In the first quarter of 2026, obesity and diabetes licensing commitments reached $22 billion, already surpassing the total for all of 2025. Pfizer completed its $10 billion acquisition of Metsera after a bidding war with Novo Nordisk. Roche invested $1.65 billion upfront to co-develop petrelintide with Zealand Pharma. These decisions were made at speed, in a therapeutic area with more than 100 active development programmes, where the difference between identifying an asset before a competitor and arriving second can be measured in billions. [1] The CI teams supporting those decisions are the ones that had moved beyond reactive landscape monitoring to proactive, scenario-grounded, AI-powered intelligence infrastructure.
At Pienomial, we built KnolAI as the life sciences competitive intelligence and enterprise intelligence platform that gives pharma CI functions and their consulting partners the shared infrastructure foundation of an AI Center of Excellence, without requiring a multi-year organisational transformation to access it. The AI CoE for CI is not a people and organisation project alone. It is a data and platform project: establishing the governed knowledge layer, the automated monitoring architecture, and the scenario intelligence capability that allow human CI analysts to operate at the strategic level rather than spending the majority of their week on data collection. This post explains how to build that CoE in a way that delivers value from the first sprint rather than from the conclusion of an eighteen-month change programme.[9]
1. What an AI CoE for Competitive Intelligence Actually Is
An AI Center of Excellence for competitive intelligence is not a team of data scientists who build bespoke AI models for CI use cases. It is a governed organisational and technology structure that enables the CI function to scale its analytical output, maintain consistent intelligence quality, and distribute insights to clinical, HEOR, regulatory, and commercial stakeholders systematically, rather than through ad hoc briefing documents that vary in quality and coverage based on individual analyst availability, whether those analysts are internal or part of a consulting engagement." [5]
A mature AI CoE for pharma CI centralises four functions that are typically fragmented across a CI function without CoE structure. First, governance: consistent policies for source selection, citation standards, data classification, and output quality review. Second, knowledge: a single governed intelligence layer rather than separate analyst-maintained databases and spreadsheets. Third, vendor and partner management: consolidated tool assessment and procurement, and a shared intelligence layer that consulting partners can access alongside internal teams without duplicating infrastructure." Fourth, capability development: career paths, training frameworks, and competency standards for CI analysts working with AI-assisted tools.[6]
Without this structure, most pharma CI functions end up with eight to twelve different AI and data tools, redundant monitoring efforts across teams, inconsistent citation and quality standards between senior and junior analysts, and institutional knowledge that sits in individual analyst files rather than in a shared, queryable system. [6] The CoE resolves each of these problems structurally rather than managerially.
2. The Diagnostic Question: Where Does Your CI Function Stand Today?
Before building a CoE, the honest diagnostic question is which of the five CI workflow maturity stages your function currently operates at, because the build path and the priority investments differ materially depending on the starting point.[1]
Stage 1, Reactive and ad hoc: Intelligence is produced in response to requests, from manually assembled sources, with no systematic monitoring between request cycles. The diagnostic question is: how long after a competitive event occurs does your CI team become aware of it?
Stage 2, Structured but manual: The function has defined source sets and monitoring schedules, but the monitoring and assembly work is manual and consumes most analyst capacity. The diagnostic question is: what proportion of CI analyst time is spent on data collection and formatting versus analysis and stakeholder communication?[3]
Stage 3, Automated monitoring, manual synthesis: The function has automated alert systems for specific signal types, but synthesis, scenario development, and stakeholder formatting remain manual. The diagnostic question is: does your CI function regularly produce forward-looking scenarios with named conditions and recommended responses, or does your output primarily describe the current competitive state?[1]
Stage 4, AI-assisted synthesis and scenario development: AI tools support both monitoring and synthesis, with human analysts focused on interpretation, framing, and stakeholder communication. The diagnostic question is: does your intelligence regularly reach the functions that need to act on it, in the format and timeline they need?[2]
Stage 5, Governed enterprise intelligence platform: A single governed knowledge layer serves all CI consumers across clinical, HEOR, regulatory, and commercial functions simultaneously, with institutional knowledge compounding over time and scenario intelligence informing portfolio decisions in real time. The diagnostic question is: is your CI function a strategic asset that influences major investment and development decisions, or a reporting service that responds to requests?[9]
3. The Three Architecture Decisions That Define the CoE Foundation
The technology architecture of an AI CoE for pharma CI rests on three foundational decisions that determine whether the CoE delivers compounding strategic value or becomes another layer of tool complexity on top of the existing fragmented infrastructure.[5]
Architecture Decision 1, Single knowledge layer vs multi-tool aggregation: The most consequential architecture decision is whether the CoE will be built on a single governed intelligence layer that all CI analysts and CI consumers query from, or on a collection of tools that individual analysts use separately and whose outputs are manually combined. The single knowledge layer approach is operationally harder to establish initially but delivers compounding value: every piece of intelligence added to the knowledge layer by any analyst becomes available to every analyst and consumer immediately, institutional knowledge compounds with every query rather than being siloed in individual files, and cross-functional consistency is structural rather than dependent on coordination.[2]
Architecture Decision 2, Continuous monitoring vs scheduled review: Manual CI programmes conduct scheduled source reviews: weekly or biweekly checks of trial registries, regulatory databases, and conference feeds. Continuous AI-powered monitoring is structurally different: sources are checked continuously rather than on a schedule, with alerts generated within hours of a triggering event rather than at the next scheduled review cycle. In a competitive landscape where a competitor's Phase III readout or a regulatory approval can shift strategic priorities overnight, continuous monitoring versus scheduled review is not a minor efficiency improvement. It is the difference between intelligence that allows proactive response and intelligence that arrives after decisions have already been made.
Architecture Decision 3, Closed proprietary intelligence vs open institutional knowledge: A CI function built on a governed knowledge layer accumulates institutional intelligence that is accessible to every authorised user — internal analysts, consulting partners, and cross-functional stakeholders simultaneously. Rather than each engagement starting from scratch, consulting partners working on active projects can draw from the same continuously updated knowledge base that the internal team uses, and their analytical contributions enrich the same layer. The value of the CI infrastructure compounds over time rather than restarting with each new engagement or team change. [2]
4. The Human-AI Division of Labour: What the CoE Automates and What It Preserves for Analysts
The most successful pharma CI functions recognise that optimal intelligence comes from human-AI collaboration rather than technology alone. The CoE must be designed around an explicit division of labour that automates the mechanical stages of CI production and preserves the high-value human contribution of strategic interpretation.[2]
What the AI infrastructure handles: Continuous source monitoring across clinical trial registries, regulatory databases, HTA assessment portals, conference abstract feeds, patent filing databases, trade press, and company communications. Automated signal classification by asset, indication, and event type. Deduplication across sources. Structured alert generation with sourced claims. Evidence table generation from extracted intelligence. Version-controlled knowledge base updates when existing intelligence is superseded by new developments.
What human CI analysts handle: Strategic interpretation of what a signal means for the organisation's specific portfolio and competitive position. Competitive framing: placing an intelligence finding in the context of the organisation's development strategy. Stakeholder narrative: translating an intelligence finding into the specific decision implication for a clinical development team, a HEOR team, or a board portfolio review. Red team analysis: building the strongest possible case that a competitor programme will succeed, even when the organisation's internal teams are invested in the opposite conclusion. Consulting partners often lead this work, applying independent perspective that internal teams find difficult to generate when they are close to the programme.[3]
This division of labour is not a temporary arrangement pending further AI capability development. It is the permanent optimal design for a pharma CI function: AI handles the data volume and monitoring frequency that human capacity cannot match, and human analysts handle the contextual judgment and stakeholder relationships that AI cannot replicate.[2]
5. Building the CoE Governance Framework
A pharma CI CoE without a governance framework will replicate, at a larger scale and higher cost, the inconsistency problems of the fragmented CI function it was designed to replace. The governance framework addresses five operational domains that determine whether the CoE delivers consistent, defensible intelligence or variable-quality outputs depending on which analyst produced them.[6]
Source governance: A formally maintained source taxonomy covering which source types are included in each CI programme scope, the quality tier assigned to each source type, and the update and validation frequency applied to each. Analyst-selected sources must be added to the formal taxonomy before their outputs are treated as CoE-quality intelligence.
Citation standards: All intelligence outputs must carry claim-level source attribution linking each specific claim to its primary source, not document-level reference lists. This standard satisfies both internal defensibility requirements and the expectations of senior stakeholders who will challenge the evidence behind a CI recommendation.[1]
Output quality standards: Defined criteria for what constitutes a CI output ready for stakeholder distribution, covering source verification status, claim attribution completeness, strategic framing depth, and human review completion. Outputs that do not meet the standard must be flagged as preliminary rather than distributed as completed intelligence.
Data classification: CI intelligence covering unpublished development strategy, pre-announcement competitive information, or material non-public intelligence must be classified and access-controlled appropriately. The CoE governance framework must include data classification procedures that prevent inappropriate distribution of sensitive competitive intelligence.[9]
Performance metrics: The CoE must define and track measurable performance indicators that demonstrate its strategic value: average time from competitive event to CI team awareness, proportion of analyst time spent on analysis versus data collection, proportion of CI outputs that influenced a documented strategic decision, and stakeholder satisfaction with intelligence timeliness and depth.
6. Cross-Functional Distribution: Making CI Intelligence Available to Every Function That Needs It
One of the most significant strategic failures of pharma CI functions is that the intelligence they produce is distributed to a narrow set of recipients, typically the CI team's direct stakeholders, rather than being available to every function whose decisions it should inform. Clinical development teams designing Phase III protocols need to know what HTA bodies have required of analogous products. HEOR teams building submission evidence architectures need to know what evidence the competitive landscape provides to inform indirect treatment comparisons. Market access teams preparing payer negotiations need to know what commercial positioning competitors have used and what access outcomes they have achieved.[7]
A well-structured AI CoE makes this cross-functional distribution structural rather than dependent on the CI team proactively sharing specific briefs with specific functions. KnolAI's shared knowledge layer means that any authorised user across clinical, HEOR, regulatory, and commercial functions can query the CI intelligence directly, in the format their specific function requires, without requiring the CI team or their consulting partners to produce a bespoke deliverable for every cross-functional request. The single source of truth established by leading pharma CI CoEs ensures consistent information access across the organisation while preserving institutional knowledge, a significant evolution from the siloed approaches that characterised most CI functions before the AI CoE model emerged.[2]
7. Measuring CoE Maturity: The Five Dimensions That Separate Leaders From Laggards
Seven global pharmaceutical companies, including AstraZeneca, Merck, Eli Lilly, Novartis, Sanofi, Novo Nordisk, and GSK, rank among the top 100 most AI-mature firms in the 2025 IMD AI Maturity Index. The index evaluates AI maturity across five dimensions: executive support, technology and infrastructure, operational excellence, workforce and culture, and ethics and risk management. [4] For a pharma CI CoE specifically, these five dimensions translate into a practical maturity assessment framework.
Executive support: Is CI intelligence formally embedded in portfolio review, clinical development governance, and commercial strategy processes? Or is it consumed informally at the discretion of individual stakeholders? CoE maturity requires formal CI intelligence integration into decision-making processes, not just availability of CI outputs.
Technology and infrastructure: Does the CoE operate on a single governed knowledge layer with continuous monitoring, or on a collection of separately subscribed tools with manually reconciled outputs? The infrastructure maturity gap is the most common barrier between Stage 2 and Stage 4 CI function capability.[6]
Operational excellence: What proportion of CI analyst time is spent on analysis and interpretation versus data collection and formatting? Is the average time from a competitive event to CI team awareness measured in hours or weeks? Are CI outputs consistently sourced at the claim level?
Workforce and culture: Do CI analysts have the AI literacy and prompt design skills required to work effectively with AI-assisted intelligence tools? Is there a formal competency development path for CI analysts moving from manual to AI-assisted workflows?
Ethics and risk management: Is the CoE governance framework active and enforced? Are data classification and access controls appropriate to the sensitivity of the intelligence being handled? Is the source verification standard consistently applied to all CoE outputs?[5]
8. The Scenario Intelligence Capability: The Most Valuable CoE Output
A diagnostic framework published by Clarivate in 2026, examining the five stages of pharma CI workflows, identifies scenario intelligence as the most valuable, and most commonly missing, capability in pharma CI functions. The diagnostic question for this capability is precise: does your CI function regularly produce forward-looking scenarios with named conditions and recommended responses, or does your output primarily describe the current competitive state? [1] Most CI functions, even technically mature ones with good monitoring infrastructure, operate primarily as landscape designers rather than as scenario intelligence providers.
The AI CoE model enables scenario intelligence at scale because KnolAI's continuous monitoring and synthesis capabilities provide the evidence base, and KnolPersona's assessor simulation provides the stakeholder perspective grounding, that make scenarios defensible rather than speculative. Scenarios are only as defensible as the data they are built on. A portfolio committee that asks 'How confident are you in this?' needs the CI team to point to primary sources: patent expiry timelines, trial design records, regulatory precedent. The quality of those sources determines whether a scenario is treated as a strategic input or a speculative one.[1]
9. How Fast Can Your Team Build an AI CoE for CI with KnolAI?
Building a genuine life sciences competitive intelligence CoE does not require a multi-year transformation programme before the organisation sees value. KnolAI's pre-built intelligence infrastructure means the governed knowledge layer, continuous monitoring, and structured briefing capabilities are available from Sprint 1, not from the conclusion of an infrastructure build project.[9]
Sprint 1, Weeks 1 to 2, Continuous monitoring and knowledge layer live: KnolAI is configured for your competitive intelligence scope: therapeutic areas, competitor sets, signal source types, and alert priority classifications. Continuous monitoring is activated across clinical trial registries, regulatory databases, HTA assessment portals, and conference feeds. The first automated intelligence alerts are delivered to your CI team within days. The knowledge layer begins accumulating validated intelligence from the first monitoring cycle.
Sprint 2, Weeks 3 to 4, Cross-functional distribution and governance framework established: Access controls and output quality standards are configured for your CI CoE. Cross-functional access is established for clinical, HEOR, regulatory, and market access teams with role-appropriate intelligence views. The governance framework covering source standards, citation requirements, and data classification is documented and activated.[6]
Sprint 3, Weeks 5 to 6, Scenario intelligence capability and CoE metrics live: KnolPersona scenario simulation is configured for your primary competitive decision contexts. The first CoE performance metrics are tracked: competitive event to awareness latency, analyst time on analysis versus data collection, and cross-functional query volume. The CoE is operational as a governed enterprise intelligence platform delivering strategic CI capability to every function in the organisation from this sprint forward.[1]
Conclusion
The pharma CI functions that supported the most consequential portfolio decisions in 2025 and 2026, including the deals and acquisitions reshaping the obesity, oncology, and rare disease landscapes, were not the ones with the most analysts or the most database subscriptions. They were the ones that had moved from manual landscape description to governed, AI-powered, scenario-grounded intelligence infrastructure: a single source of truth for the organisation's competitive knowledge, continuous monitoring rather than scheduled reviews, and scenario intelligence that stakeholders could defend in a portfolio committee with primary source confidence.
At Pienomial, we built KnolAI as the life sciences competitive intelligence infrastructure that enables any pharma CI function to operate at this level, regardless of team size or existing technology investment. The AI CoE for pharma CI is not a future aspiration. It is available now, and it is what separates the CI functions, and their consulting partners, that inform the decisions that matter from the ones that report on decisions that have already been made. . [9]
CTA: See how KnolAI powers an AI Center of Excellence for pharma competitive intelligence. Book a demo with the Pienomial team today.
