Outcomes, Not Applications: Announcing the Adarga Mission Agents Framework
This is the third in a series of articles setting out Adarga's technology strategy, following Own the Decision: The Sovereign AI Stack and Actionable, Safe and Resilient AI: The Adarga Research Agenda.
Our customers are having to do even more in less time, always seeking to make decisions at the speed of relevance. Simultaneously, the world is going through a profound technological acceleration...If Defence, Security and Resilience organisations do not seize the opportunity afforded by this acceleration, advantage will be ceded to adversaries who certainly are. We shall all be less safe - and the cost borne across all of our society.
Those words opened our announcement of the Sovereign AI Stack for Defence, Security and Resilience (DSR). Today I want to set out what happens above that stack, where the value actually gets delivered, and announce the capability that puts Adarga, we believe, further ahead in our market than we have ever been.
Where AI tech is moving to
Seizing the opportunities presented by AI is one of the most pressing challenges facing organisations across the DSR sector, not least because adoption must happen alongside the ever-changing dynamics of geopolitics. The unrelenting demands of present operational requirements force that change to happen within the context of an inherited brownfield site where legacy infrastructure, processes and software cultures impede organisational-scale transformation and the ability to realise the value afforded by the new AI-native paradigm. In addition, and as we have previously outlined, any organisation choosing to work within DSR elects to operate within the confines of a series of unique operating requirements, necessary as they are across functional areas such as security, deployment, audit/compliance and more nebulously, trust.
Adarga’s Research Agenda, highlighted in a previous blog post, demonstrates the creative opportunities these unique operational requirements offer for organisations working at the intersection of AI and DSR. Since our inception 10 years ago, Adarga’s story has been one of going the extra mile to bring the most bleeding-edge capabilities available across the global AI market, to those organisations that uphold the values that we care about most in society. A licence to operate within DSR requires the additional thought and creativity in going that extra mile. Adarga’s two research pillars are examples of this. AI Assurance provides the justified confidence, and Adaptive Knowledge Representations ensure agents reason over the right information, with the right provenance, at the right moment. Together these pillars are designed to drive clarity and confidence in decision making for our customers.
But clarity and confidence in todays and future operating environments are only good enough if accompanied with decision making ‘at the speed of relevance’. It is becoming clear, the speed of relevance is the speed of machine intelligence and future conflicts shall be avoided or, when joined, won and lost on the basis of it. We believe that ‘winning’ is predicated on the ability to hold decision advantage, the operational edge gained by making and implementing higher-quality choices faster than an adversary. To get there, we must force the human further up the value chain in workflows and decision making, building agentic systems that can operate at the speed of machine intelligence.
Recent high-profile cases of agentic systems breaking out of imposed guardrails challenge the assumption these same systems can deliver the confidence in decision making demanded by DSR organisations. But this is an example of the constraints of the domain demanding additional inventiveness and care in order that customers can realise the value of cutting-edge technological breakthroughs. The question “can agents be trusted within high-stakes decision making?” is not answered agent by agent, prompt by prompt, it must be answered at the platform level.
Announcing the Mission Agents Framework
Today we are announcing the Mission Agents Framework: an agentic system built, secured, hosted, run and served on our AI fusion platform Catalyst, engineered from first principles specifically for DSR.
Adarga’s Catalyst AI platform acts as the agentic harness. Assurance is not a wrapper applied to each agent; it is a property of the platform. Every agent operating within the Mission Agents Framework inherits, by construction:
• Agent identity and permissioning: every agent is a known, credentialed actor with explicitly scoped authority;
• Tool mediation and retrieval control: agents access data and take actions only through governed interfaces, with provenance preserved;
• Logging and approval gates: every reasoning step and action is auditable, and human authority is engineered into the workflow at defined points, not assumed around it.
Because assurance sits at the enterprise level rather than on an agent-by-agent basis, our customers can experiment at speed. With Catalyst acting as this agentic harness across a range of deployment environments (cloud, on-premises or edge) we have opened up the opportunity for our customers to drive the organisational and transformational change required to operate at the speed of machine intelligence. Our customers can rapidly build, test and field new agentic workflows at speed, challenging existing ways of working to equip themselves appropriately as the contest for decision advantage comes to define success simultaneously across the tactical, operational and strategic levels of conflict.
Catalyst enables this, behaving as the organisational semantic layer. It is the AI platform purpose-built for DSR data, that has evolved out of every paradigm shift since we started building in 2016. Each generation of the Adarga stack was re-architected around the new capability rather than imposed onto an old platform. That distinction, AI-native versus AI-compatible, is the single most important determinant of who will capture the agentic opportunity, and it cannot be retrofitted.
Outcomes; not applications
At Adarga, we concentrate our efforts to support four core Mission Threads that reflect our deep delivery experience, and we have continued this focus with our agentic framework. These are Targeting, Situational Awareness, Supply Chain Analysis and Intelligence & Investigations. Today we are building agents with customers across all four, proving how the units of work, end-to-end processes traditionally delivered by humans can be devolved to agentic systems. Examples include a common operating picture analysis agent, an investigations hypothesis analysis agent and an anomaly detection agent monitoring subsea cable traffic. In In all three instances, the agents operate with configurable autonomy levels, with varying authority to task, cue, recommend, or initiate but with an output designed as an assessment product to enter an existing human process at a defined point. The initial ambition of these use cases is deliberately narrow, especially within our targeting process products, but they are examples of the atomic unit of work a human might otherwise have done being replaced by an agent, thereby moving the human up the value chain.
Adarga’s ambition is certainly not constrained, and we are determined to stay ahead of the challenges our customers face and deliver systems that enable them to act at the speed of relevance, when called upon. With the technology capable and ready, the challenge is adoption and implementation – our biggest competitor is no other software company, but the analyst and their notepad, the planner and their doctrinal frameworks, the commander and their 30 years of institutional knowledge. Regardless, we still have time to act. With the Sovereign AI Stack beneath it, a research agenda driving it forward, and the Mission Agents Framework now delivering within the outcome layer, we intend to define the next epoch of decision-making in Defence, Security and Resilience.
To find out more about the Mission Agents Framework, reach out at hello@adarga.ai
This article was written by Ollie Carmichael, Adarga's VP Product and Technology. Ollie leads the teams responsible for Catalyst, the Mission Layer and Keystone, the three layers of Adarga’s Sovereign AI Stack, and for translating Adarga's research agenda into deployed capability for customers across Defence, Security and Resilience (DSR).