2025: The Year Chat Data Became a Strategic Asset

In our 2024 round-up, we argued that the most valuable information on a trading desk sits inside chat but has never been fully utilised as an asset. It moved fast, vanished quickly and could not be extracted easily or governed well enough to drive automation. Firms knew the value was there, but had no reliable way to do anything useful with it.

2025 is the year that changed. Extraction has now reached a production standard. Conversations could finally be converted into structured data with the fidelity required for front office execution and downstream processing. The shift was not about novelty; it was about unlocking a source of value that had always been present but never economical to use.

ipushpull was already built for this moment. While AI matured, we spent years constructing the rails automation depends on: capture, transformation, governance, orchestration and delivery. When extraction became reliable, the architecture absorbed it immediately. Chat data is now not just something to read, but something to use.

This round-up explains how that shift played out across markets in 2025 and why firms are now treating chat as a strategic asset.

2024’s bets, 2025’s reality

1. Extraction became dependable enough for production workflows

The constraint that held automation back was not intent or demand. It was extraction fidelity. In 2025, that barrier was lifted. Conversations that once produced unstructured noise now produce structured workflow objects with clear fields, lineage and the control required for regulated environments.

The clearest example is our physical gas booking agent. It interprets compressed trader language, validates the details, and books trades with human oversight. The workflow is identical to how desks communicated before, except the output is now a compliant trade record rather than a message thread needing to be manually re-keyed.

Read: ipushpull launches LLM-enabled agent to automate physical gas trade booking via chat

This is the step that turned chat from just communications and into workflow infrastructure.

Partnerships as evidence of the shift

Adoption in 2025 was driven by increased demand from desks with immediate commercial need.

StoneX selected ipushpull to improve their clients' experience by enabling them to send data into chat, where the interactions occur.

Read: ipushpull selected by StoneX to enhance client experience

Partnering with Koch Energy Services,  a physical gas booking agent, demonstrated that chat-to-execution automation is not a concept but a live production workflow.

Read: ipushpull launches LLM-enabled agent to automate physical gas trade booking via chat

These partnerships did not validate AI. They validated that firms now treat chat data as a strategic asset worth structuring and reusing.

2. The platform foundations were already in place

Firms discovered this year that successful AI deployment depends less on the model and more on the infrastructure around it. Because the foundation work was complete, the improvements in extraction simply unlocked capabilities we had already designed for:

• ingestion and normalisation across all major chat platforms

• transformation from free-form messages into structured outputs

• workflow orchestration suitable for front office and post-trade processes

• enterprise-grade governance, permissions and audit

• security credentials that satisfy regulated institutions

We did not add automation to chat; we made chat safe to automate.

Recognition aligned to workflow impact

Awards this year reflected the impact in the market.

• Best Price Sharing and Publishing Solution, TradingTech Insight Awards Europe

• One of the most influential financial technology firms of 2025, The Financial Technologist

Each reflects the same point: structured, governed chat data now underpins real trading and distribution workflows.

3. Chat data became structured, governed and automated

Chat as a structured trading input

RFQs, quoting, negotiation and confirmation continue to take place in chat. The difference in 2025 is that these conversations now produce the structured booking objects required for straight-through workflow.

Read: Turning Conversations into Trades: How RFQs Inside Chat Are Changing Buy Side Workflows

Governance as the enabler

Our Capture layer now absorbs chat, files, spreadsheets and APIs, standardises them and preserves lineage. This is the prerequisite for scaling AI safely and consistently.

Read: Bringing Order to the Chaos: The Capture Layer Under the Hood

Orchestration as the unlock

With data reliable and structured, the bottleneck shifted to validation, enrichment, routing and controlled execution directly from chat.

Read: Turning Forgotten Chat into Tomorrow’s Data

Agents as workflow users

Delivery now spans both humans and AI agents through standards such as MCP. Chat data feeds more than screens. It feeds automated decision makers.

Read: Delivering Data and Services in the Age of AI Agents

Across fixed income, commodities and OTC markets, desks have begun treating chat as a primary workflow surface, not an informal side channel.

Read: Fixed Income Leaders Summit 2025: Chat-Driven Automation – The Real Front Office Revolution

2025 Product Highlights

Each highlight below reflects a single shift: conversational intent is now structured, actionable and production-safe.

Chat to booked trade automation
Read: ipushpull trade capture and booking from chat to booked trade in near real time

RFQs and quoting inside chat
Read: Turning Conversations into Trades: How RFQs Inside Chat Are Changing Buy Side Workflows

Capture layer industrialised for AI-ready workflows
Read: Bringing Order to the Chaos: The Capture Layer Under the Hood

2026

2025 unlocked strategic value. 2026 will expose the operational demands of running AI-driven workflows in production. It's not about model selection - these will continue to improve and improve. The challenge will be governance, lineage, permissions, orchestration and the need to support both humans and agents without fragmenting the workflow.

Our focus is to deepen automation, reduce manual intervention, strengthen human-in-the-loop controls and extend delivery to where our customers’ clients work. The objective is straightforward. Turn conversational intent into governed action with minimal friction.

The takeaway

Chat data was always valuable. It simply wasn’t usable.

2025 changed that. Extraction became reliable. Governance and orchestration made it safe. Firms began treating conversation as structured workflow material rather than transient text.

We built the foundations early. This year showed why that decision mattered.