from absci import de_novo_model model = de_novo_model.load_latest() antigen = model.load_pdb("7olz.pdb", chain="A") antibodies = model.predict(antigen, N=300000) from absci_library import codon_optimizer library = codon_optimizer.reverse_translate(library) library.to_csv("covid-antibody-designs.csv") library.to_wet_lab(assay="ACE") from absci import lead_opt_model lead_optimizer = lead_opt_model.load_latest() library.naturalness = lead_optimizer.naturalness(library) lead_optimizer.optimize(library).to_wet_lab(as say="SPR") from absci import genetic_algorithm; parameters=["maximize|binding_affinity:pH=7.5", "minimize|binding_affinity:pH=6.0", "maximize|human_naturalness"]; library = genetic_algorithm.multiparametric_optimization(library, parameters, evolutions=100); library.to_wet_lab(assays=["ACE", "SPR", "Bioassays"]) C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . CORPORATE PRESENTATION SPRING 2025

2C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Disclaimers Forward-Looking Statements Certain statements in this presentation that are not historical facts are considered forward-looking within the meaning of Section 27A of the Securities Act of 1933, as amended, and Section 21E of the Securities Exchange Act of 1934, as amended, including statements containing the words “will,” “may,” “anticipates,” “plans,” “believes,” “forecast,” “estimates,” “expects,” “predicts,” “advancing,” “aim,” and “intends,” or similar expressions. We intend these forward-looking statements, including statements regarding our strategy, our expectations regarding the clinical, therapeutic and market potential of product candidates discovered and developed through our platform; the potential advantages of our technology and the assets in our internal pipeline; our ability to achieve catalysts in our preclinical and clinical development programs, such as the initiation of IND-enabling studies and Phase 1 clinical development and the receipt of clinical data; the anticipated timing of such events; the expected evolution of our portfolio over time; guidance regarding cash, cash equivalents and our projected cash runway, our future operations, internal research and technological development activities, estimated speed and cost advantages of leveraging our AI drug creation platform; our expectations regarding the status and progress of our existing partnerships and our plans for potential new partnerships; our expected operational efficiencies, research and technology development collaboration efforts, growth plans, prospects, plans and objectives of management, to be covered by the safe harbor provisions for forward-looking statements contained in Section 27A of the Securities Act and Section 21E of the Securities Exchange Act, and we make this statement for purposes of complying with those safe harbor provisions. These forward-looking statements reflect our current views about our plans, intentions, expectations, strategies, and prospects, which are based on the information currently available to us and on assumptions we have made. We can give no assurance that the plans, intentions, expectations, or strategies will be attained or achieved, and, furthermore, actual results may differ materially from those described in the forward-looking statements and will be affected by a variety of risks and factors that are beyond our control, including, without limitation, risks and uncertainties relating to the development of our technology as well as the assets in our internal pipeline, our ability to secure milestone payments and royalties, and our ability to effectively conduct research, drug discovery and development activities with respect to our internal programs and to collaborate with our partners or potential partners with respect to their research, drug discovery and development activities; along with those risks set forth in our most recent periodic report filed with the U.S. Securities and Exchange Commission, as well as discussions of potential risks, uncertainties, and other important factors in our subsequent filings with the U.S. Securities and Exchange Commission. Except as required by law, we assume no obligation to update publicly any forward-looking statements, whether as a result of new information, future events, or otherwise. Market and Statistical Information This presentation also contains estimates and other statistical data made by independent parties and by us relating to market size and growth and other industry data. These data involve a number of assumptions and limitations, and you are cautioned not to give undue weight to such estimates. We have not independently verified the data generated by independent parties and cannot guarantee their accuracy or completeness. Trademark usage This presentation/document/webpage contains references to our trademarks and service marks and to those belonging to third parties. Absci®, ®, SoluPro®, Bionic SoluPro® and SoluPure® are Absci registered trademarks with the U.S. Patent and Trademark Office. We also use various other trademarks, service marks and trade names in our business, including the Absci AI logo mark ( ), the Unlimit with us mark ( ), Denovium, Integrated Drug Creation, HiPrBind, and IgDesign. All other trademarks, service marks or trade names referred to in this presentation/document/webpage are the property of their respective owners. Solely for convenience, the trademarks and trade names in this presentation/document/webpage may be referred to with or without the trademark symbols, but references which omit the symbols should not be construed as any indicator that their respective owners will not assert, to the fullest extent under applicable law, their rights thereto.

3C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . P L A T F O R M V A L I D A T E D T H R O U G H I N D U S T R Y - L E A D I N G P A R T N E R S H I P S I N C L U D I N G W I T H A S T R A Z E N E C A , M E R C K , N V I D I A , & A M D D I F F E R E N T I A T E D L A B - I N - A - L O O P : ‘ D A T A T O T R A I N ’ , ‘ A I T O C R E A T E ’ , & ‘ W E T L A B T O V A L I D A T E ’ I N R A P I D 6 - W E E K C Y C L E S I N T E R N A L P I P E L I N E O F P O T E N T I A L L Y ‘ B E S T - I N - C L A S S ’ & ‘ F I R S T - I N - C L A S S ’ A S S E T P R O G R A M S L E A D I N G A I M O D E L S W I T H P O T E N T I A L T O A C C E S S ‘ H A R D - T O - D R U G ’ T A R G E T S A N D U N L O C K N O V E L B I O L O G Y ABS-101: “Best-in-class” potential anti-TL1A antibody entering clinic 1H 2025 ABS-201: nomination of drug candidate for androgenic alopecia addressing significant clinical and commercial opportunity ABS-301 & ABS-501: lead and candidate ID on novel and differentiated programs designed using AI Absci is a data-first generative AI Drug Creation™ company

4C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . LEADING AI MODELS Leading de novo AI model for antibody design with proof- points in internal and partnered programs DATA ADVANTAGE Proprietary ultra-high throughput data generation in 77,000+ ft2 lab Amassing high quality data at scale since 2020 ‘MULTILINGUAL’ EXPERTISE World-Class Cross-disciplinary discovery and AI team >10 Drugs Approved under current leadership COMPUTE AT SCALE Compute at scale enabled by partnerships with AMD, NVIDIA & Oracle Ingredients for Success Absci’s leadership in AI de novo antibody design

5C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Since 2020 Absci has been amassing high-quality data at scale for AI model training and validation DATA TO TRAIN Proprietary High throughput screening assays generate high- quality data for generative AI model training WET LAB TO VALIDATE 77,000 Sqft+ lab to validate AI-generated designs AI TO CREATE Advanced generative AI models create antibodies and next-gen biologics through de novo design and AI Lead Optimization 6 WEEK ‘LAB IN A LOOP’ CYCLES CONTINUOUSLY IMPROVE AI MODELS

6C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Leadership in AI de novo design of antibody-based therapeutics EVQLSEVGA . . . de novo antibody design model creates epitope-specific binders given a target structure Designed in framework of choice or multiple frameworks INPUT EMBEDDING STRUCTURE PREDICTION (DIFFUSION) . . . ARCPSIWKFPDEEGACQPC . . . Antigen Structure/Sequence (Epitope) PROTEIN LANGUAGE MODELS Co-optimization enables improvement of antibody attributes while maintaining developability Precise engineering of molecule pharmacology AI LEAD OPTIMIZATIONDE NOVO ANTIBODY DESIGN

7C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . ENABLING FEATURES: MULTI-VALENCY, pH-DEPENDENT BINDING ABILITY TO ADDRESS DIFFICULT TARGET CLASSES, E.G. GPCRS EPITOPE-SPECIFIC DESIGN + EPITOPE INTERFACE OPTIMIZATION BROAD IP: 100S TO 10,000S OF FUNCTIONALLY VALIDATED SEQUENCES ENABLED BY PROPRIETARY WET-LAB VALIDATION ENHANCED POTENCY AND MOA We use AI to create novel & differentiated therapeutics

8C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Leveraging AI throughout the end-to-end drug discovery process I N T E G R A T E D D R U G C R E A T I O N P L A T F O R M TARGET SELECTION AI reverse immunology target discovery OR Additional Target Selection approaches AI-GUIDED DRUG CREATION De novo antibody design AI-GUIDED LEAD OPTIMIZATION Affinity & developability Pharmacology engineering TM

9C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . de novo Model v1 Absci was the first to design and validate novel antibodies using zero-shot generative AI in BioRxiv preprint de novo Model v2 Demonstrated de novo design model’s broad applicability to multiple therapeutic antigens in Neurips publication de novo Model v3 Successfully designed high affinity binders to an epitope without known binder in Large Pharma partnership de novo Model v4 and continued development Successfully de novo designed against previously “undruggable” target in HIV “Caldera” program in collaboration with Caltech 2022 2023 2024 2025 Since publishing the first work in AI de novo antibody design, Absci has continued to rapidly progress and lead the field

10C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Platform Case Studies AI LEAD OPTIMIZATIONDE NOVO ANTIBODY DESIGN Goal: create universally neutralizing HIV antibody by binding conserved epitope within “caldera” region of HIV gp120 Absci’s de novo design platform can successfully address difficult to drug target epitopes Goal: Co-optimize antibodies for pH sensitive binding to increase efficacy and reduce Absci’s lead optimization platform enables molecules with differentiated pharmacology AI LEAD OPTIMIZATION FOR PH SENSITIVITY WHICH MAY REDUCE TOXICITY AND/OR IMPROVE EFFICACY OF THERAPEUTIC mAbs VIEW THE FULL CASE STUDY DE NOVO ANTIBODY DESIGN PROGRAM IN COLLABORATION WITH CALTECH FUNDED BY THE GATES FOUNDATION VIEW THE FULL CASE STUDY Model searches a massive space of ~1019, identifying functional and developable antibodies in one step.

11C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . ”Multilingual” team with expertise in AI and drug creation Sean McClain Founder, CEO & Director Andreas Busch, PHD Chief Innovation Officer Zach Jonasson, PHD Chief Financial Officer & Chief Business Officer Amaro Taylor-Weiner, PHD SVP, Chief AI Officer Karin Wierinck Chief People Officer Christian Stegmann, PHD SVP, Drug Creation Christine Lemke, DVM SVP, Portfolio & Growth Strategy Shelby Walker, JD Chief Legal Officer Karen Mcginnis, CPA Former Chief Accounting Officer, Illumina Amrit Nagpal Managing Director, Redmile Group Joseph Sirosh, PHD Former CTO, Compass VP, Amazon & Microsoft Dan Rabinovitsj VP Hardware Engineering, Meta Frans Van Houten Chairman of the Board Former CEO, Royal Phillips Sir Mene Pangalos, PHD Former EVP R&D AstraZeneca Sean McClain Founder, CEO & Board Director Ian McInnes, PHD Vice Principal and Head of College University of Glasgow Hubert Truebel, MD, PHD, MBA Chief Medical Officer AiCuris Luis Diaz, MD Head, Division of Solid Tumor Oncology Memorial Sloan Kettering Cancer Center John Wherry, PHD Director, Institute for Immunology & Immune Health, University of Pennsylvania Victor Greiff, PHD Associate Professor University of Oslo Sir Mene Pangalos, PHD Co-Chair SAB Former EVP R&D AstraZeneca Andreas Busch, PHD Co-Chair SAB Chief Innovation Officer L E A D E R S H I P T E A M B O A R D O F D I R E C T O R S S C I E N T I F I C A D V I S O R Y B O A R D E X P E R T I S E & B A C K G R O U N D F R O M Penelope Chief Morale Officer O U R P E O P L E “M ultilingual” Expertise Trademarks, service marks or trade names referred to herein are the intellectual property of their respective owners. Use of this IP does not imply affiliation, endorsement or sponsorship of any kind

12C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . INFL. Bowel Disease / TL1A Androgenic Alopecia / PRLR Immuno-oncology / Undisclosed 3 New Early Discovery Programs Target Lead ID Candidate ID IND-Enabling ABS-101 ABS-201 ABS-301 Phase 1 Oncology / HER2 ABS-501 IND* Therapeutic area/target DC Advancing and expanding our pipeline of novel & differentiated assets designed using AI Lead *or equivalent ex-US filing A B S - 1 0 1 FiH in 1H25 and Ph1 Interim data readout 2H25. New preclinical data support potentially superior immunogenicity profile. KEY HIGHLIGHTS A B S - 2 0 1 Category defining PRLR antibody for androgenic alopecia. IND- enabling activities initiated, with potential to be first in U.S market. A B S - 3 0 1 Potential first-in-class asset with target validation and initial preclinical efficacy readouts in 1H25. A B S - 5 0 1 Candidate ID phase for novel HER2 program designed using de novo AI A I P I P E L I N E

13C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . AI Drug Creation™ Partnerships Data & compute collaborations 25+ PARTNERED PROGRAMS SCALING COMPUTE 4 NAMED INTERNAL PROGRAMS ADDITIONAL PROGRAMS IN EARLY DEVELOPMENT IMPROVING MODELS INCREASING EFFICIENCIES Trademarks, service marks or trade names referred to herein are the intellectual property of their respective owners. Use of this IP does not imply affiliation, endorsement or sponsorship of any kind Driving Growth Through Industry-Leading Collaborations P A R T N E R S H I P S

14C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . 1H 2025 2H 20251H 2025 Leading AI platform driving numerous near-term value inflection points ABS-101 Ph 1 double-blind, placebo-controlled Clinical Trial initiation ABS-101 Phase 1 Interim Data Readout ABS-201 IND-enabling activities initiated 4Q 2024 ABS-301 Immuno-oncology Development Candidate 2025 Partnerships: • New partnerships including Large Pharma • Milestones on current partnerships • Potential internal asset transaction 1H 2026 ABS-101 Ph1b/2a Clinical Trial Initiation ABS-201 Ph1 Trial Initiation with opportunity for rapid PoC C A T A L Y S T S

15C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . A B S - 1 0 1 Continued progress of TL1A asset with FiH in 1H 2025 • New preclinical data supporting superior immunogenicity profile • Phase 1 Interim data readout in 2H 2025 A B S - 2 0 1 Development Candidate for PRLR (prolactin receptor) nominated early December 2024 IND-enabling activities initiated A B S - 3 0 1 Progress of first-in-class asset with target validation readout in 1H 2025 N E W : A B S - 5 0 1 Nomination of a potential best-in-class HER2 asset C o n t i n u e d a d v a n c e m e n t o f l e a d a s s e t s D i s c o v e r y o f n e x t a s s e t s Absci’s progress in Drug Creation I N T E R N A L P I P E L I N E

16C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Potential best-in-class TL1A mAb designed using generative AI ABS-101 designed to achieve competitive therapeutic properties and potential for clinical differentiation A B S - 1 0 1 T L 1 A Higher affinity and potency Bind monomer and trimer TL1A High bioavailability Expected low immunogenicity Favorable developability High convenience based on half-life extension and sub-Q dosing

17C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Inflammation Fibrosis Clinically Validated Mechanism of Action in Large Underserved Market A B S - 1 0 1 T L 1 A D A T A H I G H L I G H T S T L 1 A : D R 3 S I G N A L I N G C L I N I C A L L Y S H O W N T O I N D U C E P R O - I N F L A M M A T O R Y R E S P O N S E S 1 P O T E N T I A L R E L E V A N C E I N W I D E R A N G E O F A U T O I M M U N E I N D I C A T I O N S DC MΦ Membrane TL1ADR3 T-Cell Signaling Signaling DR3 Soluble TL1A 2 Wang 2023 http://dx.doi.org/10.1136/bmjopen-2022-065186 3 Dahlhamer, James M., et al. "Prevalence of inflammatory bowel disease among adults aged≥ 18 years—United States, 2015." Morbidity and mortality weekly report 65.42 (2016): 1166-1169. 4 Evaluate Pharma Oct 2023. 1 Adapted from Takedatsu 2008 doi: 10.1053/j.gastro.2008.04.037 Significant market opportunities beyond Inflammatory Bowel Disease $22B+ Global Market4 $4.5B for TL1A 0.8- 3M U.S. Inflammatory Bowel Disease Prevalence3 5M Global Inflammatory Bowel Disease (IBD) Prevalence2

18C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . H I G H A F F I N I T Y M A B S W I T H B I N D I N G T O B O T H T H E T L 1 A M O N O M E R A N D T R I M E R A I - O P T I M I Z E D L O W p M A F F I N I T Y T R A N S L A T E S T O S U P E R I O R O R E Q U I V A L E N T P O T E N C Y # Estimated performance of a putative clinical competitor molecule In cr ea si ng p ot en cy IC 50 (n M ) 0 2 4 6 8 10 12 ABS-10 1 RVT-3 10 1# (R oche) MK-7 240 # (M erc k) APOPTOSIS INHIBITION ASSAY IN TF-1 CELLS AFFINITY BY BIOLAYER INTERFEROMETRY (BLI) AI platform designed advanced leads with high affinity and superior potency A B S - 1 0 1 , T L 1 A Increasing affinity In cr ea si ng a ff in ity 1101001000 1 10 100 1000 Monom er Bind ing (pM) Tr im er B in di ng (p M ) No M onomer Bind ing Weak monomer binding ABS-101 RVT-3101# (Roche) MK-7240# (Merck) TEV-48574# (Sanofi)

19C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . ABS-101 and MK-7240 show reduced TL1A complex internalization than RVT-3101 A B S - 1 0 1 , T L 1 A mAb only ABS-101 RVT-3101MK-7240 10x 10x 10x 10x 10x 10x 63x 63x 63x 20 µ m 20 µ m 20 µ m 20 µ m 20 µ m 20 µ m 20 µ m 20 µ m 20 µ m mAb-TL1A complex M A B : T L 1 C O M P L E X I N T E R N A L I Z A T I O N I N T H P - 1 C E L L S Internalization of mAb:TL1A complexes potentially contributes to immune activation and formation of ADA **p<0.001, Mann-Whitney test Reference, doi: 10.1053/j.gastro.2019.08.009 # # # 0 200000 400000 600000 800000 ABS-101 MK-7240 RVT-3101 ns ** ** C or re ct ed T ot al C el l Fl uo re sc en ce ( C TC F) # Estimated performance of a putative clinical competitor molecule ABS-10 1# MK-7240# (M erck) RVT-310 1# (R oche)

20C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Latest Non-Human Primates & CMC data confirm compelling ABS-101 competitive profile A B S - 1 0 1 , T L 1 A 2-3x extended half-life in NHPs over clinical competitors to support Q8W-Q12W dosing interval ABS-101 shows enhanced biodistribution in NHPs, compared to antibodies in clinical development based on in silico modelling Optimal developability profile allowed successful development of high-concentration formulation at 200mg/mL suitable for subcutaneous injection C M C - H I G H C O N C E N T R A T I O N F O R M U L A T I O N 2 - 3 X L O N G E R H A L F - L I F E I N N H P s C O M P A R E D T O C L I N I C A L C O M P E T I T O R S m A b se ru m c on c (µ g/ m l) Time (days) 0 10 20 30 40 50 60 10¹ 10² 10³ ABS-101 MK-7240 RVT-3101 Preliminary 13-week GLP-tox shows no treatment- related adverse findings during in-life phase and necropsy; histopathology pending High subcutaneous bioavailability in NHPs at ~80% N H P - P K & P R E L I M I N A R Y 1 3 - W E E K N H P G L P - T O X

21C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Latest NHP data confirm compelling profile and target engagement Confirmatory target engagement N o n - H u m a n P r i m a t e s ( N H P ) s i n g l e d o s e P K / P D s t u d y h i g h l i g h t s : Significant improved target engagement vs. competitor molecules at comparative dosing regimen Dose dependency of target engagement including ceiling effect T A R G E T E N G A G E M E N T O V E R T I M E P R O F I L E ( T O T A L s T L 1 A A F T E R S I N G L E D O S E ) A B S - 1 0 1 T L 1 A

22C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . AI-designed for potentially optimal therapeutic profile A B S - 1 0 1 , T L 1 A A T T R I B U T E ABS - 1 0 1 M K - 7 2 4 0 ( M E R C K , P R O M E T H E U S ) R V T - 3 1 0 1 ( R O C H E , R O I V A N T ) T E V - 4 8 5 7 4 ( S A N O F I , T E V A ) High affinity/potency ++ - + + Trimer TL1A binding ++ + + ++ Monomer TL1A binding ++ + - - Low Immunogenicity potential + + - NA Bioavailability/ Biodistribution ++ + - NA Sub-Q injection + + + - Q8W to once quarterly dosing ++ - - --

23C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Continued progress with FiH expected in 2025 A B S - 1 0 1 T L 1 A AI-designed Development Candidate ü High affinity ü High potency ü Long half-life ü Favorable manufacturability 1 Q 2 0 2 4 I N I T I A T E D F E B 2 0 2 4 1 H 2 0 2 5 IND-enabling studies to evaluate: ü GMP manufacture of sub-Q formulation at high concentration ü Favorable PK and long half-life ü High Bioavailability in NHPs • Low ADA ü 13-week GLP tox: No treatment- related adverse findings during in-life phase and necropsy observed. Histopathology pending Phase 1 double-blind, placebo-controlled trial initiation Phase 1 interim data readout 2 H 2 0 2 5

24C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . A B S - 1 0 1 Continued progress of TL1A asset with FiH in 1H 2025 • New preclinical data supporting superior immunogenicity profile • Phase 1 Interim data readout in 2H 2025 A B S - 2 0 1 Development Candidate for PRLR (prolactin receptor) nominated 12/2024 IND-enabling activities initiated A B S - 3 0 1 Progress of first-in-class asset with target validation readout in 1H 2025 N E W : A B S - 5 0 1 Nomination of a potential best-in-class HER2 asset C o n t i n u e d a d v a n c e m e n t o f l e a d a s s e t s D i s c o v e r y o f n e x t a s s e t s Absci’s progress in Drug Creation I N T E R N A L P I P E L I N E

25C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . ABS-201 has the potential to unlock a wholly new category of therapy in hair “re-growth” Significant unmet clinical need for androgenic alopecia Large market: 80-90M patients in U.S., which is a highly motivated patient population CLINICAL AND COMMERCIAL UNMET NEED Straightforward clinical development path with option for early Proof of Concept Low competition, potentially first to U.S. market DEVELOPMENT PATH Highly validated target (efficacy & safety) for treatment of androgenic alopecia Supportive pharmacological profile of ABS-201 SCIENTIFIC RATIONALE

26C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Significant underserved patient population looking for therapeutic innovation FEMALE ANDROGENIC ALOPECIA ~30M women in U.S. Only 1 FDA approved therapy for women ~50M men in the U.S. Only 2 FDA approved therapies 80 - 90 MILLION AMERICANS LIVE WITH ANDROGENIC ALOPECIA MALE ANDROGENIC ALPOPECIA Growing patient population with limited therapeutic options and side-effect concerns Last FDA approved therapy for androgenic alopecia was in the 1990s Patients and clinicians need better treatment options for “hair re-growth” Hair re-growth, not just slowing of hair loss Safe and minimal side effects Durable effect Convenient administration frequency FDA approved A B S - 2 0 1

27C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Prolactin Receptor inhibition MOA: stimulation of anagen hair growth phase Proposed impact of ABS-201 on Hair Cycle Stages ABS-201 has the potential to: Catagen ↑↑ Apoptosis & Regression Telogen Resting Phase Hair falls out PRLR Anagen Active Growth & New Hair ABS-201 Anagen ↑↑ Active Growth & New Hair 2-6 years Telogen Resting Phase Hair falls out PRLR Catagen Apoptosis & Regression Shift the balance in hair cycle stage towards anagen phase1,2 with: • active and new hair growth • prevention of telogen effluvium Restore hair pigmentation2 Promote a long-lasting effect after treatment cessation 1 doi: 10.1016/S0002-9440(10)64295-2 2 doi: 10.2353/ajpath.2006.050468 Prevent prolactin mediated telogen effluvium1,2 A B S - 2 0 1

28C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Prolactin induces catagen-like phase in cultured human hair follicles A B S - 2 0 1 1 doi: 10.2353/ajpath.2006.050468 2 doi: 10.1056/NEJMoa1805171 P r o l a c t i n - d r i v e s h a i r f o l l i c l e r e g r e s s i o n i n h u m a n e x v i v o c u l t u r e Prolactin prematurely induces a catagen- like stage in organ-cultured human hair follicles1 characterized by: Apparent cessation of pigmentation Condensed shape of the dermal papilla Diminishment of the hair matrix volume Vehicle 400 ng/ml prolactin Catagen IIIAnagen VI HS M DP IRS IRS HS ORS DP MK Inhibition of hair shaft elongation Human genetic evidence suggests no safety liabilities targeting PRLR2

29C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Kobayashi, 2018 NEJM Female compound heterozygous PRLR loss-of- function carrier lacks complete PRLR signaling A 35-year-old woman with postpartum agalactia and hyperprolactinemia, otherwise in good health, with no apparent impact on fertility and completely normal serum electrolytes and hormone levels (except PRL). NV = Nonvariant ND = Not determined Family with dominant negative PRLR loss-of-function mutation leads to hypo-prolactenemia and reduced PRLR signaling In a study of two generation of women, all women had postpartum agalactia and hypoprolactinemia, otherwise in good health, normal breast development, no hemorrhage or hypotension at time of deliveries, and no reported abnormalities of other hypothalamic- pituitary axes Moriwaki, 2021 JCEM PRLR Inhibition anticipated to be safe and well tolerated, as supported by Human Genetics data A B S - 2 0 1

30C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Treatment with an anti-PRLR mAb promotes and sustains long-term hair growth in NHP TOP HEAD VIEW OF STUMPTAILED MACAQUE’S SHOWING PHENOTYPIC CHANGE OVER TIME 40mg/kg s.c. Q2W for 28 weeks Disclosure from competitor Hair density & thickness improved with short treatment duration in primate model of androgenic alopecia Hair growth remains several years post cessation Hair regrowth observed for both male and female animals M al e Fe m al e Baseline 12 weeks 28 weeks 6 months 2 years 4 years Post-treatmentTreatment Translational Model validates PRLR Target

31C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . ABS-201 shows superior efficacy vs 5% topical minoxidil in 21d hair regrowth model Administration: mAbs i.p. biweekly; Minoxidil topical daily Untreated (n=11) Isotype (n=11) Minoxidil 5% (n=11) ABS-201 30mg/kg (n=11) ABS-201 60mg/kg (n=10) ABS-201 vs minoxidil/untreated/isotype **p<0.05; ***p<0.0001 - 2way ANOVA Error bars= SEM *** *** *** * A B S - 2 0 1

32C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Straightforward path for ABS-201 clinical development C L I N I C A L T R I A L S F O R H A I R T R E A T M E N T S A R E S T R A I G H T F O R W A R D • Ease of patient recruitment • High level of KOL Interest • Ability to conduct multi-center trials • Non-invasive trial conduct W E L L D E F I N E D E N D P O I N T S W I T H V A L I D A T E D M E A S U R E S Primary Endpoints: Quantitative measurements with follicular dermatoscope (trichoscopy) • Terminal Hair Growth • Total Hair Count • Total hair density (per cm2) Secondary Endpoints: • Patient Reported Outcomes as measured by validated scales accepted by the FDA (HairDex; Hair Specific Skindex-29 (FPHL); The Men’s Hair Growth Questionnaire (MHGQ)); Women’s Hair Growth Questionnaire (WHGQ) • Re-pigmentation B E N C H T O B E D S I D E

33C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . ABS-201 represents a significant untapped new market opportunity Consumers across income levels invest in aesthetic treatments, driving steady demand. STRONG WILLINGNESS TO SELF-PAY ACROSS DEMOGRAPHICS UNLOCKS WHOLLY NEW CATEGORY OF THERAPY Lack of innovation and effective treatments in the hair loss space Hair is long considered the last frontier of medical aesthetics market Potential 2-3X larger depending on clinical profile and additional indications such as hair re-pigmentation $14B+ MARKET WITH SIGNIFICANT UPSIDE POTENTIAL Source: BCG

34C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . A B S - 1 0 1 Continued progress of TL1A asset with FiH in 1H 2025 • New preclinical data supporting superior immunogenicity profile • Phase 1 Interim data readout in 2H 2025 A B S - 2 0 1 Development Candidate for PRLR (prolactin receptor) nominated early December 2024 IND-enabling activities initiated A B S - 3 0 1 Progress of first-in-class asset with target validation readout in 1H 2025 N E W : A B S - 5 0 1 Nomination of a potential best-in-class HER2 asset C o n t i n u e d a d v a n c e m e n t o f l e a d a s s e t s D i s c o v e r y o f n e x t a s s e t s Absci’s progress in Drug Creation I N T E R N A L P I P E L I N E

35C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . The presence of TLS is associated with longer progression-free survival and better response to immune checkpoint inhibitors2,3. Rapidly growing evidence illustrates correlation between TLS-derived antibodies in the tumor microenvironment and positive clinical outcomes2. TLS-derived antibodies have been shown to be associated with apoptosis of cancer cells in patients2. Tertiary lymphoid structures (TLS) are centers of immune activity, such as B-cell proliferation and antibody production, that develop in chronically inflamed tissues1. Antibodies from TLS are specialized for local antigens and play a significant role in the progression of chronic diseases and cancer, setting them apart from the general population of antibodies in the peripheral blood2. 100 75 50 25 10 15 20 0 0 5 High Ig Staining Low Ig Staining Pr og re ss io n- fr ee ( % ) Time (months) P= 0.019 Tertiary Lymphoid Structures (TLS): The focus of Absci’s Reverse Immunology approach T A R G E T D I S C O V E R Y 1 doi: 10.3389/fimmu.2018.01952 2 doi: 10.1016/j.immuni.2022.02.001 3 doi: 10.1038/s41586-019-1922-8

36C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . LIGHT CHAINS HEAVY CHAINS Samples collected from patients with high expression of TLS markers Antibody expression Immunoglobulin reads from RNAseq data Target identification using high throughput proteomics Assembled Ig chain sequences Target antigen confirmed through SPR or BLI Computationally reconstructed antibodies Fully human antibody and target antigen identified Reference, doi: 10.1101/2021.02.06.430058 A B S - 3 0 1 | Reverse Immunology platform identifies the antigens targeted by endogenous antibodies produced in tumor lymphoid structures, TLS

37C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . A B S - 3 0 1 r e s c u e s p r o - i n f l a m m a t o r y s i g n a l i n g t h r o u g h i n h i b i t i o n o f i m m u n o s u p p r e s s i v e c y t o k i n e A B S - 3 0 1 | A patient-derived antibody discovered by reverse immunology blocks an immunosuppressive cytokine Restoring signaling by blocking immunosuppressive cytokine which may promote immune-mediated tumor cell killing Target cytokine is suggested to maintain an immunosuppressive environment through signaling inhibition Re ce pt or s ig na llin g (O D 6 20 nm , A U ) ABS-301 Isotype control Isotype control Concentration (M) Immunosuppressive Activator Immune-mediated tumor killing activation Cytokine: T a r g e t b i o l o g y a n d p r o p o s e d A B S - 3 0 1 m e c h a n i s m o f a c t i o n

38C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . In vivo Target Validation: Pathway Activation Drives Potent Anti-Tumor Response A B S - 3 0 1 Key Findings: Activation of the ABS-301–targeted pro-inflammatory pathway triggers a robust anti-tumor immune response. Study Overview: Mouse melanoma cells were genetically modified to activate the ABS-301–targeted pro-inflammatory pathway via Activator expression. Tumor progression was assessed in immunocompetent mice injected with either engineered cells or unmodified parental cells. T U M O R G R O W T H M O U S E M O D E L

39C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Expression of ABS-301’s target suggests broad potential in squamous cell carcinomas A B S - 3 0 1 Distribution of ABS-301 target expression across squamous cell carcinoma cohorts. ABS-301 target expression log2(TPM+1) Values shown are log2(TPM+1) normalized. Multiple biopsies from a patient are included in the analysis. Source: Tempus

40C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . ABS-301 | Expression in Lung Squamous Cell Carcinoma (LUSC): no change with treatment and strong negative correlation with CD8+ T cell infiltration In LUSC, univariate analysis of ABS-301 expression indicate only a minor change in expression between pre- and post-treatment suggesting opportunity for combination therapy. S u s t a i n e d t a r g e t e x p r e s s i o n i n L U S C ABS-301 target expression shows a strong negative correlation with CD8+ T cell infiltration with a minimal effect on Treg infiltration supporting immunosuppressive activity of target in vivo. C D 8 + I n f i l t r a t i o n n e g a t i v e l y c o r r e l a t e d w i t h t a r g e t e x p r e s s i o n i n L U S C Source: Tempus Source: TCGA Pre-treatment Post-treatment A BS -3 01 ta rg et e xp re ss io n (lo g2 (T PM +1 )) A BS -3 01 e xp re ss io n le ve l (lo g2 T PM )

41C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . ABS-301 | Broad potential in immuno-oncology Based on literature and potential competitive molecules, the following indications could be of interest: *dependent on stage of diagnosis References provided in appendix Indication US Prevalence Estimated 5-year survival rate* US Sales in 2030 NSCLC Calculated: ~202K in 2023 28% $27B SCC 30% of NSCLC cases Calculated: ~61K 24% Calculated Sales: $8.1B Head and Neck SCC ~54K in 2022 68.5% Calculated Sales: $2.3B Esophageal Cancer ~21K in 2022 20% $1.5B SCC ~20% of cases Calculated: ~4.2K Calculated Sales: $0.3B Cervical Cancer ~14K in 2023 $0.6B SCC 90% of cases Calculated: ~13K 67% Calculated Sales: $0.6B Skin Cancer, non-melanoma Incidence = ~3,300K 95-100% $1.0B SSC Incidence = ~700K 95% Calculated Sales: $0.2B

42C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . A B S - 1 0 1 Continued progress of TL1A asset with FiH in 1H 2025 • New preclinical data supporting superior immunogenicity profile • Phase 1 Interim data readout in 2H 2025 A B S - 2 0 1 Development Candidate for PRLR (prolactin receptor) nominated early December 2024 IND-enabling activities initiated A B S - 3 0 1 Progress of first-in-class asset with target validation readout in 1H 2025 N E W : A B S - 5 0 1 Nomination of a potential best-in-class HER2 asset C o n t i n u e d a d v a n c e m e n t o f l e a d a s s e t s D i s c o v e r y o f n e x t a s s e t s Absci’s progress in Drug Creation I N T E R N A L P I P E L I N E

43C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . • Hits with edit distance of up to 12 amino acids in HCDR3 region (13 aa, search space of 2013) were screened • Selected 50 hits with <10 nM affinity were expressed as mAbs for binding affinity determination • Top 11 antibodies were characterized in vitro and 3 leads evaluated in vivo ABS-501, HER2 | Deploying de novo AI model on HER2 led to discovery of antibodies displaying molecular interactions distinct from trastuzumab Variant # Edit distance KD (nM) Epitope mapping view Loop 581-590 Trastuzumab 0 1.07 1 7 4.16 3 7 9.75 4 2 6.66 Partial Critical Not critical Z e r o s h o t d e n o v o A I d i s c o v e r y o n H E R 2 AI-designed antibodies: same epitope, different HER2 contact preferences Epitope of interest Hits Leads bioRxiv 2023.01.08.523187; doi: https://doi.org/10.1101/2023.01.08.523187

44C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Luciferase signal driven by NFAT transcription factor positively correlates to ADCC activation against JIMT-1 ABS-501, HER2 | AI-designed antibodies demonstrate measurable enhancement of ADCC activity compared to trastuzumab ADCC assay principle Trastuzumab Variant 1 Variant 3 Variant 4 EC50 (nM) 0.062 0.056 0.028 0.040 R squared 0.93 0.97 0.97 0.95 P value N/A Not significant <0.0001 0.0015 0 .0001 0 .001 0 .01 0 .1 1 10 100 0 2000 4000 6000 A n t ib o d y , [ n M ] Lu m in es ce nc e, A U Trastuzum ab V a r ia n t 1 V a r ia n t 3 V a r ia n t 4 Iso y p e c o n t r o l JIMT-1 Jurkat cell

45C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . ABS-501, HER2 | AI-designed antibodies suppress growth of trastuzumab-sensitive & resistant HER2+ breast tumors Partial Critical Not critical Trastuzumab WT Variant 1 Variant 3 Variant 4 Mouse xenograft model using EFM192A (HER2+ BC; Tz sensitive) Mouse xenograft model using JIMT-1 (HER2-amp BC; Tz resistant) Trastuzumab-sensitive EFM192A and MDA-MB-361 tumors respond to both trastuzumab (Tz) & AI-designed antibodies Xenograft studies conducted by Dr. Dennis Slamon’s team at UCLA Isotype control Trastuzumab Variant 1 Variant 3 Variant 4 ! "! #! $! ! #!! %!! &!! D() *+ , -. /0 -1 +, 2/ 3, , $ 4 ! "! #! $! %! &! ! #!! %!! D!! (!! "!!! "#!! )*+ ,- . /0 12 /3 -. 41 5. . $ 6 *** ****** *** *** Mouse xenograft model using MDA-MB-361 (HER2+ BC; Tz sensitive) ! "! #! $! ! #!! %!! &!! D!! "!!! ()* +, - ./ 01 .2 ,- 30 4- - $ 5 *** *** ****** 2-way ANOVA ** P<0.001 and ***P<0.0001 vs isotype control JIMT-1 tumors are trastuzumab resistant but sensitive to variants 3 and 4 **

46C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . ABS-501, HER2 | AI-designed variants create opportunities to address unmet medical need Modality switch or combination opportunities under consideration to address unmet medical needs Later-line treatment regimens for HER2-positive cancer: • Monotherapy • Combination therapy with targeted small molecules M u l t i p l e p a t h s p o s s i b l e f o r t h e r a p e u t i c d e v e l o p m e n t : C u r r e n t l y e x p l o r i n g b r e a s t c a n c e r a s o p p o r t u n i t y : a l t e r n a t i v e t o o r p o s t E n h e r t u® Despite Enhertu’s good efficacy, leading oncologists are only moderately satisfied due to toxicity (e.g. interstitial lung disease); less toxic therapy and effective treatment post-Enhertu are key unmet needs. “Post-Enhertu is really where the action is right now in the field. I think the first company that comes up with something that has significant benefit in Enhertu progressive disease is going to win.” – KOL Enhancing efficacy and expanding indications (e.g. Enhertu resistance): • Antibody-drug conjugates (ADCs) • Multi-specific antibodies +

47C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Leading AI models to create novel & differentiated therapeutics “Smart” biologics Enhanced Potency & MOA Engineer selectivity, minimizing off target toxicity Agonism vs. Antagonism Bind Specific extracellular domains Target Specific conformations Address difficult target classes e.g. GPCRs ADDRESS COMPLEX AND PREVIOUSLY “HARD TO DRUG” TARGETS INTRODUCE PRECISE CONTROL OVER ANTIBODY DESIGN

48C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Leadership in AI de novo design of antibody-based therapeutics EVQLSEVGA . . . de novo antibody design model creates epitope-specific binders given a target structure Designed in framework of choice or multiple frameworks INPUT EMBEDDING STRUCTURE PREDICTION (DIFFUSION) . . . ARCPSIWKFPDEEGACQPC . . . Antigen Structure/Sequence (Epitope) PROTEIN LANGUAGE MODELS Co-optimization enables improvement of antibody attributes while maintaining developability Precise engineering of molecule pharmacology AI LEAD OPTIMIZATIONDE NOVO ANTIBODY DESIGN

49C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . AI Platform designs antibodies Wet Lab confirms AI- designed antibodies maintain drug-like properties Wet Lab data improve models D E N O V O A N T I B O D Y D E S I G N Our AI platforms are enabled by our 6-week ‘lab-in-the loop’ active learning cycles A I P L A T F O R M S L A B - I N - T H E - L O O P L E A D O P T I M I Z A T I O N AI guided lead optimization enables tunable pharmacology de novo design of epitope-specific antibodies against targets without requiring a known binder

50C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . AbsciGen: antibody<>antigen complex structure and sequence design AbsciBind: antibody design scoring and filtering Antigen AbsciGen AbsciBind AbsciBind High Rank RMSD = 2.3 Å Confidence = 0.95 AbsciBind Low Rank RMSD = 5.3 Å Confidence = 0.64 AbsciDesign comprises two categories of AI models for de novo antibody design D E N O V O A N T I B O D Y D E S I G N

51C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . STEP 1. Define design parameters STEP 2. Fine-tune and deploy AbsciGen and AbsciBind to generate hundreds of thousands of variants and filter to a subset that are likely binders STEP 3. Wet lab screening and model performance validation The AbsciDesign AI platform delivers de novo antibodies via an end-to-end design- validation workflow Cloning Expression Surface Plasmon Resonance Sequencing

52C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . C A S E S T U D Y de novo design of an antibody that binds the Caldera region of HIV-1 trimer

53C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . No natural or synthetic antibody for HIV exists today because immune system cannot derive an antibody that is universally neutralizing against HIV Design challenge: create universally neutralizing HIV antibody by binding unique and conserved epitope within “caldera” of open conformation of gp120 to prevent HIV from entering host cells Numerous attempts to target this epitope have failed-previous efforts have identified antibodies, but none bind the “caldera” and none are universally neutralizing. de novo design antibody that binds to the highly conserved caldera region of HIV gp120 D E N O V O D E S I G N 17b epitope Caldera HIV gp120 trimer (open)

54C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . HIV-Caldera: Determine inputs and design D E N O V O D E S I G N HIV Env Trimer Challenge : • Highly glycosylated • Extremely high sequence diversity among isolates • High mutation rate at common neutralizing epitopes Model inputs: 1. Antigen structure 2. Framework of 17b 3. Epitope selected conserved across HIV strains (Clades A, B, and C) Design of CDRs: • Condition the model to design long HCDR3s to reach into open caldera region (>20 residues) • Designed HCDR2 and LCDR3 to bind to HIV surface HIV Env trimer (open) HCDR3 LCDR1 LCDR3

55C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . 4 best structures selected from 10,000+ structures generated by de novo model D E N O V O D E S I G N 17b Structure S1 HCDR3 Structure S3 HCDR3 Structure S2 HCDR3 Structure S4 HCDR3 HeavyLight

56C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Applied molecular dynamics simulation to de novo designed antibodies D E N O V O D E S I G N

57C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . H A ta g Antigen Clade A Env trimer Closed Open Enriched de novo library binds open, not closed, Env trimer conformation in YSD D E N O V O D E S I G N Closed Open Clade B Env trimer

58C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . SPR data demonstrate binding characteristics consistent with binding of caldera D E N O V O D E S I G N

59C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . HIV-Caldera: SPR demonstrates no binding of de novo designs to GP120 monomer D E N O V O D E S I G N Hypothesis: If the designed mAbs are binding to the caldera region we should not observe binding to monomeric GP120 since the caldera is only present in the Env trimer Key results: ü 17b showed high affinity binding to monomeric GP120 as expected ü Absci mAbs showed no binding to monomeric GP120, suggesting these binders are targeting an epitope that is only present in the Env trimer ASN3013 ASN3014 ASN3015 ASN3016 17b IgG

60C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . HIV-Caldera: demonstrating AI de novo design for challenging target H I V D E N O V O D E S I G N SUMMARY de novo design model created a novel and diverse antibody which binds multiple clades of HIV indicating successful targeting of the caldera epitope Screening cascade enabled selection of differentially binding variants NEXT STEPS Binders from this study will be selected for affinity maturation Structure of de novo binder and epitope specificity will be experimentally solved to confirm fidelity with designed structure and targeted epitope

61C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . CASE STUDY A I O p t i m i z a t i o n f o r p H s e n s i t i v i t y

62C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . C A S E S T U D Y – A I L E A D O P T I M I Z A T I O N f o r p H S E N S I T I V I T Y AI lead optimization platform for ‘smart biologics’ T H E C H A L L E N G E : The diversity of antibodies is vast, making it impossible for traditional methods to explore effectively. A B S C I S O L U T I O N : Our AI can search a space of ~1019, a million times larger than traditional methods, identifying functional, developable antibodies in one step.

63C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . TUMOR SPECIFICITY IMPROVES EFFICACY AND REDUCES ”ON-TARGET OFF-TUMOR” TOXICITIES pH sensitivity may reduce toxicity and/or improve efficacy of therapeutic mAbs C A S E S T U D Y - A I L E A D O P T I M I Z A T I O N f o r p H S E N S I T I V I T Y Binding occurs in the acidic pH of the tumor microenvironment No binding occurs in the neutral pH surrounding healthy cells DISSOCIATION IN THE ENDOSOME DRIVES ANTIBODY RECYCLING AND EFFICIENT CLEARANCE OF SOLUBLE TARGETS Dissociation at acidic endosomal pH favors antibody recycling Binding at physiological pH drives internalization of the immune complex

64C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Models identify pH sensitive Fab variants from the same lead for either indication C A S E S T U D Y - A I L E A D O P T I M I Z A T I O N f o r p H S E N S I T I V I T Y 1. Library for model training sampled 60 positions on heavy chain framework and CDRs with up to 7 substitutions biased for ionizable residues (H, K, R, D, E) 2. Library screened for antigen binding at pH 7.4 and pH 5.8 3. Model trained and used to generate antibodies with tuned pH dependency Pr ed ic te d bi nd in g sc or e @ p H 5 .8 Predicted binding score @ pH 7.4 AI affinity scoring of variants within a large combinatorial space SPR KD (nM) @ pH 7.4 SP R K D (n M ) @ p H 5 .8 Lab measured affinities of Fab variants predicted to have tighter binding at neutral pH Parental lead Model predictionSP R K D ( nM ) @ p H 5. 8 Lab measured affinities of Fab variants predicted to have tighter binding at low pH Parental lead Model prediction SPR KD (nM) @ pH 7.4 SP R K D (n M ) @ p H 5 .8

65C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . Hits reformatted as mAbs show desired binding profiles C A S E S T U D Y - A I L E A D O P T I M I Z A T I O N f o r p H S E N S I T I V I T Y AI optimized leads achieves variants with pH sensitive binding up to 100x differential pH-sensitive leads had no liabilities for stability, aggregation and polyreactivity1 Model proposed mutations use all 6 ionizing residues in heavy chain CDRs and framework region Sequences were proposed from a >1013 combinatorial space pH 5.8 not p H 7.4 bind at b oth pH 7.4 not p H 5.8 Pare ntal 0.01 0.1 1 10 100 1000 Modeling Strategy SP R K D 7 .4 / 5. 8 1 Data provided in appendix

66C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D . D E N O V O D E S I G N de novo design model created molecule binds multiple clades of HIV suggesting successful targeting of the caldera epitope Represents second disclosed target success for our de novo platform in the 2nd half of this year Absci’s de novo design platform can successfully address difficult to drug target epitopes A I O P T I M I Z A T I O N Models identify unseen variants with 10x-20x pH sensitivity in both directions, and up to 100x differential compared to parental molecule after only one round Designed leads had no liabilities indicating the ability to successfully search a fitness landscape Absci’s lead optimization platform enables molecules with differentiated pharmacology Summarized platform case studies

Better biologics for patients, faster 67C O P Y R I G H T © 2 0 2 5 A B S C I C O R P O R A T I O N . A L L R I G H T S R E S E R V E D .