Add us as a preferred source
Narayanan Ramaswamy, head of advanced drug development at Tata Consultancy Services, explains how AI agents are reshaping drug development
Drug development relies on thousands of clinical, scientific and operational decisions being made accurately and on time. Study protocols need to be drafted and reviewed, study sites prepared, data checked, safety information assessed and submissions assembled. When these activities become slow or fragmented, this delays the time taken for life-saving treatments to reach patients.
AI agents are emerging as one way to reduce this friction. Unlike standalone tools, they can support defined activities across connected workflows, helping teams manage information, complete routine tasks, and respond to changing trial conditions. Their role is not to replace clinical, scientific, or regulatory expertise. It is to give that expertise more capacity by reducing the administrative and coordination burden around it.
Current industry data indicates that 75% of clinical trial protocols require at least one substantial amendment. Each amendment can create a ripple effect across documentation, sites, systems and stakeholders, adding time and complexity to an already demanding process. As life sciences organisations look to make development more resilient and scalable, the question is no longer whether AI can support isolated tasks, but how it can be used safely across the wider R&D workflow.
The operational work behind every clinical milestone
For healthcare professionals working in clinical development, the delays that matter most are often the ones that accumulate quietly in the background.
Protocol development is a clear example. Before a study can begin, teams may work through multiple drafts, reviews and alignments across clinical, regulatory, medical, safety and operational stakeholders.
The same applies to study setup. Sites need to be configured, data needs to be prepared, and teams need to coordinate across multiple systems and geographies. Submission preparation adds another layer of complexity, with large volumes of information needing to be turned into precise, compliant artefacts.
In each case, the challenge is not a lack of expertise. It is the amount of repetitive, document-heavy work required to keep everything moving.
AI agents can help address this challenge by taking on repeatable tasks within clearly defined parameters. They can interpret documents, draw together relevant information and work across digital systems, while operating under controls designed for regulated environments.
In practice, this could include supporting protocol drafting, clinical data processing, the creation of submission artefacts, monitoring activities and the identification of patterns across clinical and operational data. This reduces the manual effort required to prepare documents and helps teams surface relevant information faster.
AI cannot simply be inserted into a clinical process and left to operate without direction. In drug development, every output must be reliable, traceable, and capable of standing up to scrutiny. Effective use therefore requires a clear context of use, defined quality standards, human supervision, and lifecycle controls.
This is where an AI agent hub can play an important role. For example, our agentic AI platform is designed to support agentic AI across pharmaceutical R&D workflows, including clinical data management, study build and safety case processing. Its focus on role-based access, auditability, and a Human + AI operating model reflects the wider requirement for AI to operate within established governance frameworks, rather than outside them.
For clinical teams, this means retaining responsibility for the decisions that require professional judgement. AI agents may support the execution of defined tasks, but people must continue to set direction, review exceptions and remain accountable for quality, compliance and patient safety.
Making more space for scientific judgement
Today, many scientists and clinicians spend much of their time on preparatory and administrative work, such as analysing datasets, standardising them, reviewing reports and managing handoffs. These are essential tasks, but they do not always make the best use of highly trained expertise.
By taking on repeatable work, AI agents can free people to focus more on judgement, insight and decision-making. That means more time can be spent on scientific discovery and identifying risk and opportunity, and less on moving data through the process.
There is also a broader workforce implication. Life sciences organisations are under pressure from talent shortages, rising demand, and increasing regulatory complexity. A trusted AI workforce can help create more elastic capacity, especially in areas where expert time is limited, and demand can spike unexpectedly.
That does not reduce the importance of clinical teams. It makes their role more valuable. As AI takes on more routine execution, human expertise becomes even more central in setting direction, reviewing exceptions and making the decisions that shape trial quality and patient outcomes.
Pharma has invested heavily in connected data and digitising clinical research. AI agents offer an opportunity to build on those foundations by linking operational activities more intelligently across the R&D environment.
Their value will lie not in acting as an isolated technology layer, but in supporting a governed digital workforce with agents completing specific, repeatable tasks, while people provide the scientific, clinical and regulatory judgement that remains essential.
If implemented with the right controls, this model could help life sciences organisations reduce avoidable operational delays and scale development activities with greater confidence, creating a more responsive and resilient route to delivering the essential treatments patients need.
