What AIforPharma is, in long form.
A canonical description of the consortium: who it is for, what it is, what it is not, and who is convening it. Written by the organising team and dated, so the version of record is unambiguous.
Two groups of senior professionals, working on the same problem, in separate rooms.
India is the third-largest pharmaceutical producer in the world by volume. It has more than three thousand pharma companies and supplies roughly twenty percent of the global generics market. It is also, by widely cited industry observation, mid-way through its AI adoption: somewhere near half of Indian pharma is still at proof-of-concept stage as of 2024 to 2025.
The gap between where the industry is and where it needs to be is large, and closing. It exists, in part, because the two groups of professionals who could close it operate in separate rooms. Pharma people do not fully understand what AI can do for their specific function. AI and technology people do not fully understand what pharma professionals actually need. The result, observable everywhere else, is slow adoption, bad implementations, and significant wasted potential on both sides.
AIforPharma exists so that conversation happens in a single room, deliberately, on a schedule.
A recurring industry consortium. Senior, curated, by application.
AIforPharma is a recurring consortium convening senior decision makers from India's pharma industry with the technology companies and AI engineers building for them. It is being convened in 2026. The intended cadence is three to four editions per year, alternating between Indian cities. Each edition is single-day, single-room, and small enough that the room can recognise itself: fifty to ninety senior attendees, in person, under a closed Chatham House rule.
Attendance is by application. Both pharma and technology applicants go through the same review by the organising team. Accepted applicants are extended a formal invitation along with the date, venue, and agenda. Members of the private group continue the conversation between editions; the group is moderated, off the record, and joined after a first edition.
The consortium is in its founding period. The first edition has not yet been held. Applications are open and reviewed on a rolling weekly basis. Specific organisations get named publicly in edition recaps, with permission, after they have actually been in the room.
Five things we explicitly are not.
- 01
Not a conference.
Conferences are one-off, broadcast-shaped, and measured by attendee count. AIforPharma is recurring, conversation-shaped, and measured by who is in the room and what they argue about. There is no main stage, no plenary keynote sponsor, no badge-scanning hall.
- 02
Not a product showcase.
PharmaOS, the product built by the organising team, gets demonstrated when relevant. So do other systems. Vendor pitches are not the format. The room can tell the difference between a demonstration of a working system and a sales presentation, and the format favours the former.
- 03
Not an open or paid event.
Attendance is by application and by invitation. There is no general ticket on sale. Pricing for accepted applicants is announced privately when the date and venue are confirmed; the intent is to keep the room senior and the conversation honest, not to monetise the event.
- 04
Not a webinar series.
Editions are in person, single-room, and deliberately small. The conversation that makes the consortium useful does not happen in a Zoom rectangle. Recordings, when sessions warrant them, are shared with attendees only.
- 05
Not a trade association.
We do not lobby, certify, or publish industry standards. We convene a specific conversation between pharma and AI on a schedule. Organisations with overlapping mandates (NASSCOM, IDMA, OPPI, IPA) are peers we are happy to coordinate with, not bodies we replicate.
The seats we are designing the invitation for.
From pharma, the audience the consortium is built for is the function leadership of operating pharma companies in India: Chief Information Officers, Chief Digital Officers, Heads of R&D, Heads of Regulatory, Heads of Medical Affairs, Heads of Pharmacovigilance, Heads of Commercial, Heads of Manufacturing and Quality. The kind of company we are designing the invitation for is the top forty by domestic revenue and the next twenty by AI investment ambition. The room is who is in it; we invite by role and contribution, not by logo.
From technology, the audience the consortium is built for is the senior leadership at the foundation-model companies, the enterprise cloud platforms, the data and observability vendors, and the pharma-focused AI startups already building for the industry. Same rule: roles, not logos. A founding engineer with a real system shipping into Indian regulatory teams is more interesting in this room than a logo without a workflow.
A secondary, important audience: the working practitioner, the R&D scientist, the regulatory manager, the AI engineer at a CDMO. They are who the public site is also written for, even when they do not yet have the seniority to apply, because they read it and they decide whether to forward it to their leadership.
Single day. Single room. Four working sessions, two demonstrations, one dinner.
An edition runs for a single day in a single room. The intended shape is four working sessions on R&D, regulatory, commercial, and medical affairs respectively, plus two live demonstrations of AI applied to actual pharma problems. The evening before the edition has a curated dinner for attendees.
Sessions are working sessions, not panels. Each one is opened by a pharma function leader naming the workflow they are trying to improve, paired with an engineer or AI leader who has built or is building for that workflow. The room takes the conversation from there. We hold the format under a closed Chatham House rule: attendees are free to use what was said, the speakers are not attributed publicly without permission.
Demonstrations are real systems against real data, not slideware. Failures are discussed as candidly as successes. PharmaOS is regularly one of the demonstrations; it is never the only one. We have a standing preference for systems already shipping in production over systems that are interesting in theory.
The edition is the highlight. The conversation is the product.
The most useful version of this conversation does not start and end on the day of the edition. Between editions, the consortium will live in a private group of attendees. The intent is straightforward: a question on RAG over a regulatory document set gets answered by the person who built the retrieval pipeline at a peer company. A new CDSCO guidance lands and the group parses it together in real time. The next edition's agenda is shaped by what the group has been arguing about for three months.
Membership in that group is extended to attendees after their first edition. It is moderated, off the record, and small enough to recognise itself. The group does not exist yet; it begins with the founding edition.
The team behind PharmaOS, who were already in these conversations.
AIforPharma is organised by the team behind PharmaOS, a Super AI product that lets pharma companies query their own data in plain language. The consortium grew directly out of conversations the PharmaOS team was already having with R&D heads, regulatory leaders, and Chief Information Officers across the Indian pharmaceutical industry.
PharmaOS is a product. AIforPharma is not. The two share an organising team because the team is closest to the conversation and willing to host it, but the consortium is editorially independent of the product: PharmaOS is one demonstration among many at any given edition, and it is held to the same standard the room holds anyone else's system to.
For organising-team contact: room@aiforpharma.ai.
Apply. The application is the room.
The way into AIforPharma is to apply. The application takes about seven minutes and is read in full by the organising team within seven days. Accepted applicants receive a formal invitation along with the date, venue, and agenda for the founding edition. Those we cannot accept for the founding edition are told plainly, and told when we expect to be in touch next.
If the application is going to feel like trouble, that is the point: we are building a room that earns its seriousness from the care with which we assemble it.