Nexscient, Inc. · OTC:NXNT
Nexscient — Investor Presentation (April 2026)
38 slides from Nexscient's investor presentation, April 2026
Slide 1: Engineering the Human Data Foundation of Embodied Intelligence
Delivering the next dimension of AI Engineering the Human Data Foundation of Embodied Intelligence Deep Dive Rev 2.0
Slide 2: Vertically Integrated System for Embodied Intelligence
Nexscient, Inc. Vertically Integrated System for Embodied Intelligence Pioneering the Data & Model Infrastructure for the $500 Billion Robotics Economy Ticker Symbol: Nexscient, Inc. [OTCQB: NXNT] Isl\ NotebookLM
Slide 3: Investment Thesis: The Infrastructure Layer for Physical AI
Investment Thesis: The Infrastructure Layer for Physical Al "' "'A ::l.!J@ The Market The Critical The Nexscient The Operational The Investment Opportunity Bottleneck Solution Moat Vehicle Embodied Al is the Robotics is stalled We provide the Through the Public market foundation of AGI by a 'Data Sc
Slide 4: The Next Frontier: From Generative AI to Physical AI
The Next Frontier: From Generative Al to Physical Al Traditional Al (GenAI) Embodied Al (Nexscient Focus) •- - --- ---- --- - _._____ _ .. "I • :--.,....� ·=- ' 0-....> ..... � --· � ,.,.....,. /:l,9.l l -\ ·------ •• ... • ....• . �J......._ .··.··· � ...... · · ·.,: --....:::---... Input: Interne
Slide 5: The Macro Shift from Digital Text to Embodied Intelligence
@000TO NBO The Macro Shift from Digital Text to Embodied Intelligence Generative Al (LLMs) Physical Al (Embodied Intelligence) Inputs: Text / Images @®- A inputs: Vision & Environment 2 2 Output: Digital Content && > © output: Physical Action 3 >> @ _ data Source: The Web (Scraped) &&$ > Data Source
Slide 6: The Critical Bottleneck: Hardware is Ready, Data is Not
A Industrial Precision Next-Generation Engineering The Critical Bottleneck: Hardware is Ready, Data is Not Widespread utility is stalled by a lack of high-quality, diverse training data. i al ale aM Te = Lab Scarcity High Acquisition Costs Zero Diversity (SIP! FF, 2f; ie pal ae y —K OG bf bs im be =
Slide 7: The Barrier to Scale: The Robotics Data Bottleneck
The Barrier to Scale: The Robotics Data Bottleneck Data Volume Comparison E LLM Training Data 15 Trillion Tokens © The Data Scarcity Gap E Robotics Real-World Data 2.4 Million Samples (Max) The Data Scarcity Gap: Lab data is expensive; simulation lacks real- world physics. The Generalization Gap: Cu
Slide 8: Three Structural Barriers Blocking the Path to AGI
Three Structural Barriers Blocking the Path to AGI \ i] S Le 1. The Data Scarcity Gap 2. The Generalization Gap 3. The Embodiment Internet data (text/images) does not Existing models are brittle and Diversity Gap translate to physical action. Current overfitted. Without high-volume, Hardware fragmen
Slide 9: The Humanoid Revolution
The Humanoid Revolution Has Arrived, But the Minds Are Missing Hardware has outpaced intelligence. While companies like Tesla, Figure, and Unitree have solved the mechanical engineering of humanoid robots, the industry faces a critical failure point: “Dumb Robots.” The Alpha Layer. Embodied intellig
Slide 10: Capturing Value in a $500 Billion Market
• 0 Capturing Value in a $500 Billion Market Strategic value capture through recurring utility, not one-time sales. Data Licensing: o-- ....... r---o Custom Collection: Selling access to Client-specific ,,,,,, "---.../ the VLA BaseMind campaigns (e.g., (production-ready 'Collect 5,000 hours Data is
Slide 11: Investment Summary: The Utility for the Robotic Age
Investment Summary: The Utility for the Robotic Age • Market: Physical AI is the next trillion-dollar wave, moving beyond LLMs. • Problem: Data is the bottleneck; real-world robots need real-world memory. • Solution: A scalable, end-to-end operational engine (Senscient TaskAlpha + Flipside AI). • As
Slide 12: A Vertically Integrated Engine for Physical AI
A Vertically Integrated Engine for Physical Al Nexscient is building a comprehensive ecosystem spanning high-quality training data, software platforms, infrastructure, and applied Al. We serve enterprise and emerging technology customers worldwide. STRATEGY (The Head) ENGINEERING (The Hub) OPERATION
Slide 13: Defining the Niche
Defining the Niche Embodied Intelligence & VLA Language What is VLA? Vision-Language-Action models integrate visual perception, linguistic understanding, and physical action into a single system. What is Physical Al? Al systems designed to interact with the physical world, not just digital text. The
Slide 14: Why VLA Models Demand 'Ground Truth' Over Simulation
Why VLA Models Demand ‘Ground Truth’ Over Simulation Simulation / Lab Data (The Old Way) TaskAlpha Data (The Senscient Way) li : “ ~ ) € NEF Sterile, limited physics, “uncanny valley” of motion. Messy, chaotic, real-world physics, complex dexterity. The Science: Vision-Language-Action (VLA) models r
Slide 15: VLAs are Basis of Robot Foundation Models (RFMs)
VLAs are Basis of Robot Foundation Models (RFMs) RFMs are the "universal intelligence layer" for robots, designed to generalize across tasks and embodiments. OpenVLA Whole-Body Control Dexterous Manipulation -----➔ I I ' I I -�- -�-- I - - ... ... I �- • I Vision-Language-Action. Links visual Locomo
Slide 16: Robot Foundation Models: The Building Blocks of AGI
Robot Foundation Models (RFMs): The Building Blocks of AGI Just as GPT transformed language, RFMs serve as the universal intelligence layer for physical agents. Senscient isn't just selling ingredients; we are selling the recipe. = I ernargorit Capabilities: OpenVLA . These models enable zero-shot V
Slide 17: Strategic Advantage: Vertical Integration & The Data Flywheel
Strategic Advantage: Vertical Integration & The Data Flywheel Supervise (Flipside AI) Industrial supervision turns raw video into Ground Truth Capture Deliver (Senscient TaskAlpha) (Senscient Cortex) Scalable real-world Production-grade data supply , assets for customers 'o ,J ,.,,..,..,--..,.-';,-�
Slide 18: The Full Lifecycle Workflow
The Full Lifecycle Workflow. DATA FLOW> QA GATEWAY> CUSTOMER INTEGRATION> l l CAPTURE & ENRICHMENT ENGINEERING & QA PRODUCTION DELIVERY (7DVN$OSKD) (Flipside AI) (Senscient Cortex) Raw data col lection, deep Curation pipelines and "Braintrust" High-integrity, training-ready annotation, and labeling
Slide 19: An End-to-End Engine for Physical AI
An End-to-End Engine for Physical Al 1. Capture (TaskAlpha) 2. Refine (Flipside Al) Charter: The distributed "gig Charter: The operational economy" network capturing SENSCIENT Al muscle ensuring Ground-Truth CORE raw video from thousands of Humanot<STr3 ining 03UI BellavOn,alRObotics Labeling and Hu
Slide 20: The Engineering Hub: Senscient AGI
The Engineering Hub: Senscient AGl SUBSIDIARY OF NEXSCIENT, INC. I SINGAPORE • Data Engineering: STROCTUR:f:O - DATA COLLECT[ON -- • The Bridge: Specialized structured --J- GIJAlffi'·COtffllOllEO OOIIJTION Connecting raw data data collection and 2.8S YLA IIOfll(flOII lKTEORATlOM collection with qua
Slide 21: Operational Layer: Flipside AI
Operational Layer: Flipside Al End-to-End Control of the Al Data Lifecycle. Senscient creates the supply (TaskAlpha), but Flipside Al creates the quality. We acquired Flipside to own the operational machinery required for production-grade Al. Capture __ (Raw FPV Video) Supervision __""" (Human Judgm
Slide 22: Flipside AI: The Operational Moat
Flipside Al: The Operational Moat Acquisition is not just about technology; it’s about operational maturity. Flipside Al is the infrastructure operator running the data lifecycle. Orchestration Capture Supervision Enrichment Delivery Ya we, Capture Supervision Delivery Human-in-the-Loop Multi-stage
Slide 23: From Raw Pixels to Embodied Intelligence
From Raw Pixels to Embodied Intelligence 1. Ingest 2. Anatomy Extraction 3. Semantic Enrichment 4. Training Asset (Raw FPV Video) (3D Pose) (Intent & Object) (VLA Dataset) ..J We convert human motion into the mathematical language of robotics. � Isl\ NotebookLM
Slide 24: Quality Control as an Operating System
Quality Control as an Operating System Quality Loop. Automated Contributor > Check Upload (R Framerate) Specific Feedback (Coaching Moment) Next-Generation Engineering k >) Supervision esolution/ (Flipside) Rejected a a Philosophy: Quality is not a manual checkpoint; it is a system. Task Templates:
Slide 25: The Technology
The Technology Proprietary platforms enabling seamless workflow between Engineering and Operations. TaskAlpha Mobile App Senscient Cortex Training-Ready Data •·- -- • - •- o- OOc==:J.'1 • ·- D - .,_ • • o := ... · � ::..-.. =• - 0 o- • � - a o=. • <> •- •- II 8 E> 0 =- ) Interface for structured d
Slide 26: TaskAlpha: The Distributed School for Robots
TaskAlpha: The Distributed School for Robots Turning everyday household chores into supervised training data. The Concept: A contributor-powered, paid platform that operates like the “Uber for Robot Training.” We solve the scale problem by decentralizing collection to everyday people. The Edge: Unli
Slide 27: TaskAlpha: The 'Gig Economy' Engine for Robotic Data
TaskAlpha: The ‘Gig Economy’ Engine for Robotic Data Building the ‘Alpha Layer’—The Collective Intelligence of Humanity. TaskAlpha Global First-Person Video Wes Anonymization Contributors (FPV) Capture & Processing = Focus on Fine-Manipulation Tasks (cooking, cleaning, assembly) and Real-World Varia
Slide 28: Step 1: Capture - Senscient TaskAlpha
0 C Step 1: Capture - Senscient TaskAlpha Decentralizing data collection through a contributor-powered platform. Concept: Gig-Economy - Pay people for performing everyday chores. Mass Participation: Everyday people recording chores (folding laundry, loading dishwashers) via smartphone. r Ego-Centric
Slide 29: Step 2: Refine — The Flipside AI Operational Engine
Step 2: Refine — The Flipside Al Operational Engine Transforming raw video into ‘Ground Truth’ through industrial-grade supervision. =a Privacy Somubb og e Flipside Al is an infrastructure operator. Workflow Orchestration: Multi-stage QA and rejection loops to prevent quality drift. e Privacy Scrubb
Slide 30: Step 3: Deliver - Senscient Cortex
SMSC - /\ ------<> � Step 3: Deliver - Senscient Cortex Production-grade engineering assets ready for direct ingestion. • The Product: Not just video, but high ! fidelity 'Action-Outcome' pairings. • Structured & Versioned: Auditable datasets aligned specifically for Physical Al & VLA models. • Tra
Slide 31: Proprietary Infrastructure & Tooling
➔o-? Proprietary Infrastructure & Tooling TaskAlpha Mobile App JetBrains Mono Senscient Cortex ,,- JetBrains Mono � CONHAND CENTER c;;i IIORAfU)W ORCHESTRATION "• GtMLllY N£TRtCS ------r- f@ ,,a > I >L---..J ...,. 99. SI\ IH'TEBAnv ' - � 1HGCSTlOlf �lTY QIECX CUlATIOlf OEUVHlf "" .. 1' I I I .....;
Slide 32: Senscient MindMart: The 'App Store' for Physical AI
Senscient MindMart: The ‘App Store’ for Physical Al The Problem: You buy a $20k robot, but it can only walk. Building new skills is impossibly hard. The Solution: One-click deployment of pre-trained skills via Senscient Cortex. The OpenAlpha Paradigm allows developers to \-build, publish, and moneti
Slide 33: Use Cases: Powering the Next Generation of Machines
Use Cases: Powering the Next Generation of Machines Humanoid Robotics iil me _ a _ Task learning (laundry, Fine-grained manipulation | dishes). & tool usage. Vision-Language-Action Systems. S \ Next-Generation Engineering = / DATA INTEGRATION | REAL-WORLD Al\NotebookLM
Slide 34: Defensibility & Competitive Moat
Defensibility & Competitive Moat Operational depth is difficult to replicate. yj = £ N if - a) ©Capability Labs Marketplaces Senscient Al ) Scale (Int a only) High ye High (Global Crowd) : ae . Low High (Flipside QualityStpervision High (Unverified) Q Industrial QA) Operations Manual None Diversity
Slide 35: Business & Revenue Model
Business & Revenue Model Monetizing the "School for Robots". ----1/ ----1/ Senscient Sales Custom Collection Recurring Supply Programs Partnerships / \ I \ Licensing o f f -the-shelf, Tailored data campaigns for Ongoing dataset refreshes training-ready VLA datasets. specific enterprise needs and and
Slide 36: The Evolution of the Physical Agent
The Evolution of the Physical Agent Phase 1: Copilot / Teleoperation Phase 2: Specialized Autonomy Phase 3: Generalized AGI | ent Focus: Bridging Phase 1 & 2 to lay the tracks for Phase 3. _y Phase 1: Copilot / Teleoperation Phase 2: Specialized Autonomy Phase 3: Generalized AGI Human-in-the-loop co
Slide 37: Roadmap to Global Infrastructure
Se-nscien A Roadmap to Global Infrastructure / ,r 1/ Near Term ••• Mid Term LongTerm • ----� 82B Roboto Mono Roboto Mono Roboto Mono -J.,//:.o <>'\\ cl'b!0o ...,....... Phase 1 Phase 2 Phase 3 1 Inter Mono Inter Mono Inter Mono I �; Target Q1 Target Q4 Target 2027+ Integration & Beta Marketplace The
Slide 38: Nexscient, Inc.
Nexscient, Inc. 7 '] NC (OTCQB: NXNT) Subsidiaries: Senscient Al | Flipside Al To unify complementary capabilities across the value chain NN of embodied intelligence, enabling customers to accelerate product development and improve real-world outcomes. senscient.ai | flipside.ai | nexscient.ai Next-