Avalara launches agentic AI platform to automate tax compliance globally

Avalara has introduced a platform for agentic tax and compliance, featuring AI agents capable of initiating and executing compliance workflows from beginning to end.

The new platform, termed Agentic Tax and Compliance, is supported by the ALFA framework, Avalara's proprietary large language model (LLM) set-up designed specifically for agentic applications. By integrating trusted content, advanced language models, agentic middleware, and scalable technology infrastructure, the system aims to deliver compliance execution in real time.

Scott McFarlane, Chief Executive Officer and Co-Founder of Avalara, highlighted the purpose-built nature of the technology:

"The most trusted AI breakthroughs are purpose-built. Just as Harvey AI transformed legal workflows and PathAI redefined diagnostics, Avalara is now setting the standard as the domain-specialised leader in agentic tax and compliance."

From assistance to execution

While many companies leverage AI as copilots to assist professionals, Avalara's AI agents move beyond simply aiding tasks to completing them autonomously.

These agents, embedded within environments such as ERP and eCommerce systems, can observe, advise, and directly execute compliance work that previously required significant departmental resources.

The platform's structure combines AI agents, Model Context Protocol (MCP) servers, private LLMs, and small, proprietary language models, all configured using the ALFA framework, in an attempt to redefine how tax and regulatory automation is approached.

Agentic compliance model

Avalara has centred its offering on the concept of agentic compliance, where compliance mechanisms are embedded into operational processes from the outset, rather than applied as an afterthought.

These AI agents repeatedly integrate with business systems, monitor data and workflows, advise on compliance actions and risks, and execute calculations, filings, and document validations without human intervention.

This shift aims to convert compliance from a resource-heavy, highly manual activity to an automated, continuous function, managed by scalable AI agents.

Demonstrating automation

For instance, the Returns AI agent is built to ingest transaction data, apply relevant forms and jurisdiction jurisdictional logic via Avalara's Returns APIs, and file returns following approval, thus functioning as a digital compliance professional which can operate round-the-clock while staying current with changing rules and jurisdictions.

Technology and infrastructure

Avalara's platform has an active-active, multi-cloud foundation spread across multiple geographies and cloud service providers, including AWS, Azure, GCP, and OCI. This infrastructure supports horizontal scalability and currently averages response times of 15 milliseconds, due to its distributed architecture. The company has built over 1,400 partner integrations to facilitate seamless embedding of its AI agents in business applications, inboxes, workflow systems, and data streams.

Jayme Fishman, Executive Vice President, Chief Strategy and Product Officer at Avalara, emphasised the new capabilities for clients:

"Compliance has always been complex and resource intensive, and Avalara has worked for years to make it easier for businesses of all sizes. Now, we're delivering the first-ever digital compliance professionals, AI agents that advise and execute at global scale, making it even easier to handle the most onerous and time-consuming tasks-so businesses can move faster and focus on growth."

Agent-to-agent collaboration

Avalara's AI agents are designed for interoperability across enterprise systems. They can collaborate with agents from various origins-such as ERP, Point of Sale or eCommerce platforms-to complete compliance tasks quickly and accurately. Avalara refers to this as a force multiplier effect, where agentic flows are enhanced across the enterprise through seamless collaboration.

These agents are made to be portable, operating wherever business work is conducted. Integration points include development environments, productivity tools, file storage systems and user devices, enabling tax validation, product classification, and form submission to take place at the point of data creation rather than requiring data to be sent elsewhere for processing.

Agentic infrastructure coverage

The agentic infrastructure is designed to serve multiple compliance scenarios, including preparation of VAT returns in Europe, facilitating marketplace rules in the United States, excise tax processing in Brazil, and licence validation in Canada. Avalara claims to provide the only comprehensive agentic infrastructure spanning the full breadth of compliance activities.

Focus on content-driven accuracy

Avalara has applied its extensive compliance content repository to train its own small language models, which are intended to enhance private LLMs by embedding comprehensive domain expertise. These models do not learn from untrusted content, which is intended to further support compliance accuracy in output.

Platform architecture

The ALFA framework, which underpins all of Avalara's AI functionality, includes enterprise LLMs with strict data isolation from major providers, proprietary SLMs trained on billions of compliance data points, libraries of prompts, guardrails for reasoning, RAG databases, and APIs. The extensible agentic architecture is open to developers, with MCP server access available for integration and development purposes.

User experience

With the implementation of agentic automation, Avalara aims to simplify compliance tasks, allowing users to delegate work to AI agents through intuitive queries, eliminating the need for users to interact with complex interfaces.

The platform now supports a range of tax and compliance solutions, including AvaTax, AvaTax for Accounts Payable, Returns, VAT Returns & Reporting, E-Invoicing, Exemption Certificate Management, Business Licences, Tax Research, Item Classification, Cross-Border, Tariff Code Classification, 1099 & W9, and Property Tax. According to Avalara, each of these is being adapted to benefit from agentic automation, resulting in higher automation rates and reduced friction for users.

Strategy and future direction

Avalara's strategy includes open networks, sandbox testing environments, low-code partner possibilities, and robust operation across cloud platforms. The agentic platform is designed to be embedded, consistently compliant, and accessible for users ranging from developers to executives. The focus is on integrating deep domain knowledge with agentic infrastructure and actionable models to automate compliance at scale.