Three Languages, Three Identities

I always had an identity fragmentation problem.  Since my career has been Japan centric since 98, an International presence (somewhat) since I joined Adobe in 2013, I have media presence in Japanese, English and Marathi.

Internationally am Manish Prabhune

In Japan マニッシュ プラブネ 

Marathi, and I am मनीष प्रभुणे

With each language my personality also is different. ChatGPT thinks

1. マニッシュ プラブネ equals 

• Author, Adobe, インド式計算

2. Manish Prabhune equals 

• Director Sales, Adobe, Vedic Mathematics, PokemonGO 

3. मनीष प्रभुणे equals

• Photographer, Marathi Blog

A Real World Prompt Example

In Gemini I typed the following prompt.

• Blue Note Tokyo Manish Prabhune : Returns nothing

• Blue Note Tokyo マニッシュ・プラブネ : Just changed the name to Katakana Japanese. Returns result that I facilitated a Fireside Chat for Verizon Business Group executives Mike Cingari and Nitin Ahuja in 2022 at the location! It was one great location that Adobe chose. Legendaries take the stage there!

Japanese Name and English name did NOT resolve

The article is in Japanese and it referes to me by my Japanese Katakana name. When the LLM took the english name, it drew a blank, when I gave it the Japanese name, it located the news item and also resolved the name in English. Net-net the point is if you use inconsistent and fragmented identities you are essentially causing the AI to hallucinate. Not its (AI's) fault. Rather this is where LLMs are helping us, DISCOVER these gaps. The fragmentation of our brand and the messaging.

Fortunately I have a photograph to prove that I was working on that day at Blue Note Tokyo., Adobe Experience Makers Live 2022. By writing this blog, am sure that now in some days time when indexing will happen, the prompt will also say that I presented on the stage at Blue Note Tokyo ecen if I prompt with English name.

Name Address Phone (NAP) Consistency

In the above example, the impact is IDENTITY DRIFT which is a real problem brands face where just a NAME, relocated ADDRESS, an invalid PHONE inconsistency can lead to a massive visibility drift. In SEO this is as we know NAP or Name Adress Phone consistency is critical for consistent information delivery to consumers. But this was why the NAP Consistency solutions were born.

Wrong messaging can dilute a brand

Outdated phone numbers, even addresses (in case of relocation) are a nightmare for businesses since these data points percolate beyond the brand website into Online Forums, Google Maps, Yahoo Answers and what not. Keeping a track of all that and resolving them with one click change is where NAP Consistency solutions, were popular where it could update all external services. That still does not cover brand messaging gaps. They still exist, and the reasons can be multiple. A very simple Automotive customers problem is below

Use Case : Global Brand with multilingual and multiple websites

As an example, in automotive comapnies usuallt have global presence, hundreds of sites and multple languages, the messaging "feels" all over the place. The primary message can be something like SDV (Software Defined Vehicle) or EV or maybe just DESIGN. They want to understand if all dealer portals are leading to the same messaging or causing a drift. Now, with Adobe Brand Visibility they can now audit across languages and sites. You can use ABV to run prompt testing across different language sets. For example,

• The Test: Query ABV’s dashboard for English prompts ("Best passenger EV car for a four person family"), Japanese prompts ("4人家族に最適な乗用EV"), and Spanish prompts ("El mejor coche eléctrico de pasajeros para una familia de cuatro personas")

• The Goal: Check share-of-voice data to ensure the LLMs are successfully associating the regional names with your global brand metrics, rather than treating them as disconnected local competitors. You can setup Adobe Brand VIsibility to also check on each individual market to check drifts due to linguistic nuances within a particular market (e.g. EN and JP version of sites inside Japan) 

• Solution: ABV flags a gap in Japanese AI responses, it tells the marketer that the Entity-to-AEM data integration for the Japanese profile needs more local schema authority signals. It also shows external sites responsible for this drift through the Opportunties Tab in Adobe Brand Visibility, "EARNED" media and not just your OWNED ones. 

Another example : Preventing the "drift" in identity with AEM Content Fragment and Adobe Brand Visibility

Step 1 : The "Dealership Address" Use Case

The Multi location automotive brand also wanted to make sure that LLMs do not treat local dealer (e.g. in Naka Meguro) referenced online as  AutomotiveCompany, 中目黒 as an isolated unverified entity, but the authorized dealer for the group/corporate brand.

Step 2 : Setting up the JSON in AEM

Location metadata on the dealers landing page hosted on Adobe Experience Manager (AEM) utilizing content fragments needs to be done as per the below JSON. This helped the company eliminate reliance on third-party location platforms and publish structured local landing pages. This helps AI search engines recognize authorized dealer of the global AutomotiveCompany while preserving local and global naming context.

Step 3 : Validate and Monitor Visibility with Adobe Brand Visibility

Once AEM serves the schema, Adobe Brand Visibility (ABV) provides the automated validation loop by testing registered prompts in its Prompt Library (e.g., "Where is AutomotiveCompany in Nakameguro?" or "AutomotiveCompany Meguro phone number") across generative AI engines like ChatGPT, Perplexity, and Copilot. ABV tracks LLM outputs to verify that AI models correctly attribute the local dealership to the global brand, cite the local AEM landing page URL, and parse the underlying schema without hallucinating contact details. If an AI model fails to establish the parent-child relationship, Adobe Brand Visibility flags the representation gap so teams can optimize the AEM content.

So what have I been doing this past four months to resolve my own Fragmentation

The reason why I have started the website www.manishprabhune.com is to essentially become an anchor to resolve all these fragmentations. With one article like this I can now make sure that Manish Prabhune and マニッシュ プラブネ at Blue Note Tokyo are the same person.  What I want to achieve is below!

LinkedIn and Controllable Sources

For this I have gone to LinkedIn and informed it my hybrid identity as below, The Japanese name becomes the integrator. It's not perfect, but with all the circle of influence parts (Adobe Blog, Press coverage as much possible), 

Updating and cross referencing my media presence

I have started asking them to use both my Japanese and English names in brackets so that AI can find these footprints and resolve them as much possible. Maybe when I gave my prompt with Japanese name at the start of this article, it resolved it to Manish Prabhune my English name with one of these data points. Today we have to look at content as data entities, not just the UI UX, look and feel. See one more example of the SendenKaigi Event I did last month. Making it sure that both my Japanese and English names enter the content structure