Every Desk Has a Retyping Problem. Trade Booking Is Just the Most Expensive One.

This is a front-office automation problem: the everyday manual re-entry that follows a negotiation in chat. A trader agrees a price with a broker over chat. Then someone retypes it into another system by hand: a pricing blotter, a booking system, a lending desk's spreadsheet. The system may change, but the problem doesn't. It exists on every desk that negotiates in chat, every day.

Inefficient, or non-existent, retyping of information contained in chat can lead to missed trading opportunities and poor pricing. Delays in booking trades can lead to significant unnecessary risks, out-of-date trading limits, and suboptimal use of capital. If information is incomplete, post-trade compliance queries can take days to trace back to the original chat message, while missed reconciliation and reporting deadlines can lead to big fines.

Chat platforms have been opening up for a reason

The direction of travel over the last few years has been the same across every major chat platform: more of what happens inside a conversation becomes something a system outside the chat window can actually access and act on. That's part of what's made this possible to build. It still isn't enough on its own.  While deterministic approaches like regex are useful for highly structured message data and general-purpose LLMs are increasingly useful for interpreting semi-structured chat messages, neither can cope with  the shorthand a desk uses to confirm a trade: that vocabulary is private, and it varies by desk, by counterparty, by asset class. Why Trade Capture is Still a People Problem in 2026 covers why that combination kept this manual for so long. 

What changes: the conversation becomes the data

ipushpull is a workflow automation platform combining seamless integrations, powerful message interpretation and mapping, human-in-the-loop approval and auto-learning. Whatever the source- a chat message, an email, a spreadsheet- it gets captured, turned into structured data, and sent wherever it needs to go next. A person still confirms the result before anything moves downstream. When someone corrects a field, that correction trains the model on their own shorthand and their own counterparties, so it needs less correction next time. 

Trade booking: the clearest version, and the most expensive

Trade capture and booking is where inefficiencies of manual capture, reformat and re-entry have the biggest impact and where getting it wrong costs the most. Koch Energy Services, one of North America's largest physical gas traders, replaced manual booking with exactly this approach. Trade-entry time is down 85%, and booking accuracy is in the high 90s, across more than 50,000 trades processed. As Koch's Tim Flynn put it: "This is about freeing traders to focus on markets, not mechanics. Less typing, fewer errors, faster outcomes."

None of that required traders to change how they talk to their brokers. Traders keep chatting with brokers exactly as before; the retyping is what disappears.

The same problem, other desks

At one investment manager's FICC desk, traders used to copy portfolio manager instructions into chat, negotiate across several dealers, then key the agreed prices back into internal systems by hand. Automating that chain removed the same re-entry step: the desk moved faster and carried less operational risk. A securities lending desk fielding availability and rate updates from agent lenders over email faces the same shape of problem: someone still keys those updates into a blotter by hand.

Trade booking gets the attention because the fields are standard and the cost of getting it wrong is highest. It isn't the only place a desk still copies information from one format into another by hand.

A desk with a retyping problem already has the technology to fix it. What's missing is the decision to use it.

Related reading: