How we approach generative AI at Tradeshift to transform our users’ experience

With instant, multilingual answers, Tradeshift’s AskAda showcases how generative AI transforms support for global sellers

In today’s digital economy, the finance manager’s day is defined by speed, accuracy, and the ability to act on information without delay. At Tradeshift, our mission has always been to remove friction from business transactions and make trade seamless for everyone. Over the past two years, we’ve been exploring Generative AI, one of the most exciting developments in technology, and how it can reshape the way our users interact with our platform.

That exploration led us to AskAda, our AI-powered support assistant. AskAda is more than just a chatbot. She’s designed to understand user intent, provide accurate, context-aware answers, and guide users to the right resources when needed. Her job is simple: make every interaction with Tradeshift faster, easier, and more productive.

👉 AskAda for Sellers is not the only conversational AI available on the Traeshift platform. Read about the Q functionality available in the Reporting & Analytics App.

Why Generative AI matters for our users

Generative AI (GenAI) has been one of the most transformative technologies in recent years. Unlike traditional automation tools, GenAI can generate responses, summaries, and insights in real time based on a deep understanding of context.

For the sellers using Tradeshift every day, this means less time searching for answers and more time focusing on strategic work. Invoices get submitted faster, errors get fixed sooner, and questions can be answered without a maze of clicks and calls.

AskAda: Conversational AI built for sellers, beneficial for buyers

Tradeshift connects buyers and sellers around the world, but we designed AskAda specifically for sellers: the companies sending invoices to their buyers through Tradeshift. Sellers are the ones navigating our workflows daily, checking invoice statuses, fixing rejections, and trying to keep cash flow predictable.

When we help sellers work more efficiently, we’re also helping buyers. Fewer errors, clearer communication, and faster resolutions benefit both sides of the transaction. And in a network where collaboration drives value, those small wins add up quickly.

The journey of AskAda: Three waves of innovation in AI-powered support

We rolled out AskAda in phases, using each stage to learn and improve before expanding her reach.

Wave 1: November 2024 – Launch to subscription sellers

We started small, offering Ada to sellers on subscription plans. This group gave us fast, detailed feedback that helped us refine her tone, accuracy, and guidance before scaling further.

Wave 2: April 2025 – Beta for a select group of sellers

By April, we were ready for a bigger audience. A select group of seller users joined the beta, allowing us to see how Ada performed across different industries and regions. This was a real test of her scalability and versatility.

Wave 3: August 2025 – Full Rollout with Smarter Capabilities

 In August, Ada became available to our full network of active sellers. This wave also brought major upgrades:

• A richer, more up-to-date knowledge base so she can answer a wider range of questions accurately.

Better guidance when she can’t help directly, with a direct link to our Support page so users can escalate without losing time.

👉 Forecast when your invoice will be paid with Payment Predictor for Sellers.

What AskAda can do today and the technology powering its conversational AI

Right now, AskAda delivers a high-impact AI-powered support experience for sellers by:

• Answering product and process queries using a curated knowledge base.
• Guiding users to the right features, reports, or help resources.
• Responding in more than 50 languages. She’s handled questions in French, German, Spanish, Japanese, and, memorably, even Fijian.
• Keeping conversations secure, respecting the same privacy standards as the rest of Tradeshift.

If a seller asks, “When will my invoice get paid?”, Ada will point them to the ML-powered Payment Predictor report, which aggregates key signals to estimate payment dates. She then guides them through where to find that prediction so they can plan accordingly.

The technology powering AskAda’s conversational AI

AskAda runs on a generative AI engine powered by an advanced Retrieval Augmented Generation (RAG) system tuned specifically for the Tradeshift ecosystem. Unlike general-purpose AI tools, she understands our workflows, product terms, and the nuances of supply chain and finance language.

We’ve built her with a focus on:

• Accuracy: Answers come from verified, up-to-date resources.
• Deep, domain-specific intelligence: AskAda’s RAG system allows us to build expertise specifically tuned to Tradeshift’s workflows, product terminology, and the complex nuances of accounts payable, supply chain and finance language, making her a true specialist.
• Responsible AI: She won’t guess on sensitive topics and will escalate to human support when needed.
• Continuous learning:  Interaction data (anonymized) helps us identify where she needs to improve and which topics to prioritize.

Real-world impact and what’s next for AskAda for Sellers

The response from sellers has been overwhelmingly positive. Many have cut down on the number of support tickets they open, resolving their issues directly in-platform in minutes instead of hours. 

The result: faster turnaround on payments, fewer costly mistakes, and more time for work that drives growth.

What’s next for AskAda

We’re proud of how far Ada has come, but we’re just getting started.

• Multilingual support is already here. Whether a user writes in English, Mandarin, or Fijian, Ada replies in the same language instantly. This is a huge benefit for our global user base, making support feel personal and accessible everywhere.
• Smarter answers with Payment Predictor. Today, when a seller asks “When will my invoice get paid?”, Ada explains where to find that information in Payment Predictor and helps them access it. In the future, she’ll go a step further: pulling the predicted payment date directly into the chat. Instead of pointing you to the report, she’ll be able to say, “Your invoice INV-12345 is predicted to be paid on October 8, 2025. Want to see the full breakdown in Payment Predictor?”
• Even more seller-focused intelligence. Ada will continue to be tuned to the specific challenges sellers face, from submission errors to navigating compliance requirements, while still delivering value to buyers through smoother, faster seller interactions.

Why this matters for finance managers

For finance leaders, time really is money. Every delayed invoice payment impacts cash flow and supplier trust. AskAda gives sellers (and by extension, buyers) a way to get answers instantly, in their own language, and without the usual bottlenecks.

By embedding this AI-powered support directly into Tradeshift, we’ve made AI a practical, everyday partner that reduces friction and keeps transactions moving.

Generative AI is about amplifying human intelligence. At Tradeshift, we’re committed to using AI responsibly, with a laser focus on making our users’ lives easier. AskAda is proof that when you combine advanced technology with a deep understanding of user needs, you can transform the everyday experience of doing business.

Explore more about Tradeshift’s AI capabilities and see how we’re shaping the future of digital trade, or directly get in touch with one of our representatives.