Holistic surveillance is moving beyond data integration towards an operating model that can surface cross-channel evidence, assign escalation ownership, govern approved channels, organise analysts around risk and keep AI-supported triage accountable to experienced human review.
That was the central message of A-Team Insight’s recent webinar, Holistic Conduct & Surveillance: Unifying Trade, Voice, and eComms Data with AI, sponsored by NICE Actimize.
The panel framed regulatory scrutiny around the ability to connect records and reconstruct the event. As one panellist put it: “Regulators say it is all data, and you should be able to connect one to the other. Regulatory expectations nowadays are that various surveillance and monitoring functions within a firm will communicate. The regulator doesn’t care that they are in separate locations or separate countries. The regulator says you have the data and you must bring it together.”The Case Manager Becomes the Control Point
Trade monitoring, voice recording and eComms review have developed as separate control processes, often with different owners, systems and review teams. That legacy model persists, leaving many firms dependent on manual hand-offs. A trade alert fires; an investigator then searches communications records, listens to calls, builds a timeline and looks for evidence that supports or weakens the case.
The panel emphasised that holistic surveillance does not mean attempting to force every control into a single platform. One panellist described bringing the holistic view together at the case-manager level, while retaining best-in-class systems for trade surveillance, eComms and voice.
The value lies in what appears when an alert is opened: relevant communications, market news, voice records, related activity and prior behavioural indicators surfaced inside the investigation workflow.
Defensibility Starts With Escalation
The panel’s minimum viable model was not perfect capture or perfect detection. It was structure: ownership, escalation, accountability, cross-functional review and evidence that surveillance teams do not remain trapped inside their own data domains.
“You may not capture or detect everything,” one panellist said. The key is being “organised, structured, with proper ownership and accountability”. The panel was direct on this point: “You cannot sit in front of a regulator and say you have an eComms analyst who only looks at eComms all day.”
A firm may still depend on manual investigation, but the process has to show how evidence is connected, challenged and escalated. Where full technical integration is not yet in place, investigation procedures still need to force cross-referencing between trade surveillance, voice, eComms, compliance, front-office supervision and relevant control teams.Behavioural Detection Is a Taxonomy Problem First
Detecting behavioural signals is often treated as an advanced AI use case. The webinar offered a more grounded interpretation: surveillance models already point to the behaviours they are designed to detect, and firms can add value by tagging controls according to those behaviours. As one panellist put it: “If you tag all your controls with their intended behaviour, it is quite mechanical and easy to run AI over that.”
Behavioural surveillance is therefore a taxonomy and metadata challenge before it becomes an AI challenge. Firms need to know which controls map to which typologies, how those typologies relate across channels, and how individual alerts contribute to a cumulative picture of risk.
That taxonomy also changes how firms should judge AI. Fewer alerts may not be the first sign of success. The better test is whether AI accelerates context gathering, improves first-level triage and directs experienced analysts towards cases that require market knowledge, judgement and escalation. AI may surface more signals before it reduces the burden on investigators.
Governance Determines Whether Surveillance Is Defensible
The panel framed governance as the difference between connected surveillance and defensible surveillance. Timestamp alignment remains a basic but critical control. “You can have the best AI tools,” one panellist said, “but if your data has incorrect timestamps, you have a problem.” If event sequencing is unreliable, AI-enabled correlation can produce a misleading reconstruction of activity.
New chat features continue to appear inside trading platforms, collaboration tools and workflow systems. Static lists age quickly unless application owners, compliance teams and the front office maintain an active review process. New product, product-enhancement and platform-change procedures should test whether a channel can be captured, retained and surveilled before approval. As one panellist put it: “You simply cannot start communicating on a new channel without that being approved.”Culture also matters. The panel described the need for “positive attitude governance”: a culture in which salespeople and traders come to compliance first when a client wants to use a particular channel, rather than hiding demand for it. The aim is to create a route through which business need, record-keeping obligations and surveillance controls can be assessed together.
Ownership Should Follow the Risk
The operating-model question also extends to team design. One panellist described a best-practice model in which ownership is assigned at the asset-class level rather than by surveillance channel: “My preferred operating model is by asset class. This brings relevant market experience from the analysts in the asset class pods. They are tasked with owning the risk in that asset class, whether it involves eComms, voice, or trade. It is all interrelated.”
That approach moves surveillance data away from channel ownership and towards risk ownership. Analysts who understand the market, product and trading behaviour are better placed to assess whether a trade alert, a voice record and an eComms exchange form part of the same conduct picture.
Surveillance Intelligence Can Strengthen FinCrime Defence
The panel also pointed to a wider value case for surveillance data across control functions. Market abuse alerts can expose more than trading misconduct. They may reveal relationship networks, communications patterns, event timing and money-flow context relevant to financial-crime teams. That makes market-abuse surveillance data a potential intelligence asset for AML, entity resolution and network analysis, while financial-crime context can help surveillance teams assess whether a trade, message or voice record forms part of a wider risk pattern.
Check Out the Webinar
The webinar’s value lies in moving beyond familiar integration themes and into the operating details that determine whether holistic surveillance is defensible and delivers value. The panel explores where evidence should surface, how escalation should be owned, why behavioural tagging matters, how AI changes triage, and how market abuse connects to wider financial-crime intelligence. Watch the full RegTech Insight webinar, Holistic Conduct & Surveillance: Unifying Trade, Voice, and eComms Data with AI, to hear the discussion in full.
Subscribe to our newsletter
