Slide 1: Generative AI Drug Creation
GENERATIVE AI DRUG CREATION Corporate Presentation August 2026 CONFIDENTIAL AND PROPRIETARY | COPYRIGHT © 2026 ABSCI CORPORATION | ALL RIGHTS RESERVED
Slide 2: Disclosures
DISCLOSURES Disclaimers 1 2 FORWARD-LOOKING STATEMENTS MARKET AND STATISTICAL Statements contained in this press release regarding matters that are not historical INFORMATION facts are “forward-looking statements” within the meaning of the Private Securities This presentation also contains estimates
Slide 3: From Code to Clinic
ABSCI | INTRO 1. AI NATIVE PLATFORM From o Interdisciplinary Team with 10+ approved drugs and AI expertise o Integrated Lab-in-the-Loop leveraging 77k ft² automated wet-lab Code to o Leading AI platform for de novo design and AI optimization of antibody-based therapeutics 2. DIFFERENTIATED PIPELINE
Slide 4: Industrializing Drug Discovery: From Code to Clinic
FROM CODE TO CLINIC Industrializing Drug Discovery $10-15mil (Absci) DATA TO 1 INVESTMENT TRAIN $50-100mil (Large Pharma) ABS-201 FOR PATTERN HAI R LOSS CAPITAL Category re-defining opportunity for PHL in Ph1/2a trials ABS-201 FOR 2 24 Months (Absci) 6-week Lab in the ENDOMETRI OSI S loop cycles Dis
Slide 5: Advancing and expanding our pipeline
PIPELINE Advancing and expanding our pipeline of novel & differentiated assets designed using AI Lead DC IND* THERAPEUTIC TARGET ID LEAD ID CANDIDATE ID IND-ENABLING PHASE 1 PHASE 2 AREA ABS- 201 Pattern Hair Loss (PRLR) Ph1 / 2a PROLACTIN FOCUSED Endometriosis (PRLR) Ph1 PIPELINE ABS- 202 Immunolog
Slide 6: ABS-201 has the potential to unlock a new category of therapy
ABS-201 ABS-201 has the potential to unlock a wholly new category of therapy in hair “re-growth” 1. Significant clinical and commercial unmet need in Pattern Hair Loss 2. Strong scientific rationale, with validated target, de-risked Mode of Action, and pharmacology 3. Straightforward development pat
Slide 7: Underserved patient population looking for therapeutic innovation
ABS-201 PHL Underserved patient population looking for therapeutic innovation Growing patient population with limited ~80 million Americans live with therapeutic options and concerns of Pattern Hair Loss (PHL) adverse side-effects Last FDA approved therapy for Pattern Hair Loss was in the 1990s Pati
Slide 8: PRLR inhibition for Pattern Hair Loss
ABS-201 | PHL PRLR inhibition for Pattern Hair Loss is an innovative alternative to current treatment options ABS-201 HAS THE PROPOSED DIRECT IMPACT OF ABS-201 ON HAIR CYCLE STAGES POTENTIAL TO: Catagen o Shift the balance in hair cycle stage ↑↑ Apoptosis & Catagen towards anagen phase1,2 with: Regr
Slide 9: Prolactin impacts on organ-cultured human hair follicles
ABS-201 PHL Prolactin impacts on organ-cultured human hair follicles Prolactin drives hair follicle regression in human Prolactin prematurely induces a catagen-like stage in ex vivo culture organ-cultured human hair follicles1 characterized by: Anagen VI Catagen III Condensed shape of the dermal pap
Slide 10: PRLR inhibition anticipated to be safe and well tolerated
ABS-201 PHL PRLR inhibition anticipated to be safe & well tolerated as supported by human genetics I. Reduced/Loss of PRL or PRLR Signaling Dominant negative II. PRLR loss-of-function Postpartum agalactia III. Newey & Phil , 2013 NEJM Otherwise in good health: o No apparent impact on fertility I. o
Slide 11: Phenotypic change over time
ABS 201 | PHL Top head view of Stumptailed Macaque’s showing Terminal hair count ”Thick phenotypic change over time Hairs” in prior bald areas ASSESSMENT AREA = R1 TREATMENT POST-TREATMENT 300 BASELI NE 12 WEEKS 28 WEEKS 6 MONTHS 2 YEARS 4 YEARS 250 # Thick Hairs 200 MALE 150 100 50 FEMALE 0 4 8 12
Slide 12: ABS-201 shows superior efficacy vs 5% topical minoxidil
ABS-201 PHL ABS-201 shows superior efficacy vs 5% topical minoxidil in 21d hair regrowth model Hair Growth Score ABS-201 vs minoxidil/untreated/isotype **p<0.05; ***p<0.0001 - 2way ANOVA Administration: mAbs i.p. biweekly; Minoxidil topical daily COPYRIGHT © 2026 ABSCI CORPORATION | ALL RIGHTS RESER
Slide 13: Men without PHL carry genetically lower PRLR levels
ABS-201 | PHL Men without PHL carry genetically lower PRLR levels; severity of PHL scales with PRLR expression ← keeps hair loses hair → odds ratio per 1 SD PRLR No balding 0.82 grade 1 vs 2–4 · 53,076 vs 112,573 p = 3.6×10⁻⁵ Moderate balding 1.18 grades 2–3 vs 1 · 82,348 vs 53,076 p = 1.4×10⁻³ Seve
Slide 14: cis-MR + colocalization supports a causal relationship
ABS-201 | PHL PRLR cis-MR colocalization plot Expression p = 3×10⁻⁹¹ cis-MR + Hair loss p = 8.7×10⁻⁸ Causal estimate p = 4×10⁻⁶ Shared signal PP.H4 = 0.96 colocalization supports a causal relationship for PRLR in Pattern Hair Loss o More PRLR, more proportional balding. o The neighboring genes show
Slide 15: ABS-201 in human ex vivo culture study supports MOA
ABS-201 | PHL ABS-201 in human ex vivo culture study supports MOA in human scalp follicles Frontotemporal ABS-201 significantly prolongs anagen/inhibits catagen and Hairline Region stimulates hair matrix proliferation Scalp Punch Microscopic hair cycle staging 100 80 % of HFs in each hair 60 cycle s
Slide 16: Additional ABS-201 ex vivo study findings
ABS-201 | PHL Additional ABS-201 ex Stem Cell Protection & Maintenance vivo study found: (↑K15+ Stem Cells) o Prolonging anagen phase and blocking catagen, thereby inhibiting telogen effluvium ↑ FGF7 and IGF1 Expression o Protecting and promoting hair follicle stem cells and and restoring CD34+ prog
Slide 17: Phase 1/2a trial designed to provide readouts on safety, tolerability, PoC
ABS-201 | PHL Phase 1/2a trial designed to Single Ascending Dose provide readouts on safety, Cohort 1 Cohort 2 Cohort 3 Cohort 4 tolerability, and PoC in PHL 150mg IV 450mg IV 900mg IV 1800mg IV n=8 n=8 n=8 n=8 o Initiated December 2025 Design Elements: o All planned SAD cohorts dosed o Double-Blind
Slide 18: HEADLINE Trial Interim PK Data from SAD Cohorts
ABS 201 | PHL HEADLINE Trial Interim PK Data from SAD Cohorts Half life of >65 days at doses 1000 SAD Cohort 1 - 150mg i.v. SAD Cohort 2 - 450mg i.v. Mean Serum Conc. (mg/L) overcoming TMDD SAD Cohort 3 - 900mg i.v. SAD Cohort 4 - 1800mg i.v. 100 PK profile supports 10 targeted dosing interval of 2
Slide 19: Modeling of Interim PK Data Supportive of Convenient Dosing
ABS-201 | PHL Modeling of Interim PK Data Supportive of Convenient Dosing Profile ABS-201 current ~70% s.c. Bioavailability o Modeling of interim PK data from HEADLINE Trial supports the potential for 2–3 s.c. injections over six-month period (e.g., 400 mg/injection) to maintain> 90% Receptor Occupa
Slide 20: ABS-201 TPP aims to offer a new treatment category in PHL
ABS-201 | PHL ABS-201 TPP aims to offer 80 a new treatment category 70 Transplant + oral/topical in PHL based on efficacy and convenience 60 Efficacy (terminal hairs/cm2) Increased TAM o Novel, targeted regenerative hair follicle 50 mechanism 40 o Convenient, infrequent pulse therapy: 2-3 Potential
Slide 21: Consumer Research Commissioned by Absci
ABS-201 | PHL Consumer Research Commissioned by Absci 610 Participants: 306 Men | 304 Women Validates Market Potential *ALL PARTICIPANTS EXPERIENCING HAIR LOSS of ABS-201 UP TO UP TO 97% 88% Significant Unmet Need: EXTREMELY OR VERY Driven by psycho-social impacts (loss of confidence, self- LIKELY T
Slide 22: 80M Patient Funnel
ABS-201 | PHL Total PHL pts in the US 50M male, 30M female 80M Patient Funnel Income ≥$75K (US Census) Pts 17-69 yo ~39M >$25B ESTIMATED U.S. TAM Concerned / motivated about hair loss ~26M Strong interest in TPP* ~22 – 24M >$40B POTENTIAL GLOBAL TAM Strongest interest in TPP with premium price-to-pe
Slide 23: Development of ABS-201 in Endometriosis
ABS 201 | ENDOMETRIOSIS Development of ABS-201 in Endometriosis 1. Addresses Long-standing Unmet Medical Need and Poor standard of care Strong Biological And Clinical 2. Rationale: Including POC for PRLR mechanism in humans 3. Large, untapped market offers significant upside potential COPYRIGHT© 202
Slide 24: Hormonal therapies and surgical intervention treatment paradigm for Endo
ABS-201 | ENDOMETRIOSIS Hormonal therapies and surgical intervention make up the treatment paradigm for Endo Current Treatment Paradigm (US) • Endo is a chronic condition with currently no medical or surgical cure Total Estimated Endo patients • Roughly, 75% of patients are estimated to use opioids
Slide 25: PRLR antagonism is a novel MoA in Endometriosis
ABS-201 | ENDOMETRIOSIS PRLR antagonism is a novel and differentiated MoA in Endometriosis o Endometriotic lesions produce prolactin under estrogen/progesterone control. PRL and PRLR play a dual role in endometrial lesion development o Excess prolactin promotes lesion growth and pain response and se
Slide 26: PRL and PRLR increase during secretory phase
ABS-201 ENDO PRL and PRLR increase during secretory phase in healthy tissue, and is overexpressed in endometrium of patients with endometriosis Localization And Temporal Expression Of Prolactin PRL & PRLR is Elevated Ectopic Endometriotic Lesions Receptor In Human Endometrium P ES Me = menstrual P =
Slide 27: PRLR antagonism suppresses postoperative pain in female mice
ABS-201 ENDO PRLR antagonism suppresses postoperative pain in female mice and inhibits endometriosis interna formation Prolactin May Regulate Pain Responses via a Female- The Effects of Prolactin Receptor Blockade Selective Nociceptor-Specific Mechanism in a Murine Endometriosis Interna Model Depict
Slide 28: PRLR antagonism reduces lesion formation and pain
ABS-201 | ENDOMETRIOSIS PRLR antagonism reduces lesion formation and pain in endometriosis mouse model Prolactin inhibition decreases endometrial lesion formation in % Distance travelled at Week 7 female mouse interna (compared to baseline) 5 * * * Disease Score 4 3 2 lgG1 – w/o transplant 1 * 0 lgG
Slide 29: >$4.5B at peak sales
ABS-201 | ENDOMETRIOSIS ABS-201: a potentially differentiated profile targeting a large underserved market opportunity NOVEL TR EATMENT OPTI ON FOR ~ 9 M PATI ENTS I N THE U . S . AL ONE WI TH ENDOMETR I OS I S o Novel Mechanism: Non–sex- o Best-in-class Potential: steroid (peptide) hormone Superior
Slide 30: Clinical Strategy guided by a leading Endometriosis Advisory Board
ABS-201 | ENDO Clinical Strategy guided by a leading Endometriosis Advisory Board Joan-Carles Arce, MD, PhD Gaurang Daftry, MD Hugh Taylor, MD (Chair) Chief Scientific Officer & Linda Giudice, MD, PhD Chief Scientific Medical Anita O’Keefe Young Professor President of Physician Relations, Robert B.
Slide 31: Leading AI x Bio platform driving Phase readouts in next 24 months
CATALYSTS Leading AI x Bio platform driving 2 Phase 2 readouts in the next 24 months ABS-201 in PHL ABS-201 in ENDO o Ph1/2a Study initiated Dec o Ph1/2a Study initiated Dec 2025 2025 o Positive interim Safety, o Phase 2 initiation expected in Tolerability, and PK readout Q4 2026 June 2026 o Interim
Slide 32: Powered by Absci's Leading AI Platform
Powered by Absci’s Leading AI Platform CONFIDENTIAL COPYRIGHT © 2026 AND ABSCI PROPRIETARY CORPORATION | COPYRIGHT | ALL RIGHTS © 2026 RESERVED ABSCI CORPORATION | ALL RIGHTS RESERVED 32
Slide 33: Absci's de novo AI models deliver differentiated lead candidates
ABSCI | AI PLATFORM Absci's de novo AI models deliver differentiated lead candidates against challenging targets ABSCI DIFFERENTIATION • Address challenging targets – Only target structure or sequence required • Zero-Prior: Design against any epitope – No known protein binder required – Interrogate
Slide 34: We use AI to create novel and differentiated therapeutics
DE NOVO DESIGN We use AI to create novel & differentiated therapeutics EPITOPE-SPECIFIC DESIGN + EPITOPE INTERFACE OPTIMIZATION ENHANCED POTENCY AND MOA ABILITY TO ADDRESS DIFFICULT TARGET CLASSES, E.G. GPCRS ENABLING FEATURES: MULTI-VALENCY, pH-DEPENDENT BINDING POTENTIAL TO CREATE MEANINGFUL IP: 1
Slide 35: We create experimentally validated AI antibody design technology
AI PLATFORM Wet Lab Validation Wet Lab experiments confirm binding, epitope-specificity, We create developability, and function. Wet Lab data enable assessment experimentally of model performance improvements. validated AI Antibody Design (AbsciDesign) (1) Generate antibody technology that structure
Slide 36: Origin AI models design and optimize therapeutic antibodies
ABSCI | AI PLATFORM Origin AI models design and optimize therapeutic antibodies INPUT STRUCTURE & SEQUENCE DESIGN DESIGN SCORING & FILTERING OUTPUT LIBRARY Structure Design Sequence Design Design 1 Target Antigen DESIGN LIBRARY HCDR1: GFNIKDTY HCDR2: IYPTNGYT 1. DESIGN 1 Target HCDR3: SRWGGDGFYAMDY
Slide 37: Lab-in-the-loop ensures AI-designed antibodies translate into drug-ready candidates
ABSCI | AI PLATFORM Lab-in-the-loop ensures AI-designed antibodies translate into drug- ready candidates High Affinity • Screening team confirms binding to target and affinity Functional validation • Disease Biology team validates functional activity • Analytical Development team conducts developabi
Slide 38: Introducing Origin-1
ABSCI | AI ORIGIN-1 Introducing Origin-1: an AI platform for de novo antibody design against zero-prior epitopes Zero-Prior Epitope Targeting High Structural Fidelity Functional Antagonist Designed full-length mAbs against Cryo-EM confirmed designs at 3.0–3.1 Å AI-guided affinity maturation yielded
Slide 39: Absci's de novo AI models deliver differentiated lead candidates
ABSCI | AI PLATFORM Absci’s de novo AI models deliver differentiated lead candidates against challenging targets DE NOVO ANTIBODY DESIGN AGAINST DE NOVO ANTIBODY DESIGN AGAINST DE NOVO ANTIBODY DESIGN AGAINST TRANSMEMBRANE TARGET EPITOPE CHALLENGING ION CHANNEL CONSERVED HIV-1 "CALDERA" EPITOPE WITH
Slide 40: High-fidelity, customized datasets with our own AI models
ABSCI | AI PLATFORM We use high-fidelity, customized datasets with our own AI models to enrich for developable candidates Landscape-wide scoring enables selection of optimal sets of sequences Generates diverse, developable, and functional novel variants within a desired target product profile (TPP):
Slide 41: Success in AI-design and optimization of developability and pharmacological properties
ABSCI | AI PLATFORM Success in AI-design and optimization of multiple developability and pharmacological properties CROSS-SPECIES OPTIMIZATION MULTI-VARIANT SPECIFICITY PH-SENSITIVE BINDING Successful design of antibodies binding to Successful design of antibodies with Successful design of pH-sensit
Slide 42: Track record of industry-leading partnerships
ABSCI | PARTNERSHIP Track record of industry-leading partnerships AI DRUG CREATION™ PARTNERSHIPS 25+ PARTNERED PROGRAMS TO DATE OUR PARTNERSHIP FOCUS • Challenging Targets / Epitopes • Differentiated TPPs SUPPORTING COMPUTE COLLABORATIONS • Mono or Multi-specific Formats Trademarks, service marks or
Slide 43: Multilingual team with expertise in AI and drug creation
TEAM Leadership Team: ”Multilingual” team with expertise in AI Sean McClain Zach Jonasson, PhD Ransi Somaratne, MD Chief Medical Officer, Shelby Walker, JD Christine Lemke and drug Founder, CEO & Board Chief Financial Officer Chief Legal Officer SVP, Portfolio & Growth Director & Chief Business Offi
Slide 44: Absci at a Glance
SUMMARY Absci at a Glance 140 EMPLOYEES 77,000 SQUARE FEET o State-of-the-art wet and dry lab in Vancouver WA o Absci AI Research (AAIR) lab in NYC 10+ PARTNERS, INCLUDING o Innovation Centre in Zug Switzerland “Multi-lingual” AI + Drug Discovery expertise AI team drawing on experience 3 from tech l
Slide 45: Generative AI Re(Generative) Biology
Generative AI Re(Generative) Biology ABSCI CORPORATION 2026 ALL RIGHTS RESERVED
Slide 46: AI Case Study I
AI CASE STUDY I de novo design of an antibody that binds the Caldera region of HIV-1 trimer COPYRIGHT© 2026 ABSCI CORPORATION. ALL RIGHTS RESERVED. 46
Slide 47: De novo design antibody that binds to caldera region of HIV gp120
DE NOVO DESIGN de novo design antibody that binds to the highly conserved caldera region of HIV gp120 HIV gp120 trimer (open) No natural or synthetic antibody for HIV exists today because immune system cannot derive an Caldera antibody that is universally neutralizing against HIV Design challenge: c
Slide 48: HIV-Caldera: Determine inputs and design
De novo design HIV-Caldera: Determine inputs and design HIV Env trimer (open) HIV Env Trimer Challenge : • Highly glycosylated • Extremely high sequence diversity among isolates LCDR1 LCDR3 • High mutation rate at common neutralizing epitopes HCDR3 Model inputs: 1. Antigen structure 2. Framework of
Slide 49: 4 best structures selected from 10,000+ generated
DE NOVO DESIGN 4 best structures selected from 10,000+ structures generated by de novo model Light Heavy 17b Structure S1 HCDR3 Structure S4 HCDR3 Structure S3 HCDR3 Structure S2 HCDR3 COPYRIGHT © 2026 ABSCI CORPORATION | ALL RIGHTS RESERVED 49
Slide 50: Applied molecular dynamics simulation to de novo designed antibodies
CASE STUDY #1 Applied molecular dynamics simulation to de novo designed antibodies 2024 2024 2025 2025 2026 COPYRIGHT © 2026 ABSCI CORPORATION | ALL RIGHTS RESERVED 50
Slide 51: Enriched de novo library binds open, not closed, Env trimer
DE NOVO DESIGN Enriched de novo library binds open, not closed, Env trimer conformation in YSD Clade A Env trimer Clade B Env trimer Closed Open Closed Open HA tag Antigen COPYRIGHT © 2026 ABSCI CORPORATION | ALL RIGHTS RESERVED 51
Slide 52: SPR data demonstrate binding characteristics consistent with caldera binding
DE NOVO DESIGN SPR data demonstrate binding characteristics consistent with binding of caldera COPYRIGHT © 2026 ABSCI CORPORATION | ALL RIGHTS RESERVED 52
Slide 53: HIV-Caldera: SPR demonstrates no binding to GP120 monomer
DE NOVO DESIGN HIV-Caldera: SPR demonstrates no binding of de novo designs to GP120 monomer Hypothesis: 17b IgG 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 ASN3013 Key results: ASN3014 ü 1
Slide 54: HIV-Caldera: demonstrating AI de novo design for challenging target
DE NOVO DESIGN HIV-Caldera: demonstrating AI de novo design for challenging target 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
Slide 55: AI Case Study II
AI CASE STUDY II AI Optimization for pH sensitivity COPYRIGHT© 2026 ABSCI CORPORATION. ALL RIGHTS RESERVED. 55
Slide 56: AI lead optimization platform for 'smart biologics'
CASE STUDY #2 AI lead optimization platform for ‘smart biologics’ The Challenge The diversity of antibodies is vast, making it impossible for traditional methods to explore effectively. Absci Solution Our AI can search a space of ~1019, a million times larger than traditional methods, identifying fu
Slide 57: pH sensitivity may reduce toxicity and/or improve efficacy
CASE STUDY #2 pH sensitivity may reduce toxicity and/or improve efficacy of therapeutic mAbs Tumor specificity improves efficacy and reduces Disassociation in the endosome drives antibody ”on-target off-tumor” toxicities recycling and efficient clearance of soluble targets Binding occurs in the No b
Slide 58: Models identify pH sensitive Fab variants
CASE STUDY #2 Models identify pH sensitive Fab variants from the same lead for either indication 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
Slide 59: Hits reformatted as mAbs show desired binding profiles
CASE STUDY #2 Hits reformatted as mAbs 1000 show desired binding profiles SPR KD 7.4 / 5.8 100 10 SUMMARY 1 AI optimized leads achieves variants with pH 2024 2024 2025 2025 2026 0.1 sensitive binding up to 100x differential pH-sensitive leads had no liabilities for stability, 0.01 aggregation and po
Slide 60: Summarized platform case studies
CASE STUDIES Summarized platform case studies de novo Design 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 desi