Milano, Italy — July 1st, 2026 – Fixed-income traders build RFQ layouts, workflows and pricing rules in plain language, in minutes. And whatever AI a bank already runs, nexRates gives it the rails to reach the desk — with the trader always in command.
SoftSolutions has brought AI to the nexRates e-trading platform, giving fixed-income traders the agility to move at the speed of the market. A trader now describes the RFQ layout, workflow or pricing rule they need in plain language and has it generated automatically — no developers, no vendor requests, no manual configurations. The AI also analyses desk dynamics and supports decision making tailored to each client’s history and tier. It never prices, quotes, or executes — the trader stays in command of every decision — and it integrates to the AI agents banks already run in-house, leaving each client’s own AI strategy intact.
What the trader gets
The capability meets traders in their own language and handles work that used to require developers or vendor time:
- Context-dependent RFQ layouts — templates that adapt to instrument, venue and desk logic, generated from a plain description.
- Workflow management — routing, handling and automation rules expressed in plain terms and turned into working logic.
- Interface design on the desk’s terms — screens shaped around how a trader actually works.
It does not start from a blank page, and it cannot go off-piste. The capability is grounded in production-proven RFQ rules and templates already running on live desks, and it works only against a closed catalogue of the fields and actions that exist on the platform — it cannot invent one. Anything outside that catalogue routes to development, never onto a live grid. Every configuration is validated and previewed, and the trader confirms it before it applies. Risk limits, compliance gates, audit trail and supervisor approval apply throughout. This is what makes it agility a desk can trust.
Intelligence that suggests, never executes
Beyond building layouts and workflows, the AI works as an analyst on the desk. It studies how each counterparty behaves over time, factors in tier, and proposes where to sharpen pricing and response. The trader sees the options and decides.
This is decision support, not automation. The AI brings the pattern, the history and the options into one view, so the trader makes a faster, better-informed call — and every call stays in context, so the desk’s own decisions refine what the AI proposes next. It never prices, quotes, or executes on its own. Auto-reply runs only on the rules the trader sets, with MiFID II / RTS 6 alignment preserved.
Open to your own AI
The capability we ship is one expression of what nexRates now enables — not the limit of it. The platform is open by design.
A bank that has invested in its own AI — proprietary models, in-house LLMs, third-party agents in production — connects that intelligence to nexRates through open interfaces, letting it work on the desk’s layouts, workflows and strategy, within the same risk and compliance controls that govern every user. This is a deliberate departure from the closed-stack approach of monolithic platforms, which require you to adopt the vendor’s AI to use it at all.
The same engineering that made our own AI safe to run on a live desk — the closed catalogue, the validation, the audit and control layer — is what makes the platform safe to open to yours. nexRates provides the execution and workflow rails. Your AI strategy stays your own.
“Every sell-side desk is under pressure to show what AI does, and most face a poor choice: adopt a vendor’s closed model, or fund a multi-year build. We reject that choice. The hard part was never the model — it was the governed layer that lets any model touch a live grid safely. We built that for our own AI, and now we are opening it to yours.” — Giovanni Esmanech, Chief Executive Officer, SoftSolutions
Giovanni Esmanech, Chief Executive Officer, SoftSolutions
Where the budget already is
No bank has budget to replace its core OMS. Every bank has an AI budget. nexRates turns that AI budget into trading-desk value — better tools and better decisions for traders, with no rip-and-replace project to fund.
This is the first product expression of our Trading Agility program: the conviction that a competitive fixed-income desk wins on how fast it adapts, and that AI should make the trader faster — not take the trade.
Why you trust it
The reason this AI is reliable enough to live on a trading desk is the work behind it. A wrong answer here is not an inconvenience it is a risk, so reliability could not be assumed. It had to be earned.
Generic AI is easy to show and difficult to trust. Drop a language model onto a trading screen and it will happily generate something plausible — and occasionally something wrong, unbounded, or impossible to audit. On a fixed-income desk, plausible-but-wrong is the most dangerous output there is. So we did not start from a model and look for uses. We started from the desk and worked backwards, and the reliability comes from three things:
- Long, domain-specific training. The AI was trained and tuned over an extended period against the actual language, instruments, workflows and edge cases of fixed-income trading — not general finance, but the specific reality of RFQ handling, market-making and government-bond desks, and the way traders phrase what they want. Each round of refinement narrowed the gap between what a trader meant and what the system produced.
- Relentless work on response quality. Getting an AI to answer is trivial. Getting it to answer correctly, consistently and usefully — to turn a half-sentence from a trader into a configuration that is right the first time — is the work. We iterated until the output was something a desk could rely on, not second-guess.
- A platform that holds the line. The training is reinforced by hard constraints: the AI works only against a closed catalogue of what exists, every output is validated and previewed, and the trader confirms before anything goes live. Trust comes from the combination — a well-trained model inside a platform that keeps it on-piste.
The result is not an AI that looks impressive in a demo, but one that earns a trader’s confidence on a real request: a result they would have accepted from their own developer.
See what it can do for your desk
[▶ Watch the nexRates AI preview] — the capability in action on the desk.
A deep-dive session shows what it does for you — run on your own workflows, with our team walking your traders through it and answering the questions that matter most to your desk.
[Book a deep-dive session] — or get in touch with any question.
