2026 has been a big year for AI at Tradeshift. We rebuilt our invoice coding engine from the ground up, shipped a document intelligence layer that reads invoices without needing a template, moved Reporting & Analytics onto a fully conversational, agentic foundation, and started shipping the first pieces of what we call our agentic AI roadmap.
In this context, we are pleased to be recognised by independent industry observers:
That acknowledgement matters more to us than the specific placements. It’s one thing to believe, internally, that the investment in AI is paying off exactly as intended. It’s another thing when someone with no stake in the outcome looks at it from the outside and describes those same capabilities back to you, unprompted. In our view, this external recognition reflects our ongoing commitment to our capabilities, our strategy, and the investments behind them.
So here’s what actually changed this year, capability by capability, and what it means day to day for the people using it.
It combines OCR, built on AWS Textract, with large language models to read invoices the way a trained reviewer would: interpreting context and structuring the data correctly even when it has never seen that supplier’s invoice format before, whether the file is a native PDF, a scan, or an image.
Benefit: Our current extraction accuracy sits above 95% across native and image based PDFs, and since roughly 23% of the documents flowing through our platform every month arrive in these harder to process formats, that accuracy shows up directly as fewer exceptions and less manual review. It also removes the ongoing cost of template maintenance, so automation rates keep climbing instead of degrading as your supplier base grows more diverse. Read more about AI Document Intelligence.
Ada 2.0 rebuilt our original invoice coding engine, first shipped back in 2017, replacing its statistical model with a decision tree classifier trained on each company’s own historical coding data rather than one generic model shared across customers.
Benefit: It holds its accuracy steady even as your invoice population gets more diverse, and administrators can set confidence thresholds that decide how many invoices get auto coded versus routed for review, with an Automation Dashboard that shows exactly how much of the coding workload AI is handling. It also now reaches beyond coding into supplier matching, routing suggestions, and product classification. Read more about Ada 2.0.
Our Reporting & Analytics app, built on Amazon Quick, gives AP and finance teams 16 dashboards across nine domains of AP data, plus a conversational AP Auditor agent that answers plain language questions like show me pending invoices from last month, no ticket required.
Benefit: Query response times are up to 30 times faster, while users save 6 to 8 hours a week each. Spotting an operational bottleneck now takes under five seconds instead of one to two days. Together with AWS, we wrote about how the app evolved in just a year, from surfacing insights to enabling actionable intelligence: read it here.
The Anomaly Detection dashboard lives inside Reporting & Analytics and automatically flags two kinds of irregularities: high value documents that fall outside expected thresholds, where duplicate invoices and fraud usually hide, and unusual patterns in workflow behavior, like repeated approval reassignments that can point to a process or control gap. It looks back six months of data to tell a genuine anomaly from ordinary seasonal variation.
Benefit: It’s included for every buyer account with no extra setup or license, and it shifts finance teams from reviewing problems after the fact to catching them before they escalate, without anyone manually combing through spreadsheets to find what’s off. Read more about Anomaly Detection in AP Analytics.
Underneath the AP Auditor chat experience is the Tradeshift MCP Server, which we built in house rather than bolt onto a generic framework, so every AI agent inherits the exact same authentication, role based permissions, and audit trail a human user already has. It exposes 95 tools across six domains of the platform and connects directly to Amazon Quick’s agents, so someone can ask a question in plain English (or any other languages supported in Amazon Quick) and watch an agent run several analyses in parallel instead of one at a time.
Benefit: Scenario planning that used to take several days and a few people now takes one person a matter of minutes, and it’s why we believe we’re among the first B2B AP platforms running genuine agentic AI in production rather than in a demo. Our CPTO, Raphael Bres, wrote more about where this is headed in Autonomous Finance: The 2026 Vision for Agentic AI at Tradeshift, and our Director of Data Science, Robert Iordache, about how the MCP Server itself works in Inside the Tradeshift MCP Server.
None of this happened separately. AI Document Intelligence and Ada 2.0 both work on getting an invoice into the system correctly, and Reporting & Analytics, including Anomaly Detection, runs on top of the data they’ve already processed. The agentic layer inside Reporting & Analytics, powered by the Tradeshift MCP Server, is the newest piece, and it’s a preview of where several other parts of the platform are headed next.
As Raphael Bres, our Chief Product & Technology Officer, outlined in Autonomous Finance: The 2026 Vision for Agentic AI at Tradeshift, more agentic capabilities are already on the roadmap, such as: the AI Document Supervisor, a context-aware agent that spawns specialized extraction agents based on document type and complexity, building on what the AI Document Intelligence does today. And an AP Compliance Expert Agent, trained on compliance and regulatory documentation, to help review documents against local mandate requirements.
Forrester disclaimer:
Forrester does not endorse any company, product, brand, or service included in its research publications and does not advise any person to select the products or services of any company or brand based on the ratings included in such publications. Information is based on the best available resources. Opinions reflect judgment at the time and are subject to change. This report is part of a broader collection of Forrester resources, including interactive models, frameworks, tools, data, and access to analyst guidance. For more information, read about Forrester’s objectivity here .
Gartner Report, Magic Quadrant for Accounts Payable Applications, By Miles Onafowora, David Condon, June 2026. Gartner and Magic Quadrant are a trademark of Gartner, Inc. and/or its affiliates.Gartner does not endorse any vendor, product or service depicted in its research publications, and does not advise technology users to select only those vendors with the highest ratings or other designation. Gartner research publications consist of the opinions of Gartner’s research organization and should not be construed as statements of fact. Gartner disclaims all warranties, expressed or implied, with respect to this research, including any warranties of merchantability or fitness for a particular purpose.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。