How Kyriba adopted Dust while building their own AI product

Industry: Financial Services
Company Size: 1001+
Departments: Company-Wide AI Adoption
Executive

Key Highlights

1,122 active Dust seats across a 1,300-person company spanning France, the US, the UK, Poland, Singapore and Brazil
50+ formal AI champions with adoption targets built into yearly objectives
2.5 years, zero data incidents in a heavily regulated treasury and finance environment

About Kyriba

Kyriba is a 1,300-person Franco-American fintech that provides treasury management, payments and liquidity solutions to CFOs and treasury teams worldwide. With offices across six countries and a customer base of enterprise finance departments, Kyriba operates in one of the most regulated corners of financial services: the space where companies manage cash, mitigate risk and move money.

What makes Kyriba unusual in the AI landscape is that they're not just buying AI tools. They're building one. TAI, Kyriba's own agentic AI for treasury, is a customer-facing product that helps their clients automate treasury workflows. So when the company also adopted Dust as its internal AI platform, the natural question was: why buy when you're already building?

The answer, it turns out, is that buying Dust is part of how you learn to build.

The Challenge

When "we have our own AI" becomes a reason not to act

For most companies considering an internal AI platform, the presence of an in-house AI product creates a convenient excuse: we already have something. Kyriba's situation was more nuanced. TAI was built for customers. Employees still needed a way to work smarter internally, but justifying a second AI investment required a clear argument.

"I don't have a lot of budget, of course. So we have to be very innovative," Nevena Gecheva, Kyriba's Director of AI Excellence and Innovation, explains. "But we are a company of 1,300 people, and there are always a lot of enthusiasts."

From prototyping tool to platform

Nevena's team first encountered Dust in September 2023 while searching for RAG solutions. They'd already spoken to OpenAI and Google, neither of which had even published pricing at the time. In November, Kyriba ran a hackathon with Dust. By January 2024, the decision was made: Dust would be the platform.

"We first found Dust to be a very good prototyping tool," Nevena recalls. "And then afterwards, we just liked it. It happened organically."

But organic enthusiasm had limits. Marketing adopted early with high volume but simple use cases. Engineering went deep with sophisticated automations. Sales, the people whose work most directly impacted revenue, barely touched it.

The compliance gate that became a selling point

As a company handling treasury transactions and operating under rigorous compliance frameworks, Kyriba's security team had to sign off on every tool that touched company data. This was never going to be a casual adoption.

"Security was a big gate, but also one of the best arguments I had to actually choose Dust and not any American solution," Nevena says. "Being a European tool, at heart, was a very big plus for us."

The team evaluated alternatives including Claude's enterprise offerings, but the security risks were "way too high." Dust's European DNA, combined with its approach to data handling, ticked the boxes that mattered. Two and a half years in, Kyriba has experienced zero data incidents.

Solution: A Champions Program and a Platform That Coexists With Homegrown AI

Building the champions network

Kyriba's solution to the adoption gap wasn't a top-down mandate. It was a structured community.

The company launched a formal champions program, evolving from an earlier informal volunteer model. The difference: champions now have AI adoption targets written into their yearly objectives, approved by their managers. Over 50 champions operate across the company.

"Before, managers were kind of aware. But now we've implemented it in the yearly objectives of those champions," Nevena explains. "Depending on the person, they have targets for AI adoption or automation."

The tipping point came when the Chief Marketing Officer saw her team present agents at the 2024 sales kickoff. A simple knowledge management agent, basic by today's standards, was breakthrough at the time. Her response: "If you don't have the budget, I'm ready to pay for it." From there, the real breakthrough came when the GTM and sales enablement teams got involved.

The sales agent suite

Working closely with the sales enablement team, Kyriba's innovation group built a suite of agents to make salespeople more efficient. The suite has since evolved: the team is now consolidating multiple agents into a single orchestrator that handles different sales processes in one place.

"Sales might not typically use it, or they use it like a personal productivity tool. But not much sharing," Nevena observes. "So we built a sales suite of agents. It impacts our revenue directly."

Why building TAI made Dust more valuable, not less

Here's the counterintuitive part: building their own customer-facing AI product made Kyriba more committed to Dust, not less.

TAI is aimed at Kyriba's treasury customers. Dust is the internal platform. But the experience gained with Dust over two years directly informed how Kyriba built TAI. The internal teams who learned to work with AI agents, who understood prompting, tool connections and knowledge management, became the people who could design a better product for customers.

"We used a lot of the knowledge and experience we had with Dust to build TAI," Nevena says.

The cost argument seals it. Building a high-performing, well-integrated internal AI platform requires massive investment. Dust's integrations (Slack, Notion, Jira, BigQuery, MCP servers, custom APIs) are, in Nevena's words, a "complete game changer" that can't be replicated internally at the same cost. TAI, by contrast, is deliberately limited to the treasury domain.

"Dust is a platform. It's not just an AI that you use. We're trying to give the same type of experience to our customers through TAI, but it's very limited in terms of what they can connect to."

AI that passes the audit

For UK enterprise buyers in particular, Kyriba's experience with Dust's security posture has been a defining factor. Nevena's manager, who also serves as Kyriba's CISO, has become one of Dust's strongest internal advocates, publicly defending the platform in executive committee meetings.

Dust is now included in Kyriba's upcoming compliance audits, sitting alongside the company's existing SOC and GDPR certifications. For Kyriba, Dust isn't a security liability. It's part of the security story.

Results

The coexistence proof point

Kyriba is living proof that companies building their own AI products don't just need Dust; they need it more. TAI was built better because Kyriba's teams had spent two years learning how to work with AI through Dust. The internal expertise became the product advantage.

The enthusiasm has since spread beyond the original use case. Kyriba's support team wants to offer Dust to their diamond-level customers. The partner team wants to extend it to top-tier partners. The entire company, as Nevena puts it, "is actually a fan."

Visibility drove buy-in, not mandates

Kyriba's adoption was never driven by a board mandate or a finance consolidation play. It was driven by people building things and showing them off. Executive training is now scheduled for the UK office, where Kyriba plans to bring its leadership team up to speed on what their 50+ champions have built.

For enterprise companies wondering whether they can build their own AI and buy a platform, Kyriba's answer is clear: the two don't compete. One trains the muscle. The other ships the product.

Interested in learning more about how Dust can help your team? Visit our solutions page or reach out to our sales team.