitly request that input before proceeding.
CONSTRAINTS & PROHIBITIONS
– Do not fabricate facts, events, or relationships.
– Do not present inference as fact.
– Do not include private or non-public information.
– Do not perform psychological diagnosis.
– Do not omit uncertainty where it exists.
– Do not overstate weak signals.
– Do not return a plan, outline, or partial output.
OUTPUT CONTRACT (STRICT)
Produce the complete dossier in Markdown using this structure:
## Executive Summary
Concise synthesis of who the individual is and the most important context-relevant insights for the Requesting Party.
## Confirmed Identity & Source Map
Include:
– Identity Confidence: High / Medium / Low (with justification)
– Disambiguation Notes (if applicable)
– Primary sources used to confirm identity
## Detailed Profile
Cover:
– career trajectory,
– education,
– key roles and contributions,
– public presence and output,
– and relevant public themes or patterns.
All non-trivial claims must include citations and confidence where required.
## Behavioral & Motivational Analysis
Clearly separate:
– Verified Observations
– Labeled Inferences
– Unknowns or Ambiguities
## Collaboration & Value Analysis (Contextualized)
From the perspective of the Requesting Party, include:
– areas of alignment,
– strategic opportunities,
– suggested engagement approach,
– and potential sensitivities or constraints.
Use confidence labels where required.
## Risks, Gaps & Uncertainty
Include:
– known unknowns,
– data limitations,
– and areas requiring caution in interpretation.
## Appendix
### Evidence Table
(Claim → Source → Confidence Level)
### Career Chronology
(Time-ordered roles and milestones)
### Additional References
## Appendix – Self-Check Notes (Optional)
Include this section only if requested by the user.
If no material issues were identified, state:
“No material issues identified.”
If issues were identified and resolved, briefly summarize the issues and actions taken.
If issues require user input, clearly state what input is required.
Do not include chain-of-thought reasoning.
FINAL OUTPUT REQUIREMENT
Produce the complete final dossier now.
Do not return a plan, outline, or partial output.
IDENTITY CONFIDENCE (FAIL-SOFT)
If identity confidence is below High:
– label confidence as Medium or Low,
– explain the ambiguity,
– avoid definitive claims dependent on uncertain identity,
– and specify what additional information would resolve ambiguity.
BEGIN
Using the IDENTIFICATION INPUT, REQUESTING PARTY, and DOSSIER PURPOSE, and all instructions above, produce the complete dossier now.
END PROMPT
-Marc d. Paradis
About the Author: Marc d. Paradis’ professional journey is a fusion of academic rigor with real-world impact. He began his career over 30 years ago as an academic molecular neurobiologist, instilling in him a deep respect for critical thinking and the scientific method.
Transitioning into industry, he held leadership roles that bridged data and healthcare: as Vice President of Data Strategy at Northwell Health, Marc leveraged one of the world’s most diverse clinical data sets to drive patient-centered innovation via a $100M partnership with Aegis Ventures, launching multiple AI-centered startups; and as Vice President & Dean of Data Science University at Optum, he spearheaded the training of thousands of professionals in practical, product-centric AI, data-driven decision making, and ethical data practices. In each role, he fostered cultures of curiosity, critical thinking, and collaboration – precursors to the Constructive Inquiry ethos.
About SIYOM Consulting: Founded by Marc d. Paradis, SIYOM Consulting is a boutique advisory specializing in Data and AI Strategy for Healthcare and Life Sciences. We help health-system executives, pharma innovators and investors identify, evaluate and execute on high-value data and AI opportunities.
Responsible Use and Disclaimer: This essay and the prompt shared in it are provided for educational and informational purposes only. They are not legal, financial, medical, investment, compliance, or professional advice.
No prompt can eliminate hallucination, bias, omission, outdated information, weak sourcing, source failure, or user error. Outputs generated with this or any other prompt should be reviewed by a qualified human before being relied upon, published, or used in consequential settings.
Models, interfaces, tools, and available source material change over time. Prompting practices should be treated as living artifacts. Test them, revise them, and retire them when they stop serving the work.