AGENTIC OPERATIONS
Agentic Operations is an operating model that is built around the capability of agentic AI. It is this technology that enables firms to overcome the challenges legacy operating models never could.
While it harnesses other technologies as well as agentic AI, such as cloud computing and no-code functionality, it is the capabilities of agents that form the foundational ability to deal with complexity like never before.
Agentic Operations is an outcome-driven operating model. It shifts the focus towards achievement and is closely aligned with the organisational strategy. Rather than existing to complete a checklist of tasks, it drives the organisation towards its ultimate destination.
Doing so requires more than just adding a new technology platform. It requires breaking down silos, building interteam relationships, rethinking governance, and evolving your technology landscape to create a shared, interconnected stack.
Agentic Operations solves the problem of repetitive manual work in post-trade data management by replacing low-value human effort with autonomous software that can plan, act, and reason.
Who is agentic Operations for?
Firms of any size that deal with large volumes of complex trade data and currently rely on manual intervention to complete essential post-trade tasks can benefit from agentic Operations.
Our initial Pacesetter wave who piloted our agentic workspace capability, which underpins agentic Operations, consisted of banks, asset managers, a broker, an insurer, and an asset servicer of various sizes.
What problems does agentic Operations solve in post-trade?
Agentic Operations solves the problem of repetitive manual work in post-trade data management by replacing low-value human effort with autonomous software that can plan, act, and reason.
The existing operating model in post-trade could be described as ‘exceptions management-focussed’. The intricacies of financial data, and the shortcomings of the platforms designed to handle it, meant data had to be viewed as a process disruptor. The operating model was built around knowing that things would break on the way from A to B, and resourcing for armies of people to clean up the mess.
Nobody designed this on purpose, it’s just what evolved to cope with the unique challenges Operations teams in financial services businesses face.
Agentic Operations, on the other hand, is an operating model that is purposefully designed. While it harnesses other technologies as well as agentic AI, such as cloud computing and no-code functionality, it is the capabilities of agents that form the foundational ability to deal with complexity like never before.
This enables it to solve the following core post-trade challenges.
Replacing endless manual work
Traditional automation tools can’t cope with the complexity and variety of financial data. Human intuition and knowledge used to be the only way certain tasks could get done.
Agentic AI, with its ability to reason and adapt, is able to finally automate these tasks.
Scaling to keep pace with demand
Volumes are increasing, markets are moving to 24/6, and regulation is still tightening. Operations needs to be able to handle much more - yet at the same time, leaders are under pressure to keep costs down.
Agentic Operations enables you to put your valuable analysts to work on high-value tasks while agents automate the bulk of the work that was eating their time, even as volumes grow.
Overcoming agility blockers
Operations has to respond fast to changing business needs. But traditional technology, and the change management procedures that surround it, can’t do this.
Agentic Operations relies on fast, governed tools that can rapidly adapt to changing circumstances, such as ingesting a new data source or iterating on existing controls.
Replacing opacity with clarity
Traditional operating models run on a tech stack full of point solutions held together with manual work in end-user computing (EUC) solutions such as Excel spreadsheets. It’s almost impossible to see across the data lifecycle and understand everything that happened to it during its time in your organisation.
Agentic Operations changes that, automating away EUCs, standardising and centralising processes, and logging every action clearly.
What are the benefits of agentic AI in post-trade?
Agentic AI enables firms to automate complex data management tasks in post-trade that previously could only have been handled by humans. Other automation tools have relied entirely on rules-based processing, which is fine for low complexity tasks. But these rules fall over when things change regularly and they encounter an eventuality they were not programmed for.
Agentic AI replicates the natural flexibility and adaptability of the human brain, can understand the operational context in which a process exists, and make informed decisions to reach the desired outcome.
For example, agents can triage and resolve many of the false breaks produced by traditional reconciliation platforms before Operations analysts even come to work, enabling them to start their days focussed on the high priority errors that require immediate resolution.
Early pilots of agentic Operations have revealed a weighted average time saving of 76% on reconciliation tasks.
Agentic Operations FAQs
Agentic Operations automates a lot of the manual tasks that have been keeping your team occupied. This, in turn, enables them to focus on higher-value work, such as investigating genuine breaks, optimising processes and monitoring risk. Additionally, agentic AI requires a human-in-the-loop to review and approve key actions as an essential part of your control framework (see below). This is something that your team will handle, because they are the ones with the knowledge and expertise to evaluate potential actions.
Operations leaders are under constant pressure to meet the ever growing demands of the business without increasing their cost base. Agentic AI offers them a way to do much more with the resources and team they already have.
Agentic AI is safe for post-trade if it is implemented correctly, with the right guardrails and governance framework in place. We believe any agentic platform should be designed with the following principles in mind:
- Human-in-the-loop by default - people should always be involved in reviewing and approving key actions.
- Explainability as a product requirement - the tool should clearly show what decisions were made, and why.
- Immutable auditability - every aspect of your data’s journey in the platform, from rules-based transformation to agentic reasoning, needs to be clear, documented and defensible.
It’s worth remembering that, because agents essentially act like people, you already have many of the right governance principles in place, such as granular access permissions and four-eye checks. Agentic AI requires some new guardrails, but safe implementation can build upon what’s already there.
Data quality is important for any tasks where the agent is required to make a decision, but agents can work with messy and poor quality data - in fact, that is one of their key use cases.
For instance, agents can review datasets and build a reconciliation process with the necessary matching rules to identify issues. If agentic AI required perfect data in order to operate, no one would need agentic AI in the first place; the problems it can solve wouldn’t exist.
However, data quality is non-negotiable when it comes to the information any form of AI (or humans, for that matter) is using to make decisions and take actions.
AI can act at machine speed, which as we’ve already seen causes all kinds of governance headaches. AI given poor quality data is at risk of making the wrong decisions, and the consequences of those decisions can quickly spread across systems and your organisation.
Agentic Operations is built upon the principles of speed, adaptability, connectivity and transparency. The technology used to realise this operating model must therefore be able to adhere to those principles.
In practice, that means that any agentic Operations platform needs to be AI-enabled, cloud-native, data agnostic, interoperable and no-code.
Cloud-native
Being cloud-native is important because it enables rapid scaling in the face of rising business demands, reduces the total cost of ownership for IT teams, and enables seamless updates and innovation.
Data agnostic
Data agnostic technology is able to ingest data in virtually any format, from any source. This is opposed to traditional data management platforms, that require data in a particular schema in order to understand it. Being data agnostic radically changes the extract-transform-load stage of data preparation, making it quicker to adapt to new business demands.
Interoperability
Agentic Operations is built upon interoperability, meaning that your tools can connect to one another and agentic capabilities are able to utilise tools across platforms. For example, the Duco platform contains hundreds of individual tools that any correctly-permissioned agent can access and use, in order to perform tasks such as building a reconciliation process or triaging exceptions.
No-code
Finally, agentic Operations platforms must leverage no-code functionality for both users and agents. This means that all data transformation, reconciliation and validation rules are written in plain English commands, not a scripting language like Python. This is essential in creating transparency, as everyone from teams members to external auditors can read and understand what a rule is doing.
Automation is a key technique that enables you to remove human effort from post-trade tasks. Agentic Operations is an operating model that uses the flexibility of agentic AI to expand automation to tasks that have traditionally required human intervention. You can’t have agentic Operations without automation, but you need intelligent, adaptable automation that can reason in order to build agentic Operations.