Bioinvest General-purpose Predictive Trials

(BioinvestGPT)
10x more correct GO/NO-GO Decision-Making for Biopharma Decision-makers
Disclaimer
This Presentation does not constitute, or form part of, nor is it intended to communicate, any offer, invitation, inducement or recommendation to sell or issue, or any solicitation of any offer to purchase or subscribe for, any shares in
any entity affiliated with BioinvestGPT (the “Company” hereafter) in any jurisdiction nor shall it, or any part of it, or the fact of its distribution, form the basis of, or be relied on in connection with or act as any inducement to enter into,
any contract therefor.

Use of Assumptions: Preparation of this Presentation requires the Company to make certain estimates and assumptions. While the Company has used reasonable endeavours to ensure the estimates and assumptions are based
on the best available information but actual results could be materially different. The Company shall not be liable for any errors, inaccuracies or delays in the information, nor for any actions taken in reliance thereon. The Company
disclaims all warranties, express or implied, as to the accuracy or completeness of any of the content provided, or as to the fitness of the content for any purpose to the extent permitted by law. This Presentation is intended for
information purposes only and is provided without any warranty of any kind, either expressed or implied. Due to various risks and uncertainties, actual events or results or actual performance of the Company may differ materially
from any opinions, forecasts or estimates reflected or contemplated in this Presentation. These statements are not guarantees of future performance, condition or results. There can be no assurance that future results or events will
be consistent with any such opinions, forecasts or estimates. The past performance of the Company is not a reliable indication of the future performance of the Company. No representation or warranty is made as to the
achievement or reasonableness of, and no reliance should be placed on, such forward looking statements. The Company undertakes no duty or obligation to publicly update or revise the forward‐looking statements or other
information contained in this material. While some information used in this material may have been obtained from various published and unpublished sources considered to be reliable, the Company neither guarantees its accuracy
or completeness nor accepts liability for any direct or consequential losses arising from its use. Amounts and percentages may reflect rounding adjustments and consequently totals may not appear to sum.

Neither the Company, nor any of its members, directors, officers, agents, employees or advisers intend or have any duty or obligation to supplement, amend, update or revise any of the opinions, forward-looking statements or
estimates contained in this Presentation. No statement in the Presentation is intended to be, or intended to be construed as, a profit forecast or profit estimate or to be interpreted to mean that earnings per Company share for the
current or future financial years will necessarily match or exceed the historical earnings per Company share. Any investment in the Company is speculative, involves a high degree of risk, and could result in the loss of all or
substantially all of their investment. Results can be positively or negatively affected by market conditions beyond the control of the Company or any other person. As a result, no undue reliance should be placed on such
statements.
FDA 2025 Guidance of GO/NO-GO Decision-Making AI Performance Metrics

In 2024, Top 20 big pharma and major shareholders have trusted BioinvestGPT using those FDA-recommended
performance metrics in an ex-ante, real-world GO/NO-GO decision-making setting.

* Source: https://www.fda.gov/regulatory-information/search-fda-guidance-documents/considerations-use-artificial-intelligence-support-regulatory-decision-making-drug-and-biological
Figures are AS IS subject to updates from time to time without warranty of any kind.
All rights reserved. | Confidential, not intended for further distribution. | 1
Ex-ante Clinical Prediction Validation

368 real-world ex-ante clinical predictions validated by big pharma in 2024.

99% CI 100% 96% ex-ante accuracy

Having surpassed the 135 “ex-ante” means “before readout”. Ex- Legally timestamped and delivered in advance to top 5 big pharma
real-world ex-ante clinical ante clinical benefit prediction is the shareholders & top 20 big pharma companies, BioinvestGPT clinical
predictions threshold to most rigorous validation method in the benefit predictions achieved 96% ex-ante prediction accuracy
trust BioinvestGPT’s ex- industry. BioinvestGPT is the world’s (353/368 validated prospective predictions; 99% CI; P < 0.0001)
ante clinical benefit only AI platform 100% validated by in 2024 and 93.9% ex-ante prediction accuracy (682/726 trials)
prediction accuracy with ex-ante clinical benefit predictions since Sept. 2022. (Visit https://data.bioinvestgpt.com/ for details).
99% Agresti-Coull (visit https://data.bioinvestgpt.com/ for
confidence interval. details).

Ex-ante Clinical Prediction Audit Timeline

Prediction Legalization Date: 14 Feb 2024

Emraclidine Partner Delivery Rate: 15 May 2024
EMPOWER-1 &
Declassified Date: 10 Nov 2024
EMPOWER-2
Readout Date: 11 Nov 2024
Figures are AS IS subject to updates from time to time without warranty of any kind.
All rights reserved. | Confidential, not intended for further distribution. | 2
 FDA-compliant Ex-ante GO/NO-GO Decision-Making Performance Metrics WY2024
As the world’s only universal clinical benefit/risk diagnostics platform for any-stage FIC/BIC assets, BioinvestGPT’s performance
in the real GO/NO-GO decision-making environment has been thoroughly evaluated using diagnostic metrics in 100% real-world
prospective ex-ante drug clinical benefit predictions.
++Reliability ++Revenue --Impairment ++ROI
GO/NO-GO bias-free rate GO coverage rate NO-GO coverage rate GO/NO-GO cost-effect rate
Accuracy
F1-Score Sensitivity Specificity Critical Success
96%
(99% CI; 95% 97% 95% Index (CSI)
two-sided P < 0.0001) (99% CI; (99% CI; (99% CI;
96% of BioinvestGPT- two-sided P < 0.001) two-sided P < 0.001) two-sided P < 0.001) 90%
informed GO/NO-GO decisions Both BioinvestGPT-informed GO (99% CI;
BioinvestGPT-informed correct BioinvestGPT-informed correct
are correct for any-stage decisions and NO-GO decisions are two-sided P < 0.001)
GO decision identifies 97% of all NO-GO decision identifies 95% of
FIC/BIC assets, nearly 10 times almost equally likely to be correct, the possible blockbuster FIC/BIC all the possible lackluster FIC/BIC BioinvestGPT-informed GO decisions
higher than the industry’s even though blockbuster drugs are assets. assets. enable the most cost-efficient
current best level (5-10%). much rarer than lackluster drugs. blockbuster drug development.

GO/NO-GO reliability rate GO correct rate NO-GO correct rate

First-in-class Ratio
Diagnostic Odds Ratio (DOR) Positive Predictive Negative Predictive
663 Value (PPV) Value (NPV) 85%
(261/307 validated
(99% CI;
two-sided P < 0.001)
92% 98% prospective predictions)
(99% CI; (99% CI; 85% of BioinvestGPT-informed
Both BioinvestGPT-informed GO decisions and NO-GO two-sided P < 0.001) two-sided P < 0.001) GO/NO-GO decisions have been made
decisions are almost always likely to be correct (663 times
92% of BioinvestGPT-informed 98% of BioinvestGPT-informed NO-GO for the most challenging and valuable
more likely than being incorrect), even though blockbuster drugs
GO decisions result in launches decisions result in cost savings from first-in-class assets.
are much rarer than lackluster drugs.
of blockbuster drugs. not advancing lackluster drugs.

* All confidence intervals (CIs) are Agresti-Coull. BioinvestGPT Performance Data: https://analysis.bioinvestgpt.com/
Figures are AS IS subject to updates from time to time without warranty of any kind.
All rights reserved. | Confidential, not intended for further distribution. | 3
Ex-Ante GO/NO-GO Decision Correct Rate WY2024
BioinvestGPT’s ex-ante clinical benefit prediction performance has been proven to increase FIC/BIC GO/NO-GO decision-making accuracy to
96% across all four biopharma-prioritized therapeutic areas, with 92% GO decision accuracy (vs 5-10% industry standard) and 98% NO-
GO decision accuracy at scale for both the preclinical stage and clinical stages.

Biopharma-Prioritized Cardiovascular, Renal

Therapeutic Areas Oncology Immunology Neuroscience
and Metabolic

Overall GO/NO-GO 96% 97% 100% 93% 93%

(306/319 validated (114/117 validated (63/63 validated (55/59 validated (74/80 validated
Decision Correct Rate* prospective predictions) prospective predictions)
prospective predictions; prospective predictions) prospective predictions)
Accuracy (99% CI) 99% CI;
two-sided P < 0.001) **

NO-GO correct rate 98% 100% 100% 94% 97%

(196/200 validated (78/78 validated (28/28 validated (30/32 validated (60/62 validated
Negative Predictive prospective predictions; prospective predictions) prospective predictions) prospective predictions) prospective predictions)
Value, NPV (99% CI) 99% CI;
two-sided P < 0.001)

GO correct rate 92% 92% 100% 93% 78%

(110/119 validated (36/39 validated (35/35 validated (25/27 validated (14/18 validated
Positive Predictive prospective predictions; prospective predictions) prospective predictions) prospective predictions) prospective predictions)
99% CI;
Value, PPV (99% CI) two-sided P < 0.001)

* As of January 2 2025, 368 clinical readouts have been announced to validate the corresponding BioinvestGPT prospective clinical benefit predictions for all therapeutic areas, resulting in 96%
overall accuracy (353/368 validated prospective predictions; 99% CI; P < 0.0001), an 12% improvement over the 84% overall accuracy (146/173 trials) achieved by BioinvestGPT for 2020-2022.
BioinvestGPT Performance Data: https://analysis.bioinvestgpt.com// in which Tab T0 is for all indications and Tabs T2-T6 are for biopharma-prioritized therapeutic areas.
** All confidence intervals (CIs) are Agresti-Coull.
Figures are AS IS subject to updates from time to time without warranty of any kind.
All rights reserved. | Confidential, not intended for further distribution. |4
 Compelling evidence shows that BioinvestGPT-informed ex-ante GO/NO-GO decision-making
would be 663 times more likely to be correct than incorrect.

BioinvestGPT for universal clinical benefit prediction (DOR 663) even outperforms PCR Test for COVID-19 (DOR 316) in terms of
maximizing true predictions (correct GO/NO-GO decisions) and minimizing false predictions (incorrect GO/NO-GO decisions).

Conventional Drug GO/NO-GO

BioinvestGPT COVID-19 PCR Test
Decision-making Approach
Diagnostic Odds
Ratio (DOR) 663 *
316 **
0.06 ***

(2024WY) (99% CI;

two-sided P < 0.001)

The confidence in Interpretation: BioinvestGPT-informed Interpretation: As the most reliable Interpretation: Conventional GO/NO-GO
BioinvestGPT for prospectively GO/NO-GO decision-making is 663 times and adopted test for diagnosing decision-making is 94% more likely to
correctly predicting both more likely to correctly identify COVID-19 worldwide, PCR test is 316 falsely initiate the clinical development
successes and failures blockbuster FIC/BIC assets (true times more likely to correctly identify of a commercially insufficient FIC/BIC
positives) and rule out commercially COVID-19 patients (true positives) and assets (false positives) and falsely
Positive Likelihood Ratio / insufficient FIC/BIC assets (true rule out healthy population (true discontinue the clinical development of
Negative Likelihood Ratio negatives), compared with the chance of negatives), compared with the chance of a blockbuster FIC/BIC asset (false
making an incorrect decision (false making an incorrect diagnosis (false negatives), compared with the chance of
positives and false negatives). positives and false negatives). making a correct decision (true positives
and true negatives).

* All confidence intervals (CIs) are Agresti-Coull. BioinvestGPT Performance Data: https://analysis.bioinvestgpt.com/
** Arshadi et al. (2022). Diagnostic Accuracy of Rapid Antigen Tests for COVID-19 Detection: A Systematic Review With Meta-analysis. Frontiers in Medicine, 9.
https://www.frontiersin.org/journals/medicine/articles/10.3389/fmed.2022.870738
*** The DOR for current drug GO/NO-GO decision-making approach is estimated based on the more optimistic 10% nomination rate (https://www.boehringer-
ingelheim.com/science-innovation/human-health-innovation/drug-discovery-boehringer-ingelheim).
Figures are AS IS subject to updates from time to time without warranty of any kind.
All rights reserved. | Confidential, not intended for further distribution. | 5
Value For BigPharma: Bottomline-Maximizing Partnership (++Revenue & --Impairment)

Guaranteed blockbuster drug discovery and internal expertise enhancement with unlimited BioinvestGPT virtual trials.

Accelerated Global Patients

clinical development
of GO candidates in ++Revenue Market Leadership
which 9 out of 10 will Shareholders
reach blockbuster
launches

GO --Impairment
Blockbuster
Internal Drug Candidate Discontinued
Knowledge Blockbuster Hypothesis Screening further development
Internal/Public
Experimental
Graphs Drug Candidate with BioinvestGPT’s NO- of NO-GO candidates
Continously Hypothesis almost all of which
Data
Enhanced with Generation
Unlimited SoC-controlled
Virtual Clinical Trials with
GO would indeed be
BioinvestGPT underperforming drugs
Actionable Rationale

++Expertise Enhancement

Figures are AS IS subject to updates from time to time without warranty of any kind.
All rights reserved. | Confidential, not intended for further distribution. | 6
 Value For BigPharma: Use Cases

BioinvestGPT is helping pharma partners shift from a “fail often and fail late” old paradigm to a “succeed often and
succeed early” new reality now.
Discover Blockbuster Assets Acquire Blockbuster Drugs Expand Blockbuster Indications Prolong Blockbuster Sales

Candidate M&A Indication Competitive

Nomination * Screening Expansion Analysis
92% of GO decisions Glittering like morning dew before BioinvestGPT has a compelling track record of BioinvestGPT virtual patients are proven to
recommended by BioinvestGPT evaporating with the rise of the sun, bio exhaustively identifying clinically meaningful differentiate competing drug candidates via head-
lead to blockbuster drug first-in-class M&As have long been risky indications for FIC/BIC assets across all the to-head virtual clinical trials that dynamically
launches. endeavors with only a 30% success rate major therapeutic areas. rank their comparative clinical benefits.
for decades.
98% of NO-GO decisions For GLP1R agonists, BioinvestGPT maintains For example, BioinvestGPT’s metabolic disorder
recommended by BioinvestGPT With a proven track record of achieving a 100% ex-ante prediction accuracy for all the virtual patients have maintained 100% ex-ante
prevent costly investments in 90% success rate in both GO/NO-GO indication expansion clinical trials of clinical benefit prediction accuracy when it
future underperforming drugs. M&A decisions, BioinvestGPT could Semaglutide and Tirzapertide: Alzheimer's comes to running head-to-head virtual clinical
generate hundred-billion-dollar value for Disease (1520, 1521), NASH (1513, 1523), trials and ranking all the clinical-stage anti-
big pharma companies. HFpEF (1512, 1684), Knee Osteoarthritis (1272), obesity FIC/BIC assets (MariTide, CagriSema,
MACE (973, 1519), Diabetic Eye Diseases and S309309, monlunabant, semaglutide, tirzepatide,
In 2024, BioinvestGPT could have T2DM (1534), Heart Disease with T2DM (1532), VK2735, CT996, GSBR1290, ZP8396, TERN601,
saved big pharma companies such as Peripheral Arterial Disease and T2DM (1531), danuglipron, amycretin, orforglipron, AZN5004,
Novo Nordisk (1682), AbbVie (1457, Heart Failure (1525, 1526), Chronic Kidney etc.) in terms of efficacy, tolerability, muscle
1458), and Incyte (1575, 1584) at least Diseases (1515, 1524), Obstructive Sleep Apnea preservation, cardio protection, and possibly any
$12 billion in cash. (716), in addition to Obesity (992, 1517, 1518, other yet-to-be-investigated endpoints/indications.
687, 1514).

Figures are AS IS subject to updates from time to time without warranty of any kind.
All rights reserved. | Confidential, not intended for further distribution. | 8
Value for BigPharma: Blockbuster R&D Productivity
BigPharma could achieves 100x R&D productivity for blockbuster drugs compared with other big pharma
companies (10x faster blockbuster launches at 10% R&D cost)

Blockbuster BigPharma
Drugs /Year
10 BioinvestGPT-based blockbuster-discovering
BigPharma
virtual P3 head-to-head clinical trial cost
$ 1B/Blockbuster Drug
Partners
100% BioinvestGPT-based blockbuster-discovering
virtual P3 head-to-head clinical trials &
8 BioinvestGPT-derisked human clinical trials
Bi 10 10 Blockbuster Drugs / Year
6
oi
nv
0x
es
tG
PT
.
4
Industry’s Average

2 $ 10B / Blockbuster Drug

100% Human-based blockbuster-discovering
1
real P3 head-to-head clinical trials

$11B
1 Blockbuster Drug / Year
$1B $3B $5B $7B $9B
R&D Cost/
Blockbuster Drug

Sources: Mullard, A. How much do phase III trials cost?. Nat Rev Drug Discov 17, 777 (2018). https://doi.org/10.1038/nrd.2018.198
Figures are AS IS subject to updates from time to time without warranty of any kind.
All rights reserved. | Confidential, not intended for further distribution. | 10
Problem
The biopharmaceutical industry’s benchmark of clinical success rate for new drugs remains at 5% - 10% from
1980 to present.

Once we get it into human beings we have about a 5-10%

success rate... If you look industry-wide it is about 5%.*
5-10%
THE CLINICAL SUCCESS RATE OF NEW DRUGS

Despite remarkable advancements in life

science and biotechnology, the clinical
success rate of investigational new drugs
for all human diseases remains at 5-10% on
average for decades.

* Vas Narasimhan, et al. the Science and Business of Innovative Medicines, a16z Podcast, 2019 Figures are AS IS subject to updates from time to time without warranty of any kind.
All rights reserved. | Confidential, not intended for further distribution. | 11
Cause

Experimental data are poor-quality sources of clinical-scale causal knowledge.

Too siloed Too unreliable Too unstructured

Ligand-protein interaction data cannot 70% biomedical experimental data Past clinical trial data cannot
generalize to new ligand-protein
are irreproducible. generalize across even the most
interactions (Mastropietro et al. 2023
similar trials, let alone new trials
Nature Machine Intelligence**). (Baker et al. 2016 Nature*). (Checkroud et al. 2024, Science***).

*Baker, M. 1,500 scientists lift the lid on reproducibility. Nature 533, 452–454 (2016). https://doi.org/10.1038/533452a
**Mastropietro, A., Pasculli, G., & Bajorath, J. (2023). Learning characteristics of graph neural networks predicting protein–ligand affinities. Nature Machine Intelligence, 5(12), 1427–1436.
https://doi.org/10.1038/s42256-023-00756-9
*** Chekroud, A. M., Hawrilenko, M., Loho, H., Bondar, J., Gueorguieva, R., Hasan, A., Kambeitz, J., Corlett, P. R., Koutsouleris, N., Krumholz, H. M., Krystal, J. H., & Paulus, M. (2024). Illusory generalizability of clinical
Figures are AS IS subject to updates from time to time without warranty of any kind.
All rights reserved. | Confidential, not intended for further distribution. | 12
prediction models. Science, 383(6679), 164–167. https://doi.org/10.1126/science.adg8538
 Solution

Genome is the optimal source of fully working clinical-scale causal knowledge.

Unified Reliable Structured

2% coding genome
98% noncoding genome

A/T/C/G is the unified programming A/T/C/G contains all the causal knowledge of how
language of life that generalizes to every human body works that have been constantly used to A/T/C/G constitutes both the 2% coding genome
living and extinct species over the past 3.7 execute life function every second of a human’s life. and the 98% non-coding genome that enables a
billions years on earth. structured understanding of multi-scale
mechanisms from genotype to phenotype.

Figures are AS IS subject to updates from time to time without warranty of any kind.
All rights reserved. | Confidential, not intended for further distribution. | 13
 Feasibility

BioinvestGPT and embryogenesis use the data-free approach to building human bodies.

BioinvestGPT Embryogenesis
uses genome as the uses genome as the
reliable source of exclusive source of
whole-body causal whole-body causal
knowledge to build a knowledge to build a
virtual patient. human body.

No biomedical experimental data are required for building human bodies.

Figures are AS IS subject to updates from time to time without warranty of any kind.
All rights reserved. | Confidential, not intended for further distribution. | 14
 Our Advantage

The pharma industry’s best knowledge graphs are just low-dimensional projections/
shadows of the genome knowledges extracted by embryogenesis and BioinvestGPT.

Industry-leading
biomedical BioinvestGPT
knoweldge graphs human-equivalent
(grounded in truth yet virtual patients that
prone to illusions of accurately capture
non-existing causation) real complex causal
relationships in
human body.

Biomedical experimental data sample shadows of physiology and pathogenesis.

Photo Source: https://pin.it/6SSXnDilu
Figures are AS IS subject to updates from time to time without warranty of any kind.
All rights reserved. | Confidential, not intended for further distribution. | 15
Our Approach

Data-independent direct extraction of clinical-scale causal knowledge from genome.

Mathematical genome Clinical-scale Human-equivalent

understanding causal knowledge virtual patients

No biomedical experimental data are required for building BioinvestGPT.

Figures are AS IS subject to updates from time to time without warranty of any kind.
All rights reserved. | Confidential, not intended for further distribution. | 16
Data-Independent Head-to-Head Virtual Clinical Trials

Prior data independence makes BioinvestGPT super-reliable when it runs complex head-to-head
virtual clinical trials for first-in-class assets with > 92% ex-ante validated prediction accuracy.
Inputs Drug MoA Trial Design
(MNPI-free public source, prior
clinical data not necessary) Drug Target Indication
Target Modulation Inclusion/Exclusion Criteria
Drug Modality Primary Endpoints
Drug Delivery Secondary Endpoints
Mono/Combo-therapy Control Type
Treatment Type Trial Phase

Simulation
(virtual clinical trials) BioinvestGPT

Outputs
(actionable and compliant
Technical Clinical Prediction Commercial Clinical Prediction
recommendation for making
(whether primary/secondary endpoints could be (whether clinical benefits are superior over
investment decisions)
met with statistical significance) competitors for commercial successes)

No prior clinical data are required for reliably using BioinvestGPT.

Figures are AS IS subject to updates from time to time without warranty of any kind.
All rights reserved. | Confidential, not intended for further distribution. | 17
No Need of Prior Clinical Data for Clinical Trial Simulations

BioinvestGPT predicts the concrete clinical benefits (not probability of success) for novel
FIC/BIC assets from Preclinical to Phase 3 without needing prior data as inputs.
Preclinical-Stage Novel Assets Clinical-Stage Novel Assets

Target Lead Pre-IND Phase 1 Phase 2 Phase 3

placebo-controlled virtual phase 3 simulation placebo-controlled virtual phase 3 simulation

using hypothetical trial design using real/hypothetical trial design

active-controlled virtual phase 3 simulation active-controlled virtual phase 3 simulation

using hypothetical trial design using real/hypothetical trial design

No need of any prior clinical data as inputs No need of any prior clinical data as inputs
(which is also unavailable at preclinical stages)

No prior clinical data are required for reliably using BioinvestGPT.

Equal simulation reliability regardless of the actual development stage

with data-independent inputs (drug MoA + trial design) and stage-
indpendent outputs (technical and commercial clinical benefits).

Figures are AS IS subject to updates from time to time without warranty of any kind.
All rights reserved. | Confidential, not intended for further distribution. | 18
 Workflow

BigPharma initiates job requests, then BioinvestGPT delivers clinical predictions.

BigPharma BioinvestGPT

Specification Submission Simulation Delivery

All the necessary design parameters All the initial job requests from different BioinvestGPT will simulate placebo- Clinical prediction packages will
required for a clinicaltrials.gov study teams could be aggregated into a controlled and active-controlled be delivered as attachments via
record should be specified in job centralized database and further virtual Phase 3 clinical trials for each encrypted email to the point of
requests to meet the following needs: consolidated and confirmed by a received job request, in order to contact.
steering committee as the point of generate:
Portfolio Team select clinical candidates contact.
Technical Clinical Prediction
Partnering Team source external deals The final job requests can be prepared Whether key study endpoints (efficacy
and submitted to BioinvestGPT as and toxicity) will be met with statistical
attachments via encrypted emails. significance.
Alternatively, cloud-based access to the
Commercial Clinical Prediction
final job requests can be granted to
BioinvestGPT. Whether clinical benefits are
superior/non-inferior/inferior to
competing drugs and/or SoC treatments.
Figures are AS IS subject to updates from time to time without warranty of any kind.
All rights reserved. | Confidential, not intended for further distribution. | 19
Capacity

BioinvestGPT can run 3000+ head-to-head phase 3 virtual clinical trials

of unlimited scope in the next 12 months.

1-7 days / trial, max 7 trials /day Drug MoA

Indication
By running hundreds of BioinvestGPT phase III head-to-head virtual clinical
Modality
trials in 2025 (each takes 1-7 days), BioinvestGPT’s capacity is sufficient to
meet the needs of both the portfolio team for selecting internal clinical Delivery
candidates and the partnering team for sourcing external deals, such that
Combotherapy
92% of BioinvestGPT-recommended GO decisions could lead to blockbuster
drug launches and 98% of BioinvestGPT-recommended NO-GO decisions Stratification
could prevent costly investments in future underperforming drugs.
Control Arm

...
Figures are AS IS subject to updates from time to time without warranty of any kind.
All rights reserved. | Confidential, not intended for further distribution. | 20
BioinvestGPT Leadership

BioinvestGPT’s virtual clinical trial agents is 10x more than conventional clinical CROs.

2
Human Leaders

Bragi and Idonae have been co-working as the

leadership team for 12+ years, through the full
Invented lifecycle of BioinvestGPT from development,
& validation, to commercilization.
Manage

2,000,000+
Clinically Validated Human-equivalent Virtual Patients
(prospectively validated via prior-to-readout data-independent simulations
of 1800+ the most challenging human clinical trials sponsored by
biopharma companies globally)

Figures are AS IS subject to updates from time to time without warranty of any kind.
All rights reserved. | Confidential, not intended for further distribution. | 21
 Early blockbuster drug discovery with
BioinvestGPT. 10x success rate at 10% R&D cost

Use Cases: www.bioinvestgpt.com Ex-ante Data: data.bioinvestgpt.com