Accenture and Bristol Myers Squibb (BMS) have come up with “Mosaic”, a medical content knowledge hub empowered by generative AI solutions and established in Mumbai, India, and the focus is to change the manner at which BMS and other pharmaceutical companies produce, adjust, and distribute medical or marketing content to healthcare professionals (HCPs). The concept is simple: “AI + great content operations should help shift a medical-marketing content operation from a one-size-fits-all segment into a faster, smarter, and personalized segment without compromising on the elements of control and scientific accuracy.”

Following is the “everything you need to know” long description of what Mosaicism is, and then the implications thereof, the likely mechanisms involved, the changes that will occur among doctors and patients, and the risks involved.

1) What is the essence of “Mosaic”

“Mosaic is defined as a novel, end-to-end, AI-assisted medical content platform that utilizes GenAI and supplemented digital capabilities in:”

Analyze the needs of physicians in learning.

Develop patient- and HCP-centric content faster

Support the overall commercialization strategy of BMS in the international market (that is, enhancing the way in which company information related to products, diseases, and values is communicated to healthcare professionals)

Think of it as a mix of:

a creative studio,

“a medical/scientific writing factory

“a digital personalization engine”,

and a governance/compliance workflow—

. all supported by Gen AI and executed with the process rigor of a major services partner (Accenture).

Significantly, the promotion of Mosaic is not being offered in the form of “AI replacing medical writers.” Instead, the emphasis of the introduction of the AI model, called Mosaic, concerns its role in accelerating the medical writing pipeline, which includes discovering what content needs to be produced, developing variations, working on channel adaptations, and accelerating

2) Who launched it, where, and why is this significant?

The launch
The mosaic was formally inaugurated in Mumbai by:

Adam Lenkowsky, Chief Commercialization Officer | Bristol Myers Squibb

Ndidi Oteh,
Global CEO,
Accenture Song

Pharam

1

Why Mumbai?

BMS currently already has a established presence in India, and it has been increasing its large capability centres in the country (also a large centre in Hyderabad as cited from coverage). To date, India offers:
deep digital talent pools (creative, analytics, engineering, content ops)
scalable operations
cost advantage

speed and “always-on” global support opportunities

So, Mumbai will become the “hub” that can support global teams, not just the Indian one.

3) What problem does the company Mosaic aim to solve?

Pharma content writing is famously difficult since it needs to:

*medically accurate

legally compliant,

consistent with approved labeling and claims,

tailored to different audiences and different channels,

updated as new scientific knowledge emerges.

The repositories are approved via Medical/Legal/Regulatory reviews

Pain Points Faced by Many Pharmaceutical Companies Today:

Slow turnaround time: it may take weeks/months from concept to finishing.

Content overload: “thousands of assets across brands, regions, indications.”

Low reuse: Instead of teams reusing content, teams reuse content in a different or lower level of

Poor Personalisation: HCPs receive generic information that does not relate to their environment.

Channel mismatch: the content does NOT address the channel in which the professionals are active. Channel mismatch occurs when the content is

Mosaic has the goal of addressing these challenges by leveraging the power of AI and Content Operations to make content:
faster to produce,
easier to adapt,
more targeted,

more useful to clinicians,

and is consistent with governance.

The Week

4) What does the “AI-based medical content hub” entail?

While this element isn’t detailed in typical short articles, based on how pharmaceutical content hubs function (and how Mosaic has been described), here’s how Mosaic could plausibly contain elements such as the following.

“A) Real-time insight into HCP needs

“The hub hopes to recognize physician education needs in real time.”

This can involve the following analytics signals:

which topics HCPs are searching for in portals,

engagement metrics of emails/webinars,

questions being asked to medical science liaisons (MSLs),

• feedback from events and field teams,

performance data regarding the content assets (opens, time on page, click paths),

region/segment patterns (such as oncologists compared with general practitioners).

“Need detection” has become essential since the most successful content will always address what HCPs require at this point in time, rather than the original plans that were developed through marketing 3 months ago.

A) Business model framework design (“shoe box model”)
B)

For personalized content at scale, today’s hubs might turn those messages into a modular library:

“`php
<?php
$is_a_person = true;

approved claims,

scientific references,
24

safety language,

disease education modules,

visual components,

patient journey explanations,

Dosing/administration instructions, if applicable

Then, instead of creating brochures, teams assemble from modules. The assembly of the brochures could be aided by GenAI, although this is the task of human and MLR.

C) Gen AI-aided drafting and adaptation

Gen AI may expedite:

first drafts of educational materials,

semantic rewrite for clarity at different levels,

summarizing complex clinical information into a format friendly to the clinician,

adapting content to multiple channels (email newsletter, slide presentation, infographic text, brief HCP portal message),

Developing variations for specialties and contexts of practice.

“The big value: speed + volume + variation, while still requiring human review.”

D) Best practices in governance and workflows

In pharma, “governance equals everything.” Mosaic must be “end-to-end” to include at least these:

version control,

audit trails,

reference management,

claim substantiation,
controlled vocabularies

mandatory safety statements,

MLR Routing and Approvals,

Digital Asset Management(DAM).

However, the useful applications of GenAI can be possible only if it can operate within certain “guardrails” – approved content, traceable sources, and an efficient approval process.

5) What does Accenture bring to the partnership?

In Accenture’s role here, the work described is not merely “IT support.” The coverage closely links the relevance of Mosaic to the leadership of Accenture Song and mentions the extensive team of creatives and technologists.

Accenture probably adds:

Content operations expertise: “Factories” for scalable content production and quality assurance

Experience design: designing content experiences and channels

AI engineering: working on GenAI workflows, prompts, retrieval systems, guard rails

Platform integration: linking CRM, marketing automation software, DAM systems, and compliancy systems

Analytics and Measurement: Proving “What Works” and Iterating
Analytics has been

What is also worth mentioning is that BMS and Accenture are said to maintain a long-term partnership, which began with traditional services and now encompasses digital change.

6) How is it helpful to clinicians and patients?

“This is where ‘why’ trumps ‘tech’.”

For Clinicians (HCP

If the Mosaic positively impacts the healthcare community, healthcare professionals could expect the following

more timely updates (such as summary delivery of current evidence).

More relevant education (education with a focus on their specialty and interests).

less noise (fewer generic blasts, more targeted communications),
more appropriate format for the content (short for mobile, in-depth when required).

In a world filled to the brim with clinician workloads, “right message, right time, right format” is a huge improvement.

Patients

Patients don’t directly access pharma HCP marketing. However, the patient indirectly benefits if:

clinicians comprehend therapies better,
Adolescence is the

that can help to improve understanding, that are more available, and that

care pathways and adherence messages are enhanced,

Misconceptions decline.
The angle given in these publications about Mosaic is “patient centric content” and “enhancing the patient experience” with easier communication plan development and dissemination.

7) For what reason is BMS so concerned about AI within marketing and commercialisation?

Because: Commercialisation is where:

patient access efforts meet real-world adoption,
education can prevent improper use as well as facilitate proper use.

competition is intense (particularly in oncology, hematology, and immunology),

and online platforms are driving this engagement.

News outlets have also mentioned Mosaic’s connection with BMS’s strategic investment of $130 million in order to enable and enhance various AI initiatives and capabilities within marketing. Here, Mosaic appears as a part of this larger initiative.

“So Mosaic isn’t just a random pilot—it’s part of a strategic investment theme: using AI as a means to achieve speed, scale, productivity, and better decision-making.”

8) What “hyper-personalised content” means in pharma (and what it should NOT mean)

Some reports include expressions such as “hyper-personalised experiences.”

“That can sound pretty frightening, so it is a good idea to acknowledge that

What it can mean (Legit Use)

providing a list of suitable topics for the specialist to choose from, and

tailoring detail (brief overview vs. detailed clinical information),

format (video synopsis versus power point),

tailoring based on practice setting (large hospital vs. smaller clinic),

tailoring by engagement history: if the person has ever requested information on dosing, offer them the dosing administration resource.

What it SHOULD NOT Mean (high risk)

on the basis of sensitive personal data of specific patients.

manipulating prescribing behavior through opaque targeting,

making unsubstantiated claims,

the production of “AI persuasion” rather than evidence-based educational content.

A reputable pharma AI content platform would aim to enhance its relevance and usability, keeping strictly within the promotional or medical guidelines.

9) Major risks and the way a hub like Mosaic should tackle them

Every pharmaceutical content engine enabled with Gen AI faces serious challenges:

Risk 1: Hallucinations or Misrepresentations

“A GenAI can create convincing but false statements. In pharma, this is not acceptable.”

Mitigation

Based on authorized references, generation through retrievals,

hard constraints: approved claims and/or approved data only,

mandatory human medical review,

automated checks against a claims database.

Risk 2: Violation of Compliance/Regulatory Requirements

Promotional statements have to comply with labeling and local regulation requirements.

Reduction:

embedded MLR workflows,

Local adaptation rules,

controlled templates,

audit logs and approvals.

Risk #3: Bias and imbalance of information
*

AI can put too much weight on some endpoints or fail to adequately show the safety information if it is not designed well.

Mitigation

balanced templates which always contain the obligatory safe data.

governance on model outputs,

monitoring consistency of the content.

Risk 4: Data privacy and security

Insight engines can consume engagement data; such systems require protection of the data.

Mitigation

strict access controls,

anonymisation as appropriate,

compliant data handling policies.

Again, these specifics are not included in the public declarations, though they represent basic needs within any meaningful biopharma content process enabled with AI.

10) Why this announcement matters beyond BMS and Accenture

Mosaic is an important indication that there is a significant industry shift:????

Pharma is shifting from “digital assets” to “digital engines.”

“Building on building content, but building systems that learn what is required on a continuing basis and produce it at a faster rate.”

“Gen AI is becoming operational, not experimental.”

The establishment of this hub generally implies investment, process redesign, and commitment. India is becoming a global hub for premium AI-driven operations. It’s not just back office tasks. It’s not just data analysis. It’s not Marketing + medical content is becoming more dynamic. It has the potential to enhance the quality of education if implemented responsibly. 11) What success would look like – practical KPIs In order for the performance of Mosaic to be measured on a real-world basis, the following factors could be taken as typical indicators of success: Shorter cycle times for content approval mean a shorter time from the initial request to the approved Cost per asset: doing more and more with the same or lower costs As the number of projects increases, there is a likely increase in the number of projects being MLR Efficiency: Reduced cycles of rework because of improved first drafts HCP engagement uplift: increase in open rates, time spent on content, and event attendance Field adoption: ease of content use by sales/MSL groups Quality indicators: correction rate decreases, noncompliance problems decrease, consistency improves “Teach Back”
Many journals are considering a novel type of peer-review activity, which we 12) Bottom line Mosaic is the effort of BMS and Accenture to “industrialize” the development of medical and business content leveraging AI, using GenAI to read physician needs and produce and refactor communications faster and in a more personalized way, all out of a dedicated innovation hub in Mumbai. It is part of BMS’s overall use of AI in marketing capability, and a trend in the pharma industry as a whole, to use AI not just in drug development, but in the development of health-related information itself.