Chapter III
AfiniTwin: five real use cases with detailed workflow
No theory: five concrete human profiles, each with their downloaded preset, active mode and step-by-step workflow. Not selling «possibilities»: showing how this is actually used on a Tuesday morning.
Five profiles, five uses
Talking about «use cases» in the abstract is useless. What matters is who’s at the keyboard and what they want: the same AfiniTwin serves an executive decision-maker, a creative writer, a clinician who needs framing, a PhD candidate preparing a defence and a salesperson refining a pitch. Five different flows.
What repeats in all of them: the AfiniTwin loads before the first question, calibrates the LLM, and stays there until you switch topic or mode. What changes is the preset, the active mode and the frequency of switches.
Anatomy of a case
Each case below follows the same structure:
- Person profile — who they are and what they’re after.
- Recommended preset — which AfiniTwin variant they download.
- Step-by-step workflow — the actual sequence: install the system prompt, open the conversation, activate the mode, iterate.
- Expected result — what changes in their work compared with doing it without AfiniTwin.
When AfiniTwin is overkill
Knowing when not to use AfiniTwin matters as much as knowing when to. If all you ask of the AI is summaries, literal translations, boilerplate, routine email templates or transcripts, the profile adds nothing. Base AI does that well.
AfiniTwin adds value when the LLM’s answer changes depending on who’s asking. If the answer would be the same to anyone, there’s nothing to calibrate.
The five cases
Each with person, preset, workflow and result.
Executive decision-maker
Critic + Devil’s AdvocateCOO preparing an irreversible decision: reorganising the sales team. Wants to avoid the AI validating the first option she thinks of.
Workflow
- Downloads the «Critic» preset and pastes it as system prompt in Claude Projects.
- Opens conversation with: «I’m about to split sales into two teams. Activate Devil’s Advocate mode and give me three angles where this decision can blow up.»
- AI responds with three concrete failure lines, not generic ones, calibrated to her high dissent tolerance and short-term aggressive time frame.
- Iterates three rounds: each answer opens a new critical question, not applause.
Result
Walks into the committee with a decision that survived a reproducible adversarial critique, not a complacent brainstorm.
Creative professional (copywriter)
GhostwriterFreelance writer running newsletters for three clients. Needs the AI to write in his voice when delegating drafts, not in a generic LinkedIn-friendly register.
Workflow
- Loads the «Ghostwriter» preset into a ChatGPT Custom GPT.
- Pastes the client’s raw brief and asks: «Write this in my voice, keeping the humour and rhythm of the last newsletter.»
- AI mimics his declared humour (ironic, not sarcastic), preference for short sentences and tolerance for unusual metaphors.
- Reviews, adjusts three sentences, ships.
Result
Saves 90 minutes per newsletter without losing signature. The AI stopped sounding like a generic community manager.
Clinician (psychologist)
AnalystPsychologist receiving a patient with their Big Five report; wants to prep the first session without getting lost in the popular-report jargon.
Workflow
- Loads the «Analyst» preset (her own profile) in Claude.
- Pastes the patient’s report and asks: «Translate these scores into clinical operational language, no coaching tone. Mark which facets to address in the first session.»
- AI replies in technical key, no motivational phrasing, flagging high-clinical-sensitivity facets (N1, N2, N4).
- Cross-checks against her own judgement.
Result
Arrives to the session with a technical frame that respects the report language but is calibrated to her own clinical style (more direct, less exploratory).
PhD candidate
CriticMoral philosophy PhD candidate preparing his defence. Needs a critical peer who doesn’t flatter.
Workflow
- Pastes the «Critic» preset into a Gemini Gem.
- Uploads the central chapter and asks: «Argue the three angles a hostile committee will attack me on here.»
- AI produces counterarguments with real citations, calibrated to his very high disagreement tolerance and his declared philosophical frame (continental, not analytic).
- Iterates eight rounds in an afternoon.
Result
Walks into the defence having sparred with a critical partner who didn’t sugar-coat, calibrated to his own way of thinking.
B2B salesperson
Mirror + TranslatorAccount executive prepping a hard call with a client who sent a cold email. Wants to anticipate subtext and rehearse his reply.
Workflow
- Loads the «Mirror» preset into his private API via AfiniTwin’s B2B endpoint.
- Pastes the client’s email and asks: «Activate Translator mode. What does this email say between the lines according to my way of reading?»
- AI interprets subtext from the salesperson’s frame (low anxious attachment, high literal reading) and proposes three reformulations.
- Rehearses the answer in Mirror mode: AI speaks as him, anticipates objections, adjusts tone.
Result
Walks into the call with three readings of the email, not one, and a rehearsed answer that sounds like him. Closes with a confirmed follow-up.
Your case, your workflow
If you see yourself in one of the five — or a combination — the AfiniTwin downloads in five minutes and is in use the same day.