Why We're Raising Now

The window for building the Knowledge Security layer is open. It won’t stay open.

By DcB

Co-founder & Executive Chairman, Decidr AI Industries (ASX: DAI)

Foreword

A question I have been asked by investors over the last few days is ‘why now?’. Having completed a raise in September 2025 of $20 million, and with c.$17 million on the balance sheet, it is a fair point to raise. It looks condensed on the surface. But unlike other industries like mining, biotech, networks and retail, the industries that are in the AI glide path move at speeds that have never been seen before.

What we are witnessing in the AI space is unprecedented. Foundation models improve weekly, there are more AI agents than there are people on the planet. Every week, nearly 1 billion people use AI. 

When you look at the latest placement through this lens of speed and innovation, it takes a very different form. In the AI world where being first can get you the deepest moat and the widest market, strategic capital can make a critical difference.

Today’s placement announcement should not distract from the progress we are making on our operational roadmap - momentum continues to build and we are confident in our plan for 2026. In order to meaningfully grasp the white space in front of us, however, an injection of capex focused capital will help get us there first.

I’ve spent the past few months writing about what’s happening in the AI market. The venture subsidy playbook. The knowledge security gap. The IPO capital squeeze. The interoperability-versus-orchestration distinction. I wasn’t writing those articles in a vacuum. I was writing them because they explain, better than any investor deck could, why we’re raising capital right now.

This is the context around our recent placement: a $15 million raise at the DAI level to fund the next phase of what we’ve been building for two years.

Let’s start with what’s happening in the market, because the timing isn’t arbitrary.

Token pricing is collapsing. Down 99.7% from GPT-4’s launch pricing in March 2023. Anthropic just killed flat-fee enterprise billing and moved to per-token consumption two weeks ago. OpenAI moved Codex to token metering in April. GitHub tightened Copilot limits on April 10. Windsurf replaced credits with daily quotas in March. Cursor revealed that the true price of their $200 plan was actually $5,500 in real costs. The venture subsidy era for agentic workloads is ending across the board.

That repricing creates a window. Enterprises that built their AI workflows around a single provider at subsidised prices are now discovering what captured dependency feels like. They’re looking for vendor-neutral alternatives. They’re asking, for the first time, who owns the system that coordinates AI agents economically, safely (without hallucinating). Six months ago the question “who owns the orchestration layer” was academic. Today it’s becoming a budget line item.

At the same time, knowledge security has gone from a niche concern to a board-level risk. OpenAI’s desktop superapp watches how 900 million users work. Anthropic’s Cowork has eleven plugins observing sales, legal, finance, and support workflows. These tools capture the tacit knowledge of how organisations operate, the 80% living in employees’ heads that never made it into a process document. All of this is also being captured on infrastructure the enterprise doesn’t control.

France understood this. SecNumCloud exists because the CLOUD Act allows US law enforcement to compel data from US-controlled providers regardless of where it’s stored. China registers every AI algorithm with the state. Australia doesn’t have an equivalent. The companies best positioned to address both the enterprise and sovereign risk are the ones building vendor-neutral orchestration with knowledge residency built into the architecture.

The question used to be: can AI do this task? The question now is: who owns the knowledge of how your organisation does it? That’s the question we’re raising capital to answer.

Here’s what two years of building taught us. The enterprise AI problem today is not intelligence. It’s knowledge security. Every foundation model sitting on a desktop is a knowledge extraction engine. Claude Code watches your engineers. Cowork watches your sales team. ChatGPT watches your analysts and HR. Over time, these tools accumulate a detailed map of how your organisation actually operates. Not the org chart. The real version. The workarounds, the tribal knowledge, the undocumented reasoning that makes your business run. That knowledge is your most defensible asset. And right now it’s being captured on infrastructure you don’t control.

The acquisition of Sugarwork gave us the answer. Sugarwork captures tacit knowledge out of organisations before the foundation models do. AI-assisted expert interviews. We’re building an always-on meeting capture that extracts decision-structure as it emerges. The output is a reasoning graph: a structured, proprietary representation of how the enterprise actually thinks. That’s not a process map. It’s the training corpus for a sovereign model that the enterprise owns outright.

That’s the product we’re building. Sugarwork captures knowledge. The captured structure feeds into sovereign fine-tuning of open-source large-language models (LLM), running inside the customer’s boundary on infrastructure they control. Your data trains your own LLM and decision support system. Nobody else’s. This also integrates and accelerates the DecidrOS orchestration layer and is the required first step for all businesses who want to better coordinate people and AI to achieve better business outcomes.

The capital raised this week goes to turning this into a scalable product.

Tom Blomfield at Y Combinator said it last week: “every company is going to need a living map of how the company works that turns its own artifacts into an executable skills file for AI.” 

That’s what Sugarwork builds. And when that map feeds into a sovereign fine-tuned model running on infrastructure the enterprise controls, you get something no foundation model provider can offer: AI that knows how your business actually works, running in your environment, with weights you own. The enterprise lowers its reliance on frontier model spend without losing access to higher-cost intelligence for specific tasks.

We raised $11 million in March 2025 and $20 million in September 2025, funding the platform build and initial international expansion. This current placement takes it further: A$15 million, disciplined across four areas:

• ~$6 million to productise Sugarwork as a Knowledge Security Platform with enterprise go-to-market, always-on meeting capture, and ongoing methodology R&D. 

• ~A$2 million into sovereign compute infrastructure with partners like Southern Cross AI and Nebius, a patent filing programme covering DecidrOS architecture and Knowledge Security primitives, and a peer-reviewed research programme. 

• ~A$6 million reserved for disciplined M&A, strategic acquisitions that compound inside the DecidrOS orchestration layer. 

• ~A$1 million for offer costs and working capital. 

This delivers five milestones over twelve months, each compounding into the DecidrUS valuation reference from the Sugarwork acquisition.

The window is open because three things are true at the same time. The foundation model providers are repricing, which is pushing enterprises to seek vendor-neutral alternatives. The knowledge security risk is becoming visible, which is creating demand for sovereign orchestration. And the foundation models themselves are racing into verticals, launching sixteen role-specific applications in eighteen months, which makes the coordination layer between those verticals more valuable with every release.

None of those conditions existed eighteen months ago. The window for an independent, vendor-neutral orchestration platform to establish itself is now. That’s why we’re raising now. Because the market conditions that make the orchestration layer most valuable to enterprises are peaking, and the companies that scale into that window will set the standard.

We’ve spent two years building the infrastructure. Sugarwork is acquired and delivering value to customers. DecidrOS Beta is live. Annualised revenue exit rate exceeded AU$8 million at 31 March 2026, more than doubling quarter on quarter. Partnerships with AWS, Beckway, and ICON Consulting Group are executed. This placement is how DAI becomes the leader in Agentic transformation with Knowledge Security as the first step.