Summary: The Branded AI Avatar - How to Transform a Generic AI into a Collaborator Who Knows Your Brand Inside Out
A branded AI avatar is a system of hundreds of interdependent parameters that allows artificial intelligence to accurately embody a brand: your correct tone, accurate arguments, consistent behavior in every situation, and decisions aligned with the founder's values. It's not just an improved prompt. It's a brand database built on 12 layers of knowledge, 27 tone sliders, and dozens of documented situations with their own specific parameters. Wiiv, a branding and strategic packaging agency based in Paris, regularly working in Bordeaux, Lyon, and Milan, builds these avatars as part of its Deepbranding method for e-commerce food, cosmetics, fashion, and lifestyle brands. The goal: for the AI to manage 80-85% of operational interactions in the exact brand tone, freeing up the founder for decisions that require true human judgment.
Wiiv Creates Your Branded AI Avatar: Hundreds of Parameters to Make Your AI Think, Speak, and Decide Like You
Imagine you're hiring a collaborator. You explain what the brand does in thirty minutes, give them access to the website, and let them manage social media, respond to customers, and help you make marketing decisions. This collaborator will do decent work. Polite, clean, error-free. And completely interchangeable with what any other collaborator would do under the same conditions.
This is exactly what 95% of brands do with their AI today.
Now imagine something else. You hire the same collaborator. But this time, before they start, you spend six weeks transmitting everything you know about your brand. Who your real target is and what makes them tick. The words you never use. How you react when a customer is aggressive. What you refuse to do even for money. The campaigns that worked and those that failed. Your deep convictions about your industry. The 50 situations they will encounter and how you want them to behave in each.
This collaborator is a partner. They think like you. They speak like you. They decide like you. And unlike a human, they are available 24/7, they don't get tired, they never deviate from what they've learned, and they can simultaneously manage a hundred conversations without losing consistency.
This is what we call a branded AI avatar. And this is what Wiiv builds.
For Billy, our own wellness brand, we built this avatar before launch. Today, the AI manages about 80% of customer interactions in the exact brand tone. Customers discovered later that some responses were AI-generated. Their reaction: surprise, not disappointment. Because the responses didn't sound like AI. They sounded like Billy.
What a Branded AI Avatar Really Is
A branded AI avatar is not a prompt. It's not a tone of voice written in two paragraphs. It's not copying and pasting your About Us page into ChatGPT.
It's a system. A structured brand database so that an AI can ingest, understand, and apply it in any situation. The difference between the two approaches is the same as between giving an instruction to an intern and onboarding a senior collaborator who will deeply integrate the company's culture.
A prompt gives a one-time instruction. The AI executes it and forgets. The next session starts from scratch. A brand avatar is a permanent context. The AI knows the brand. It knows who it is, how it speaks, to whom it addresses itself, what it refuses, how it behaves in each situation. It never starts from scratch.
What that changes in practice: with a prompt, you get something correct that you spend twenty minutes correcting to give it some personality. With a complete avatar, you get a first draft that already sounds like the brand, which you refine in two minutes. For high production volume, this difference represents hours every week. And for customer relationship quality, it represents the difference between a consistent brand and a brand that sounds different depending on the day and who responded.
"The question we ask every client at the start of a mission is: if your AI had to replace your best collaborator tomorrow morning, would it know what to do? In 99% of cases, the answer is no. Not because the AI is bad. Because no one ever really taught it."
Philippe, co-founder of Wiiv
A complete branded AI avatar relies on several hundred interdependent parameters. Removing one layer degrades everything else. Giving a tone of voice without giving product knowledge results in an AI that speaks well but answers incorrectly. Giving product knowledge without the tone results in an AI that answers correctly but sounds like a technical manual. It's a system, not a list.
With up to 850 custom parameters, your AI avatar creates an AI dedicated to your brand, and it works with ChatGPT, Claude, Gemini.
The 12 Layers of Knowledge That Build the Avatar
A branded AI avatar is built on 12 distinct layers. Each fulfills a specific function. Together, they allow the AI to operate autonomously and consistently in the vast majority of situations a brand encounters daily.
The first layer is the brand identity. The founder's story, the true purpose of the project, what angered them in their industry before launching, the mission, the promise, the tagline, and what it truly means. This layer gives the AI its reason to speak. Without it, it produces content without anchoring, without a point of view, without a reason for being other than to fill a format.
The second layer is the real target audience. Not demographic data, but the precise psychological portrait. Their real objections to buying, the exact words they use to describe their problem (not marketing words, but the words they type into Google at 11 PM), what makes them tick, what makes them flee, the brands they admire and why. With this level of detail, the AI writes for a real person. Without it, it writes for a statistic.
The third layer is positioning. What the brand is, what it is not, how it stands against three to five main competitors, what makes it unique, how it responds when compared to a competitor. This layer ensures the AI never produces content that could belong to anyone else in the industry.
The fourth layer is personality. If the brand were a person, who would they be exactly? Their character, their verbal tics, their relationship with humor, how they react under pressure, what excites them, what annoys them. This layer is what makes the AI recognizable. Without it, the tone is correct but impersonal.
The fifth layer is proprietary vocabulary. The brand keywords systematically used, recurring phrases, forbidden expressions, industry words to avoid, industry words to appropriate, banned formulations even when they seem correct. This is what makes brand content recognizable even without a signature.
The sixth layer is situational tone. We will come back to this in detail, but it is the most complex and important layer for customer interactions. It documents how the tone changes according to the objective, the channel, the moment in the customer journey, and the nature of the situation.
The seventh layer is the editorial territory. The topics discussed and why, the topics never discussed and why, the brand's own angle and point of view, the opinions the founder is willing to express publicly, those they keep to themselves. Without this layer, the AI can drift towards subjects that dilute the brand's editorial identity.
The eighth layer is product knowledge. Complete description of each product in the brand's tone, main arguments, specific objections, what can be promised and what cannot, frequently asked questions with the real answers. This layer allows the AI to answer specific questions without improvising.
The ninth layer is marketing strategy. Priority channels, effective acquisition levers, what has been tested and abandoned and why, short- and medium-term objectives, the KPIs that truly matter to the founder. Without this layer, the AI can produce content that is correct in tone but misaligned with the actual strategy.
The tenth layer is the operational context. Team size and who does what, available budget, tools in place, what the brand can produce itself versus what it outsources, deadlines and delivery constraints, peak activity periods during the year. This layer is essential for the AI's recommendations to be realistic and executable.
The eleventh layer is the competitive context. Main competitors with their real strengths and weaknesses, what they do well that we don't want to copy, what they do poorly that we exploit, positioning opportunities that no one has taken. This layer allows the AI to produce differentiating content rather than converging towards the industry center.
The twelfth layer is historical data and decision rules. Important past decisions and their results, successful and failed campaigns, mistakes not to repeat, the founder's real operational values (not the values displayed on the website, but the true implicit rules they apply without having written them down), and the criteria for evaluating an opportunity, a creative direction, or a partnership.
The 27 Tone Sliders: How We Calibrate the Brand's Voice
"Warm and professional tone" means nothing to an AI. It's an instruction that 80% of service brands include in their brief. The AI cannot deduce what that specifically means for your brand. It will produce something that resembles the statistically most common interpretation of "warm and professional," which yields exactly the same result as the 400 other brands that gave the same instruction.
To calibrate the tone precisely, we use 27 sliders rated from 1 to 10. Each slider opposes two extremes. The rating indicates where the brand stands on that spectrum. The combination of these ratings produces a unique tone footprint, impossible to confuse with another brand.
The sliders cover the general register (formal versus familiar, serious versus light, understated versus expressive), the brand relationship (institutional versus human and personal, modest versus assertive, discreet versus self-assured and proud), the customer relationship (distant versus close and intimate, formal address versus natural informal address, neutral versus warm), the content relationship (purely informative versus purely emotional, factual versus storytelling, short and dense versus developed and narrative), the humor relationship (no humor versus constant humor, subtle humor versus direct humor, never self-deprecating versus frequent self-deprecating humor), the conviction relationship (neutral and balanced versus sharp and nuanced, never taking a stance versus assumed opinion on everything, gentle versus direct and unfiltered), the industry relationship (within norms versus against norms, industry vocabulary versus exclusive proprietary vocabulary), the sales relationship (never hard-sell versus assumed and constant selling, rational argument versus emotional argument, discreet about price versus price highlighted), the proof relationship (no proof needed versus proof and figures everywhere, discreet testimonials versus omnipresent social proof), and the error relationship (we avoid versus total transparency, we avoid sensitive topics versus we address them directly).
27 sliders. Each rated from 1 to 10. These are 27 dimensions on which the brand distinguishes itself from all others. And these are 27 dimensions on which the AI can produce with a precision that "warm and professional" will never allow to achieve.
Without sliders: "Write a response to a customer who says our product is too expensive, warm and professional tone."
With sliders: "Write a response to a customer who says our product is too expensive. Register: familiar 8/10, assertive 9/10, natural informal address 10/10, emotional argument 7/10, never apologize for the price, hold the position without being aggressive, length: 4 to 6 sentences maximum."
The first produces a generic response. The second produces a response that sounds like the brand.
The 50 Situations and Their 17 Parameters: How the Avatar Knows What to Do in Each Case
A brand avatar cannot merely know the general tone. It must know how that tone modulates in each specific situation. Because a response to an enthusiastic customer is not written like a response to an aggressive customer. And an acquisition hook is not drafted like a loyalty email.
We have mapped over 50 distinct situations an e-commerce brand encounters. These cover pre-purchase (discovery, product question, range comparison, price hesitation, promo request, competitor comparison), ordering (payment issue, post-order modification, pro invoice, gift packaging), post-order (package tracking, delay, damage, wrong product, return, refund, exchange), satisfaction (5-star review, photo share, recommendation, new product request, long-term loyalty), dissatisfaction and conflict (unfair negative review, justified negative review, aggressive customer, threat of reporting, accusation of lying), social media (question comment, critical comment, positive tag, negative tag, collaboration request, ambassador request), B2B (potential reseller, distributor, journalist, investor, commercial partnership), crisis situations (bad buzz, negative article, serial quality issue, false information), and marketing and decision situations (opportunity evaluation, creative direction validation, creative brief, campaign analysis, content recommendation).
For each situation, we define 17 precise parameters. The register on the tone sliders, the ideal response length, the displayed empathy level, the urgency level in the response, whether to ask questions or assert, whether to offer a solution immediately or listen first, whether to speak on behalf of the brand or as a person, whether to name the customer or not, whether to apologize or not, whether to make a commercial gesture or not, whether to redirect to another channel or handle it here, whether to leave the last word to the customer or conclude, the typical opening word or phrase, the typical closing word or phrase, what is never said in this specific situation, an example of a perfect response, and an example of a response to avoid.
50 situations. 17 parameters each. That's 850 situational parameters just for this layer. To which are added the 27 tone sliders, the 12 knowledge layers, the example library, and the visual guidelines. We are talking about several hundred interdependent data points that together form a precise representation of the brand.
"When we present the volume of work to our clients, the reaction is always the same: they realize that their brand exists in their head with this level of precision, but no one else ever has access to it. Neither their teams, nor their service providers, nor their AI. The avatar is the way to get this precision out of the founder's head and make it transmissible."
Cynthia, co-founder of Wiiv
The Example Library: Why Real Cases Are Worth More Than All the Rules
Rules tell you what to do. Examples show you how. Both together produce something truly precise. Examples alone, without rules, yield an AI that imitates without understanding. Rules alone, without examples, yield an AI that understands without knowing how to embody.
The example library of a complete brand avatar covers several categories. On social media: an Instagram post that perfectly matches the brand, a failed Instagram post or one that didn't match it, a product caption the founder is proud of, a LinkedIn post about the brand. On marketing emails: a post-purchase welcome email, an abandoned cart recovery email, a successful newsletter, an inactive customer re-engagement message. On customer relations: an email to a satisfied customer written by the founder themselves, a well-handled email to a dissatisfied customer, an email to an aggressive customer, a response to a 5-star review, a response to an unfair negative review, a response to a justified negative review. On product and sales: a perfectly judged product sheet, an advertising hook that worked well, an advertising hook we regret publishing. On B2B and partnerships: an email to a potential partner, an email to a reseller, an accepted collaboration request response, a rejected collaboration request response. On brand presentation: a three-sentence pitch the founder said aloud that sounded right, a description for a directory or a press kit.
What is striking about this list is the importance of negative examples. The failed post. The regretted hook. The response that should have been phrased differently. These examples are often more valuable than good examples, because they clearly define the boundaries of what the brand is not. And it is often there that the AI deviates when it does not have these guardrails.
Spontaneous customer feedback deserves special attention. What customers say about a brand in their own words, in their reviews, in their messages, is often more revealing than what the founder says about themselves. These words, once integrated into the avatar, allow the AI to speak to the target audience in their own vocabulary. This is what creates an immediate sense of recognition: "this brand truly understands me."
Translating visual guidelines into words: how to guide AI on what it cannot see
An AI doesn't see mood boards. It doesn't feel the "volcanic inspiration" of a reference board. "Being inspired by Jacquemus in his relationship with the Mediterranean" means nothing to an LLM if this instruction isn't translated into concrete textual rules.
However, an AI well-fed on visual guidelines can brief a graphic designer, evaluate whether a creative direction fits the brand's universe, describe a visual to be produced, or guide a photographer without the founder being present. The condition: the visual universe must be translated into natural language with sufficient precision.
This translation relies on three types of information. Sensory and precise words that describe the visual universe: not "modern and clean" but raw, warm, mineral, soft, contrasted, artisanal, clinical, dense, airy, saturated. These words allow the AI to describe and evaluate visuals with a common vocabulary. Explicit rules: pure white is never used as a background, the logo is never placed on a photo of a person, photos are always taken in natural light, stock photos with smiling models are never used, packaging is always photographed on a textured background. And references translated into words: rather than providing a reference image, extract what it evokes in usable terms (formats, proportions, density, palette, atmosphere, what is seen and what is never seen).
The combination of all three allows AI to discuss visual universes with a precision that most human briefings do not achieve. This is not a substitute for human artistic direction. It is a tool that allows anyone on the team to brief within the correct territory without needing the founder on hand.
What it represents in actual volume
To understand why an AI brand avatar is not just an improved prompt, you need to have an idea of the actual volume of data it represents.
Each of the 12 knowledge layers produces several tens of parameters. The 27 tone sliders define the brand's sonic footprint across as many dimensions. The 50 documented situations, each with 17 parameters, represent 850 situational parameters. The example library covers several tens of real content pieces across all registers. Visual guidelines translated into words add an additional layer of precision to the brand's universe. Operational values, decision rules, campaign history, competitive context: all additional data that allows the AI to function autonomously and consistently.
In total, we are talking about several hundred interdependent parameters. This is not a document that can be produced in one morning. It is the result of several weeks of work, structured according to a precise method, to extract everything that makes the brand from the founder's mind and make it transmissible to an AI.
This is why Wiiv's Deepbranding always precedes the construction of the avatar. One cannot accurately document what has not yet been precisely defined. Deepbranding is the work of definition. The avatar is the work of transmission. The two are inseparable.
On Billy, the AI-ready brandbook is part of the Deepbranding deliverables. Before the first post was published, before the first product sheet went online, the avatar was operational. The launch was done with an AI already trained on the brand, not an AI that we tried to catch up with afterwards. This made it possible to maintain a volume and consistency of content impossible to maintain humanly with a small team.
What it concretely changes in daily life
A well-built AI brand avatar changes a brand's operational life very concretely, on four distinct levels.
For content production, the difference is immediate. Instead of starting from a blank page and editing for twenty minutes to get something that resembles the brand, you get a usable first draft in a few seconds. The AI doesn't produce generic content that you personalize: it produces brand content that you refine. The ratio between what the AI produces and what the human modifies shifts to 80/20 instead of 20/80. For a high volume of production, this means entire days saved each month.
For customer relations, tone consistency becomes possible at high speed. A growing team, changing service providers, periods of high activity where everyone responds urgently: without an avatar, the brand's voice gradually fragments. With an avatar, every response produced or verified by the AI follows the same rules, the same tone, the same limits. The customer who interacts with the brand in January and December has the same interlocutor.
For decisions, the avatar allows for autonomous filtering of opportunities. A proposed partnership, a creative direction submitted by a graphic designer, an influencer collaboration to evaluate: the AI can provide a structured initial opinion in a few seconds, based on the operational values and decision rules documented in the avatar. This is not a final decision. It is an initial filter that eliminates bad leads without involving the founder.
Regarding the founder's time, this is probably where the impact is most profound. Responding to customers, producing content, briefing service providers, evaluating opportunities: these tasks represent a huge portion of a founder's time during growth. A well-built avatar absorbs 80 to 85% of these, allowing the founder to concentrate on the 15 to 20% that require true human judgment. That is, high-stakes decisions, real crisis situations, opportunities that deserve in-depth strategic thought.
What the avatar does not replace
We won't lie to you: an AI brand avatar, however well-built, does not produce AI "as good as a human in 100% of situations." Selling that would be lying to you, and it would harm the credibility of everything that precedes.
What remains human, and must remain so, covers several specific areas. Fine contextual judgment: a human senses when a customer is about to leave for good versus when they are just making noise. The AI applies documented rules; it does not read the emotional subtleties of a situation. Creative improvisation: when an unexpected opportunity arises, a human associate will instinctively seize it. The AI stays within what it knows. Reading weak signals: an innocuous comment that announces a trend, a customer phrasing that reveals something new about the target audience. The AI doesn't pick up on it unless told to look for it. High-stakes emotional or strategic decisions: a real crisis, a positioning pivot, a decision that commits the brand's image in the long term. And the brand's subtext: what the brand says between the lines without ever explicitly stating it, what makes it magnetic, remains difficult to document and even more difficult to transmit to an AI with total precision.
What the avatar truly produces: an AI that manages 80 to 85% of operational situations excellently, frees up the founder for the 15 to 20% that require true human judgment, and never goes grossly off-topic. This is already transformative for a growing brand. And it's honest.
How Wiiv builds your brand's AI avatar
Of course, depending on the brand's maturity level and ambitions, the avatar's precision level is adjusted, but for the biggest brands, up to 850 parameters can be defined.
The construction of an AI brand avatar at Wiiv always begins with Deepbranding: 11 strategic steps that precede all design and content. Founder analysis, real target definition, positioning, tagline, brand essence, promise, manifesto, operational values, tone of voice, brand keywords, visual guidelines, marketing levers. Each step produces data that directly feeds the avatar.
This foundational work is what distinguishes an AI-ready brandbook built by Wiiv from a hastily produced document. You don't structure what isn't defined. The precision of Deepbranding is the condition for the precision of the avatar.
Once the foundations are laid, the avatar is built layer by layer. The 12 layers of knowledge, the 27 tone sliders calibrated with real examples, the mapping of situations with their parameters, the library of examples collected from the founder (the real ones, not invented examples), the visual guidelines translated into natural language, and the decision rules derived from the founder's actual behaviors.
The final deliverable is a structured document, manageable by Claude, ChatGPT, or Gemini, operational from day one. Not a decorative PDF. A brand database that the AI can consult before each production and in every interaction.
For existing brands that already have branding but no AI avatar, we start with an audit: what does the AI currently produce, where does it deviate, what parameters are missing. This audit helps prioritize the construction work rather than starting from scratch. To evaluate the current state of your branding, Wiiv's branding score is a useful starting point.
To discuss this with Cynthia and see if building an AI avatar makes sense for your brand, you can book a slot directly on her Calendly.
FAQ: AI brand avatar, AI-ready brandbook, and AI customization for your brand
What exactly is an AI brand avatar?
An AI brand avatar is a system of several hundred interdependent parameters that allows artificial intelligence to embody a brand with precision. It covers 12 layers of knowledge (identity, target, products, personality, vocabulary, situational tone, editorial territory, strategy, operational constraints, competitive context, history, decision rules), 27 tone sliders calibrated with real examples, a mapping of 50 situations with 17 parameters each, and a library of real content. It's not a prompt, not a tone of voice written in two paragraphs: it's a brand database that the AI consults before each production.
What is the difference between a prompt and a brand avatar?
A prompt is a one-time instruction. The AI executes it and forgets it in the next session. A brand avatar is a permanent context. The AI knows the brand, its tone, its target, its rules, its typical situations. It never starts from scratch. The practical difference: with a prompt, you edit for twenty minutes to get something that resembles the brand. With an avatar, you get a usable first draft in a few seconds that you refine in two minutes.
Why 27 tone sliders instead of a prose description?
Because "warm and professional" is an instruction that 80% of brands give to their AI. The AI cannot deduce what that specifically means for your brand. The 27 sliders, rated from 1 to 10, produce a unique tone imprint across 27 dimensions: register, relationship with the customer, relationship with humor, relationship with conviction, relationship with sales, relationship with proof. The combination of these ratings is impossible to confuse with another brand.
Why document failed examples in addition to successful ones?
Failed examples are often more valuable than good ones. They clearly define the boundaries of what the brand is not. This is where AI most often deviates when it doesn't have these guardrails. A post that didn't resemble the brand, a headline that we regret publishing, a customer response that sounded off: these examples give the AI precise limits that a list of abstract rules cannot produce with the same effectiveness.
How can AI guide visuals if it doesn't see images?
By translating the visual universe into usable natural language. Three types of information: precise sensory words describing the universe (raw, warm, mineral, airy, rather than "modern and clean"), explicit rules (pure white is never used as a background, never stock photos with smiling models), and references translated into textual characteristics (formats, density, palette, atmosphere). With these three layers, the AI can brief a graphic designer, evaluate a creative direction, or describe a visual to be produced within the brand's exact territory.
Does the avatar work with all AIs?
Yes, with nuances of usage. Claude, ChatGPT, and Gemini all process text documents comparably. Performance differences between models exist, but they are marginal compared to the quality of the document provided. A precise and dense avatar produces good results on any model. A vague avatar produces mediocre results on all. We tested all three on Billy and provide usage recommendations based on the application case.
How long does it take to build a complete avatar?
Deepbranding at Wiiv takes between two and twelve weeks depending on the project's complexity. This foundational work produces the data needed for avatar construction. Once the foundations are laid, structuring the avatar takes an additional one to two weeks. In total, between three and eight weeks for a complete and operational avatar. This is a time investment, not a quick fix. That's also why the result is radically different from an improved prompt.
Does the avatar need to be updated?
Yes. A brand evolves: its target refines, its channels change, its tone matures, its products transform. A static avatar will gradually describe an outdated version of the brand. We recommend a review at least once a year and with every significant pivot (major new product, change in positioning, entry into a new market). The good news: a partial update is much faster than the initial construction.
Are the 85% of interactions managed by AI noticeable?
When the avatar is well-built, no. On Billy, customers discovered afterward that some responses were AI-generated. Their reaction: surprise, not disappointment. Because the responses sounded like the brand, not like an AI. What is noticeable is AI without an avatar: the average tone, interchangeable formulations, correct but impersonal responses. With a complete avatar, the content sounds like the brand because it is produced from precise knowledge of the brand.
Does it make sense for a small, starting brand?
Paradoxically, this is where the avatar is most valuable. A small, starting brand doesn't have the resources for a full marketing team. With a well-built avatar from launch, a single person can maintain a level of content production and customer relationship quality that would normally require two or three people. We did exactly that with Billy: launched with an operational avatar, volume and consistency maintained with a small team from day one.
How do I know if my current branding is precise enough to build an avatar?
The simplest test: load your brand documents into an AI and ask it to write an Instagram post and a response to a dissatisfied customer. If both results sound like your brand, the foundation is there. If they sound like any brand in your sector, the foundations lack precision. Wiiv's branding score helps identify priority areas for improvement with a structured evaluation.
What is the difference between a classic brandbook and an AI-ready brandbook?
A classic brandbook is designed for humans. It contains mood boards, visual references, spacing rules, and examples of typography in context. An AI cannot process this information. An AI-ready brandbook is entirely written in natural language with explicit rules, textual examples, named prohibitions, and documented situations. The information density is maximized because that is the only criterion that matters for an AI.
Can the avatar help in making strategic decisions?
Yes, in a first-filter role. When presented with a partnership opportunity, a creative direction, or a positioning decision, the AI evaluates consistency with the operational values and brand territory documented in the avatar. It doesn't decide: it structures the analysis and identifies areas of tension with the brand identity. This allows the founder to make an informed decision rather than an intuitive one, and the team to filter opportunities autonomously.
Can an avatar be built for an existing brand without redoing all the branding?
Yes. Wiiv offers audit and structuring missions for existing brands. If the branding is strong but not documented in a format usable by AI, we work on translating and structuring what already exists. If the branding has strategic flaws, we recommend going further with a complete Deepbranding. In all cases, we start with an audit of what the AI currently produces to precisely identify what is missing.
How does the avatar manage crisis situations?
Crisis situations (bad buzz, negative article, series quality issue, misinformation) are among the best-documented situations in the avatar, precisely because they are the riskiest. For each type of crisis, we define the posture (whether to acknowledge or not, whether to speak publicly or not, whether to escalate or manage directly), the exact tone with the 17 situational parameters, what is never said in this situation, and an example of a perfect response. The AI does not manage the crisis instead of the founder. It prepares, structures, and proposes response elements within the brand's exact territory.
