AI drug discovery partnerships move from promise to practice
AI drug discovery partnerships are strategic collaborations in which computational platforms and experimental laboratories form closed-loop systems that automatically design, synthesize, screen, and refine drug candidates for complex therapeutic targets, turning fragmented discovery steps into an integrated, data-driven workflow that iteratively learns from each cycle to improve potency, selectivity, and developability. These alliances have shifted from hype to hard infrastructure. Receptor.AI and Sethera have partnered to create a closed-loop drug discovery workflow that links polymacrocyclic peptide libraries with physics-based modeling and artificial intelligence. In parallel, Evotec has entered a research collaboration with Odyssey Therapeutics to identify small-molecule treatments for autoimmune and inflammatory diseases using AI-enabled screening and proprietary compound libraries. Taken together, these moves are not side experiments; they signal a deliberate redesign of how difficult-to-target diseases are tackled.
The core message is blunt: if biology is messy and targets are hard, discovery workflows can no longer afford to be linear and manual. Automated drug candidate screening and closed-loop discovery workflows are emerging as the serious alternative to traditional enrichment-led, assay-heavy approaches. Instead of isolated campaigns that throw thousands of molecules at a target and hope for a signal, these AI drug discovery partnerships treat each experimental run as fuel for the next round of design, dialing in better candidates with each iteration. That shift matters most for autoimmune and other complex diseases, where conventional hit-finding has been slow and expensive, and where small improvements in selectivity or stability can make or break a program.
Receptor.AI and Sethera: closing the loop on hard targets
The Receptor.AI–Sethera alliance is a textbook example of what a modern closed-loop discovery workflow should look like. Sethera will generate and experimentally screen architecture-diverse polymacrocyclic peptide libraries, installing one to six stable cross-links to explore polymacrocyclic, nested, in-line, and interpeptide structures across large encoded libraries. Receptor.AI then applies physics-based modeling, artificial intelligence, and multiparameter optimization to interpret sequence, architecture, enrichment, and activity data, build binding hypotheses, and prioritize candidate series. This is not AI in a vacuum; it is AI wired into real wet-lab output. As Alan Nafiiev, PhD, puts it, "Our objective is to use physics and AI not as a substitute for experimentation, but to learn from each experimental cycle and direct the next one".
What makes this partnership opinion-shifting is its commitment to iteration. Subsequently, the companies will design, synthesize, and experimentally test new candidates, using the resulting data to inform each subsequent cycle. That is automated drug candidate screening with a memory, not blind trial-and-error. Sethera’s platform deliberately avoids locking into a single constrained-peptide motif, instead letting target biology and screening data identify the most productive molecular architecture. By combining this chemical space with Receptor.AI’s computational capabilities, the partners intend to move screening-derived hits toward validated lead series with improved potency, selectivity, stability, permeability, and other developability characteristics. If they succeed, their coordinated design-make-test-learn process will set the bar for how hard-to-drug targets should be approached in the AI era.
Evotec and Odyssey: AI-enabled compound libraries for autoimmune disease
While Receptor.AI and Sethera tackle structural complexity, Evotec and Odyssey Therapeutics focus squarely on disease complexity. Evotec has partnered with Odyssey to identify small-molecule treatments for autoimmune and inflammatory diseases, applying AI-enabled screening and proprietary compound libraries across multiple targets. The collaboration will combine AI-supported screening, high-throughput experiments, and disease biology across multiple research targets. This is automated drug candidate screening aimed at conditions where pathways intersect, inflammatory signals are noisy, and traditional target-by-target approaches often stall. Cord Dohrmann, Evotec’s Chief Scientific Officer, states that this collaboration illustrates the company’s strategy of applying integrated discovery platform technologies to complex disease areas.
Unlike more speculative AI plays, this partnership is tied to an advancing clinical pipeline. Odyssey is progressing OD-001 in ulcerative colitis and expects two additional clinical studies to begin during the second half of 2026. That timeline matters because it forces the AI-enabled compound libraries and screening workflows to deliver actionable hit series, not abstract insights. Evotec can receive milestone payments for each successfully validated hit series, aligning incentives around real, validated chemical matter rather than platform buzz. The opinionated reading is straightforward: autoimmune and inflammatory diseases are no longer treated as too messy for AI. Instead, AI drug discovery partnerships are being judged by their ability to feed tangible candidates into concrete clinical programs.
From manual campaigns to automated, data-driven discovery
The most important change these partnerships signal is a move from manual, campaign-based discovery to automated, data-driven discovery systems that learn continuously. Through their alliance, Receptor.AI and Sethera aim to establish a repeatable discovery system that continuously learns from sequence, architecture, counterselection, binding, functional, and developability data generated across experimental campaigns. The resulting workflow is intended to reduce the number of design cycles required to progress from initial screening hits to differentiated peptide lead series. That is a direct challenge to the old model where each screening campaign was a semi-isolated event. In parallel, Evotec’s use of AI-supported screening, high-throughput experiments, and disease biology embeds learning into every step of autoimmune and inflammatory hit-finding.
This matters for time and cost barriers in complex diseases, even if the sources avoid specific price tags. By tightening the closed-loop discovery workflow, partners can cut wasted cycles and focus resources on candidates with better predicted selectivity and developability. Automated drug candidate screening does more than speed up hit identification; it provides richer data for smarter optimization. In my view, the real value is epistemic: the ability to understand why hits work and how they can be improved, which Sethera explicitly cites as a key contribution from Receptor.AI. That understanding is the currency that allows difficult-to-target biology to be approached systematically rather than by hunch and brute force.
What comes next: scaling closed-loop AI discovery across disease areas
These deals are not designed as one-off experiments. The initial Receptor.AI–Sethera program will focus on a mutually selected hard-to-drug target. Following validation of the workflow, the partners intend to pursue additional internal programs and jointly structured discovery collaborations with pharmaceutical and biotechnology companies across selected target classes and therapeutic areas. That intent to scale is critical: it signals that closed-loop workflows are meant to become a standard engine, not a bespoke pilot. On the autoimmune side, Odyssey’s expectation of two more clinical studies starting in the second half of 2026 indicates its pipeline will continue to demand novel candidates, creating ongoing pressure for Evotec’s AI-enabled compound libraries and screening platforms to deliver.
The conclusion is clear. AI drug discovery partnerships are no longer optional curiosities; they are quickly becoming the organizing principle for drug discovery in complex disease spaces. Automated drug candidate screening, AI-enabled compound libraries, and closed-loop discovery workflows are being stitched together into integrated systems that challenge the slow, siloed norms of the past. For autoimmune and difficult-to-target diseases, that shift is not a luxury—it is a necessity. Companies that keep treating AI as an add-on tool rather than as the backbone of an iterative design-make-test-learn process risk falling behind partners who are already building continuously learning discovery engines. The story to watch now is not whether AI will "enter" drug discovery, but which alliances will set the standard for how it is done.
