Announcement PDF Decidr Building the Orchestration Layer for the Agentic economy Capital Raising Presentation – 06 May 2026 1 Disclaimer 2 Decidr builds the orchestration layer for the agentic economy AI investment has concentrated on infrastructure and intelligence, but value will accrue to the orchestration layer that governs decisions, compliance, and economic coordination. Decidr is building that layer for enterprises and sovereign economies. 3 DAI 1 A once-in-a-decade category Knowledge Security is a new line on the CISO budget, underpinned by $100B+ of sovereign AI compute investment in 2026 and binding regulatory action across multiple jurisdictions. 2 A platform already shipping Sugarwork acquired; DecidrOS Beta live; annualised revenue exit rate of A$8.06m at 31 March 2026, more than 2x prior quarter. Partnerships executed with Beckway, AE Studio and ICON Consulting Group. 3 A strategic enterprise asset Sugarwork acquisition (implied US$250m DecidrUS valuation) adds methodology IP, Vanessa Liu's leadership and the enterprise Knowledge Security wedge. DAI holds 97% of DecidrUS. 4 Disciplined use of proceeds Knowledge Security Platform · sovereign compute + IP/R&D · M&A opportunities · costs. Five milestones over 12 months. 5 Built at the orchestration layer Value in the agentic economy accrues at coordination, not application or model. DAI sits between c.US$2T of underlying infrastructure and trillions of enterprise processes — the layer that translates probabilistic AI into governed enterprise execution. Why now, why DAI 4 4 Opportunity Frontier labs have left the model layer and are colonising the application layer — by role, industry, and workflow. The remaining vacuum: the orchestration layer between models and the enterprise. That is the layer that compounds, carries audit trails, sovereignty, and trust. 01 CAPTURE Sugarwork • Embedded Partnerships 02 COORDINATE DecidrOS platform 03 COMPOUND Agentic Graph • data substrate We sell sovereign-grade AI orchestration to enterprises who can't fully trust US hyperscalers. The orchestration layer is unclaimed DAI is building it 5 Strategy 01 Pilot fatigue is real 95% of GenAI pilots fail to reach production MIT NANDA's State of AI in Business 2025 study found only 5% of integrated AI pilots extracted measurable P&L value. Buyers want orchestrated workflows that integrate with trusted systems, not just demos. MIT NANDA · State of AI in Business · Aug 2025 02 Vertical land grab is on 16 vertical product launches by Anthropic + OpenAI Frontier labs are colonising the application layer — engineering, gov, defense, commerce, health, design. The orchestration layer - coordination, sovereignty, audit - is structurally unclaimed. Public product announcements · Oct 2024  Apr 2026 03 Sovereign AI budgets $2T in committed sovereign infrastructure spend by 2030 GCC, ASEAN, India, and EU are building sovereign cloud + AI stacks. They need orchestration partners that can air-gap workloads and respect data residency. McKinsey, BCG, Saudi PIF and UAE G42 aggregated Three forces converging in a 12-month window The market Pilot failure forces orchestration • Frontier labs occupy verticals • Sovereign capital is ready to deploy 6 01 CAPTURE Sugarwork + Embedded Partnerships Knowledge-capture • Ecosystem Led Growth via embedded partner workflows • sovereign fine-tuning pipeline Revenue today; training data for the Graph tomorrow 02 COORDINATE DecidrOS Schema-driven agent orchestration & interoperability Standardises and governs how agents read, write and audit across vertical apps 03 COMPOUND Agentic Graph Pre-boarded knowledge-node • network = moat Partners become super-nodes, Enterprises major nodes. As graph density increases pre-boarded businesses become more intelligent. One product story: Capture → Coordinate → Compound 7 Strategy The opportunity Enterprises are leaking decision-structure into black-box foundation models. Sovereignty is moving from policy debate to buying criterion - with c.US$100+ billion of sovereign AI compute investment expected in 2026 alone. DAIʼs solution Knowledge Security: a new enterprise category that captures tacit reasoning inside the customer's boundary and fine-tunes open-source models on that captured structure - delivered through Sugarwork and the DecidrOS orchestration layer. Proof in market Annualised revenue exit rate of A$8.06m at 31 March 2026, more than doubling quarter on quarter; partnerships executed with AWS, SAP and ICON Consulting Group; Sugarwork acquisition completed; DecidrOS Beta live. Capital allocation A$6.0m to commercialise Sugarwork as a Knowledge Security Platform · A$2m into sovereign compute infrastructure, patents and published research · A$6.0m M&A opportunities · A$1.0m offer costs and working capital. Proceeds fund the build-out of enterprise Knowledge Security, sovereign compute capability and disciplined M&A — positioning DAI at the highest-value layer of the agentic AI stack. A capital raise aligned to the sovereignty challenge 8 Placement A$8.06m Annualised exit rate at 31 March 2026 More than 2x the prior quarter 2.0x Q3 FY26 over Q2 FY26 QoQ growth 8.0x Year-on-year growth Q3 FY26 vs Q3 FY25 Annualised revenue exit rate has doubled in the most recent quarter - five quarters of growth in Australian test market Top 5% of SaaS trajectories Companies with similar early revenue trajectories 9 Progress Revenue Model Revenue pathways & progress 10 New Partnerships Tier 1 APAC System Integrator Teaming agreement executed across APAC, as a Sugarwork reseller. ICON Consulting Enterprise pipeline progressing across ASEAN for Sugarwork and Accounts Receivable workflow via SAP. AE Studio Co-sell motion initiated, supported by joint marketing initiatives targeting the private equity sector. Beckway Appointed as commercial reseller, including Sugarwork distribution. APJ Datacentre MOU signed covering NVIDIA chip inventory access, supporting AI positioning in ASEAN. Google Cloud Continuing to deepen relationship with Google Cloud ecosystem. Existing Partnerships 2025 FasTrack Selection AWS Highlight Marketplace Fast Tracking listing now live, with co-sell motion active across enterprise and mid-market segments. Partnerships update Building on the existing partnership network 11 11 Ecosystem-led growh model Partnerships case study Hyperscalers working with Decidr 12 ● Decidr commenced AWS partnership journey in Feb ‘25 ● Accepted onto AWS fast-track program to co-sell on platform ● Existing AWS customers can use credits to purchase Decidr applications ● Decidr has recently released its second AWS application, unlocking new sales avenues ● Co-sell motions active from Mid-March ‘26 The new way to capture data for fast, accurate decision-making Bringing order to the chaos of scattered, unstructured data so human teams and AI agents can work in concert to identify efficiencies, create new products, and increase revenue is the mission of Australian firm Decidr. "Google Cloud's advanced AI and data ecosystem, including Gemini, allows us to move seamlessly from raw data to deterministic outcomes." Michael Cindric Chief Technology Officer, Decidr Published Case Study 12 Partnership example 01 Methodology IP Codified Deep Dive question-set built over two years. Teachable, replicable, schema-aligned with DecidrOS. 02 Category creation Sugarwork creates Knowledge Security - a new line on the CISO budget. With it, DAI owns both capture and orchestration. 03 Enterprise GTM wedge Moves DAI from SME mid-market into Fortune 2000 enterprise via Security/GC entry points. 04 Cross-sell compounding Sugarwork opens enterprise pathways for DecidrOS via task identification. DecidrOS opens SME app revenue for Sugarwork-derived schemas. Acquisition Snapshot ACQUIRER DecidrUS Inc. STATUS Completed VALUATION US$250m, 97.4% DAI ownership CEO JOINING GROUP Vanessa Liu (ex-SAP.iO, McKinsey) PARTNER ECOSYSTEMS AWS · GCP · SAP · DXC Sugarwork captures the tacit decision-structure that drives a firm's competitive position. Acquired by DecidrUS, it brings the methodology IP, leadership and enterprise GTM that take DAI from SME orchestration into Fortune 2000 Knowledge Security. Sugarwork — the enterprise wedge for DAI Sugarwork customers & partner channels 13 Sugarwork Sugarwork Knowledge discovery and security in practice Enterprise is still in its infancy of understanding their need for sovereignty. Sugarwork is solving the problem that the market is uncovering The enterprise problem Prompt Leakage: 43% of employees admit sharing sensitive data; 67% of enterprise AI usage happens via unmanaged accounts. Schema leakage: Even when content is sanitised, prompt patterns reveal reasoning structure. Employees are unknowingly training LLMs on their reasoning structure. Model-dependency lock-in: Black-box integration creates compounding dependency. Lack of independence and sovereignty becomes a problem for customers. Regulatory exposure: GDPR, HIPAA, APRA CPS 230, MAS TRM. Four of these came into force in 2025 alone, requiring enterprise-controlled infrastructure. The Sugarwork solution Capture Operational Structure: Sugarwork ends leakage by capturing and retaining tacit knowledge in-house. Independence & Sovereignty: Avoids black-box lock-in by maintaining an executable map of company artifacts, ensuring the firm retains its own reasoning structure. Improved economics: Businesses operating on sovereign models lower reliance and costs from frontier models, without losing their ability access higher-cost intelligence for specific work. Compliance-Ready Infrastructure: Enables regulated decision processes to run on infrastructure the enterprise controls, meeting EU AI Act and GCC sovereign-AI mandates. “We think every company is going to need a new primitive: a living map of how the company works that turns its own artifacts into an executable skills file for AI.ˮ — Tom Blomfield, Partner, Y Combinator - April 28, 2026 14 Sugarwork: Knowledge Security Platform Structured Knowledge Capture AI-assisted expert interview programme - codified at the heart of Sugarworkʼs IP Ambient meeting capture Always-on intelligence layer that extracts decision-structure as it emerges in working meetings. Reasoning Graph Unified, proprietary representation of how the enterprise actually reasons — the training corpus. Sovereign fine-tuning Customer-selected open-source base fine-tuned inside infrastructure. Weights owned outright. DecidrOS Deterministic Orchestration Layer Agentic Schema First-principles blueprint of the organisation — reusable across domains, for AI to reason and act. AI proposes → rules constrain → humans approve Deterministic guardrails with full audit trail. Confidence thresholds gate automated actions. Multi-agent coordination Coordinated workflows across LLMs, business apps and AI agents replacing brittle integrations. Enterprise and sovereign capable VPC, sovereign cloud, on-premise, or air-gapped deployment. Compliance-as-code at execution. Sugarwork captures the enterprise's decision-structure. DecidrOS runs the deterministic agentic workflows that execute on it. Both sit inside the customer's boundary. Sugarwork and DecidrOS - one sovereign reasoning platform Technology alignment 15 Equity Raising Accelerating commercialisation 16 $15.0m Total raise Knowledge Security Platform Sugarwork) Productisation, enterprise go-to-market, always-on meeting capture, ongoing methodology R&D. Sovereign Compute, Patents & Research Dedicated inference capability with strategic sovereign compute partners; IP register build-out; peer-reviewed research programme. M&A Opportunities Opportunistic bolt-on acquisitions that compound inside the DecidrOS orchestration layer. Offer Costs & Working Capital Co-Lead Manager fees, legal, registry, listing costs and corporate working capital. Proceeds are directed to three strategic areas aligned to the sovereignty thematic, with a reserve for opportunistic M&A. A$15 million - disciplined deployment 17 Use of proceeds "Data security is solved. Knowledge Security isnʼt. Every prompt your employees send is a training example." Training data is the moat. The market already priced it. US$45B Every frontier lab is buying experts to take over industries Finance · Engineering · HR · Marketing · · · · of valuation has accreted to expert-capture companies in 18 months. Scaling Sugarwork from a completed acquisition into a productised Knowledge Security platform that supports enterprise fine-tuning of open-source LLMs inside the customer's boundary. Sugarwork as Knowledge Security Platform 18 A new category AfterQuery / Mercor / Surge / Handshake Sugarwork Customer Frontier labs Enterprises Owns the expertise Frontier labs The enterprise Where the model runs Frontier labs Customer sovereign environment Purpose Abundance for all humans Sovereignty for one firm Budget line Lab R&D Security CISO + Sovereign AI CIO TAM ~ 1020 labs globally 10,000ʼs enterprises Concentration risk Severe Meda / Scale) Diversified by design We don't compete with them. We complete them. They build expertise abundance for the labs. We build expertise sovereignty for the enterprises those experts work at. CAPTURE DEPTH → MODEL OWNED BY → CUSTOMER VENDOR SHALLOW DEEP ChatGPT / Claude Ent. Black-box SaaS chat Microsoft Copilot Embedded assistant AfterQuery / Mercor Deep capture · to labs Glean / MS Graph Retrieval on own data Writer / Cohere Thin fine-tune, vendor-hosted Sugarwork Deep capture + customer-owned Five adjacent categories. One clean whitespace. Deep-capture + customer-owned is structurally unoccupied. Chat vendors can't pivot without cannibalising model-lock-in. Expert-training firms can't enter without abandoning their customer. Retrieval and thin-fine-tune don't capture reasoning. AfterQuery & Mercor sell to labs. Sugarwork sells to enterprise The white space for Sugarwork 19 A$1.0m Sovereign inference capacity Dedicated inference targeting bare-metal performance, fixed-rate power and 24/7 managed operations. A$0.75m Patents & IP register Filing programme covering DecidrOS architecture, schema methodology and Knowledge Security primitives. A$0.25m Published research Peer-reviewed programme strengthening thought leadership and signal for academic credibility. Strategic compute partners — under commercial discussion Southern Cross AI SCX AU-based sovereign cloud Dedicated inference targeting bare-metal performance and managed operations. Nebius Global NVIDIA-aligned compute Strategic dialogue via NVIDIA for scaled global inference with sovereign deployment. Multi-national hyperscalers Country, region and global cloud Discussions on in-country deployments of DAI sovereign orchestration capabilities. Establishing dedicated inference capability through strategic sovereign compute partnerships — the substrate enabling air-gapped, regulated and sovereign environments. A$2.0m — Sovereign compute, patents & research 20 Use of proceeds - detail Disciplined M&A opportunities Completing the repeatable M&A flywheel: horizontal AI rewards integrated platforms while standalone point solutions struggle with distribution and margin compression. As the market continues to evolve DAI sees a whitespace opportunity to deliver full-stack sovereign orchestration solutions to enterprises via Sugarwork entry and to SMEs via the Agentic Graph. 21 Use of proceeds What A$15m unlocks over the next 12 months Five milestones that compound into the DecidrUS valuation reference. 22 Use of proceeds 1 Fully paid ordinary shares as of last close on Monday, 4th, May 2026 Sources, uses of funds and pro forma capital structure Sources A$M Institutional Placement $15.00 Cash and cash equivalents balance as at 31 March 2026 $17.16 Total Sources $32.16 Uses A$M Knowledge Security Platform Sugarwork) $6.00 M&A opportunities $6.00 Sovereign Compute, Patents & Research $2.00 Offer Costs & Working Capital $1.00 Total Uses $15.00 Fully paid ordinary shares1 Existing shares % pre-offer Offer shares issued Post-offer shares % Post-offer shares Decidr Group Pty Ltd 78,400,000 24.1% 0 78,400,000 22.4% JR Cranston Family Trust 7,054,794 2.2% 0 7,054,794 2.0% Lead Nation Holdings Limited 5,285,000 1.6% 0 5,285,000 1.5% Vincent Corp Pty Ltd 4,995,000 1.5% 0 4,995,000 1.4% Institutional Placement 0 0.0% 24,590,168 24,590,168 7.0% Other shareholders 229,074,040 70.5% 0 229,074,040 65.6% Total shares outstanding 324,809,534 100% 24,590,168 349,399,702 100% Other securities Options 56,972,000 17.5% 0 56,972,000 16.3% ESOP – performance rights 2,281,677 0.7% 0 2,281,677 0.7% 23 Details Equity raising timetable Indicative summary of events Date Trading halt announced Tuesday, 5 May 2026 Placement bookbuild completed Tuesday, 5 May 2026 Trading halt lifted – Securities recommence trading on ASX and announce completion of Institutional Placement Wednesday, 6 May 2026 Settlement of New Shares under the Institutional Placement Wednesday, 13 May 2026 Allotment and Quotation of New Shares under Institutional Placement Thursday, 14 May 2026 24 Placement timetable Confidential / Copyright © Appendix 25 Rare Earths Energy Chips Data Centres & Compute Data Labelling & Training Foundation Models AI Applications Orchestration AI is a horizontal technology, not a vertical market Standalone AI point solutions face distribution and margin compression Orchestration is the value-capture layer Sits between c.US$600B of infrastructure and $trillions of enterprise processes. Platform economics compound Each capability acquired compounds inside a shared orchestration layer. Network effects: more schemas → more patterns → better orchestration for all Enterprise and sovereign capable by design Deterministic guardrails prevent probabilistic outputs from corrupting ledgers, contracts or compliance records. Source: MIT NANDA (2025); Arcade.dev / Gartner (2025) — 94% of enterprises view process orchestration as essential for deploying AI effectively. Global capital is building the intelligence stack. DAI is building the deterministic layer that translates probabilistic AI output into governed enterprise execution. DAI builds the missing orchestration layer 26 DAI 26 Probabilistic AI ● LLMs predict the next most likely token — output is a probability distribution, not a fixed answer ● Hallucination rates 0.7%30%) mean even the best models fail 7 in every 1,000 prompts ● Compounding error: at 95% step-reliability, a 20-step workflow succeeds only 36% of the time Deterministic Business ● Compliance, pricing, and contracts require predictable, repeatable outcomes ● Explainability is the #1 AI adoption barrier in regulated industries 65% ● Legacy systems expect deterministic inputs, not probability distributions ● Single errors in commerce trigger regulatory action, brand damage, or financial loss 95% of generative AI pilots fail (MIT NANDA, 2025). The bottleneck is not better models — it is the coordination and reasoning layer that sits above them. Raw AI capability does not produce compliant decisions, trusted transactions or functioning economies. The gap between probabilistic models and deterministic business is where enterprises get stuck. Intelligence ≠ Coordination The problem 27 Capture stays on-shore Deploy inside your VPC, sovereign cloud, or on-prem. No fragment of captured reasoning transits through Sugarwork-controlled systems. Training runs in your environment LoRA/QLoRA jobs execute on infrastructure you control. Sugarwork operators cannot see your corpus or intermediate gradients. Inference stays inside the boundary No tokens transit to third parties. Air-gapped option available — the model operates with zero outbound egress. You own the weights Customer-held artefacts. Terminate us and the model remains yours. We retain no copies. Your data trains your model Nobody else's Enterprise knowledge value from Sugarwork 28 28 28