AskAda has grown from an AI assistant into an AI agent that checks payment, invoice, and validation status for suppliers in seconds, securely and in their own language.
Ask anyone who has worked in supplier support what the most common question is, and you will hear the same answer: “Where is my payment?”. It is a fair question. For a seller, an unpaid invoice is cash flow that hasn’t arrived, and not knowing why is worse than knowing it is late.
Until now, answering it took several steps. A supplier opened a ticket, a support agent looked up the document in both the supplier and the buyer account, confirmed whether it had been delivered, approved, or rejected, and then wrote back. For the supplier that meant waiting. For the buyer’s AP team and for our support team, it meant the same question handled by hand, over and over.
With the latest release of AskAda for Sellers, suppliers can simply ask. AskAda now looks up the live status of their invoices and payments on the Tradeshift platform and answers in seconds, with no ticket and no waiting. Asking about their payments is only the first step: sellers can now ask Ada to find a document, check an order, look up a customer, or explain what happened to an invoice, all in the same conversation.
From product guide to AI agent: how AskAda has evolved
AskAda carries a name with some history. Ada, after Ada Lovelace, the first computer programmer, has been the name behind Tradeshift’s AI since 2017, long before AI assistants became a category of their own. AskAda for Sellers is the latest chapter of that story, and it has moved quickly.
We introduced AskAda for Sellers in December 2024 as a beta: an always-on digital assistant that answered product questions in seconds, sourcing its answers from our Knowledge Base, product resources, and website. In our Spring 2025 release it became available to all paid seller subscribers, and by August 2025, as we shared in our Fall 2025 release, it was live for every active seller account on the platform, in multiple languages, with a Babelway version launched earlier that summer.
That first version was very good at one thing: explaining how Tradeshift works. What it could not do was tell a supplier anything about their own documents. If a question was about a specific invoice, AskAda pointed the supplier to the Support Center. The next step was obvious, and it is the one we have now taken: AskAda has gone from knowing the product to knowing your data.
What AskAda can answer now
The new AskAda focuses on the three questions suppliers ask us most often about their documents:
- Payment status: “Where is my payment for invoice 10452?” or “Has my customer paid this invoice yet?”
- Document status: “Was my invoice accepted?” or “Why was invoice 10452 rejected?”
- Validation status: “Why couldn’t my invoice be processed?”
Suppliers ask in plain language, in their own language, and get an answer based on the real-time data on the platform. AskAda also keeps track of the conversation, so a supplier can follow up with “and what about the one I sent last week?” without starting over. Product and how-to questions are still answered from the Knowledge Base, just as before, so one assistant covers both.
What this means for sellers:
- Answers in seconds, not days: the current status of any invoice or payment, without opening a ticket
- In their own language: suppliers ask and receive answers in the language they work in
- Clear next steps: when an invoice is rejected or fails validation, suppliers see why and can act on it right away
- Only one screen for all their answers: regardless of what they are trying to find out, sellers now don’t need to spend time searching for the place where to find their answer
How it works: an AI agent, not a scripted chatbot
Behind the same chat window, AskAda now runs on an agentic architecture. When a supplier asks a question, an orchestrator agent works out what they are asking and which document or flow they mean. If the question is about the supplier’s own data, the agent selects the right tools through the Tradeshift MCP Server, retrieves the document, payment, order, or network information, and turns it into a clear answer. If the question is about how the platform works, it draws on our product knowledge instead.
This matters because the agent is not following a fixed script. It can combine several lookups to answer one question, handle follow-ups in context, and use the same open, standards-based tools we are making available across the platform. It is the same foundation that powers our AP Auditor agent for buyers, now working on behalf of sellers.
One note to keep in mind: your data is not training data
Seller documents, questions, and AskAda’s answers are not used to train or fine-tune AI models. The AI model has no standing access to your account and for each question, AskAda retrieves just the documents relevant to that answer.
Secure by design
Giving an AI assistant access to financial data only works if it is built to be trusted. We designed AskAda so that security is enforced by the platform, not left to the AI model.
- No new logins: suppliers use their existing Tradeshift session, and AskAda only sees what that user is already allowed to see
- Company data stays with the company: the platform, not the model, decides whose data is queried, so a question cannot be phrased to reach another company’s documents
- Read-only by design: AskAda answers questions; it cannot change any data on the platform
- Short-lived memory: conversation context is kept only for the active session
- Protected and auditable: built-in defenses against prompt injection, and every lookup the agent performs is logged
Why it matters for buyers and AP teams
Every “where is my payment?” email that a supplier doesn’t need to send is one that your AP team doesn’t need to answer. When suppliers can check status themselves, AP teams spend less time on inquiries and more time on the work that needs them, and supplier relationships benefit from faster, more transparent communication.
It also works alongside the Payment Predictor for Sellers, which we upgraded this fall. Payment Predictor tells sellers when they are likely to be paid; AskAda tells them where things stand right now. Together, they give suppliers the visibility they need to plan their cash flow with confidence.
What this means for your team:
- Fewer supplier inquiries: routine status questions are answered before they reach your inbox
- A better supplier experience: instant, transparent answers build trust and support supplier adoption
- No extra setup: AskAda is built into the Tradeshift platform and uses the permissions already in place
Part of a bigger shift toward agentic AI
AskAda’s evolution mirrors where Tradeshift AI is going as a whole: from software that answers questions to agents that do useful work. Over the past year we rebuilt our invoice coding engine with Ada 2.0, launched AI Document Intelligence, introduced the AP Auditor agent, and opened up the platform through the Tradeshift MCP Server. You can read more about the road ahead in our 2026 vision for agentic AI.
What connects all of it is the same principle behind AskAda: AI that works on real platform data, inherits the permissions of the person using it, and leaves a clear audit trail. AskAda for Sellers shows what that looks like from the supplier’s side of the network.
To try it, suppliers can open AskAda from the icon in the bottom left corner of their Tradeshift dashboard. For everything else in this release, watch our Fall Release webinar on demand.