We all know the drill by now: You ask a question, and a chatbot retrieves an answer. You upload a document, and an AI tool summarizes it. But while these retrieval-focused tools are excellent for saving individual bits of time, they don’t actually push a business forward on their own. They don’t make decisions or act without direct human involvement.
But that’s all beginning to change with agentic AI: AI tools that will make decisions and execute actions according to preset parameters without direct human involvement.
M-Files is at the forefront of integrating agentic AI into document management, and they have just launched a portfolio of context-aware AI agents designed to help your AI do more for you. They’ve also released an exciting new beta feature: Custom Agents.
Following is an in-depth look at what M-Files AI agents are, the simplified mechanics of how they operate, and how your business might deploy the new Custom Agents to supercharge your productivity, governance, and automation.
What Are M-Files AI Agents?
At their core, M-Files AI agents are specialized, autonomous digital assistants built directly into the M-Files platform. Because M-Files connects every single document to the specific customers, projects, contracts, and people that give it meaning, these agents, unlike traditional chatbots, possess structural “business context.” They don’t just see a PDF invoice; they understand which client it belongs to, whether the associated contract is signed, and who needs to approve it.
M-Files has developed several AI agents designed for specific tasks:
While Custom Agents provide the ultimate flexibility, they work alongside a powerful suite of out-of-the-box agents designed to maintain a pristine, AI-ready data foundation:
- The Metadata Agent: This agent proposes immediate naming, tagging, and classification at the moment a document enters the software ecosystem. The goal of this agent is to eliminate manual indexing.
- The Metadata Agent at Scale: An extension of the Metadata Agent, this agent runs background jobs on a recurring schedule in order to transform messy historical data into structured, AI-accessible archives.
- The Insights Agent: This agent delivers reports and information to users based on their specific role and tasks. It can flag at-risk projects, identify pending tasks, or even note workload imbalances.
- The Task Agent: This agent can automatically generate tasks from user prompts or meeting transcripts. For example, after a client meeting, this agent can generate and assign action items as well as update the client’s interaction record.
- The Quality Agent: Specifically built for quality contexts, this agent can fill in fields like nonconformance descriptions, CAPA action plans, and audit responses based on context.
- The Contracts Agent: Tailored to a contract-heavy environment, this agent can review contracts against organizational policies and can automate governance and audit processes.
While these pre-configured, specialized agents can make a huge difference to most businesses, sometimes you may need something a little more customized.
M-Files Custom Agents
M-Files Custom Agents allow organizations to create their own AI agent. Instead of relying on hard-coded rules or manual human intervention for every decision, a Custom Agent acts as an intelligent checkpoint in a business process. It is a highly malleable, secure assistant tailored by your organization to handle complex tasks.
You don’t need a PhD in data science to deploy or understand an M-Files Custom Agent. The platform distills the operational mechanics into three straightforward, governed steps:
1. Natural Language Configuration
Traditionally, automating complex workflow steps required thousands of lines of code or complex scripting. With Custom Agents, configuration happens entirely in natural language. Here’s an example of a prompt you could use:
“Review this incoming invoice. Check if the totals match our agreed-upon contract milestones found in the related Project folder. If the total is under $5,000 and matches the milestone, update the ‘Status’ property to ‘Approved for Payment’ and assign it to Accounting. If it exceeds $5,000 or deviates from the terms, flag it as ‘Requires Review’ and assign it to the Project Manager.”
Of course, you can set up a Custom Agent to apply its guidelines to more than just one document! A prompt you might use could look like this:
“Review all incoming invoices that arrive in this folder. Check if the totals match our agreed-upon contract milestones found in the related Project folder. If the total is under $5,000 and matches the milestone, update the ‘Status’ property to ‘Approved for Payment’ and assign it to Accounting. If it exceeds $5,000 or deviates from the terms, flag it as ‘Requires Review’ and assign it to the Project Manager.”
2. Contextual Reading and Analysis
You can set a Custom Agent to begin work when a document enters a specific workflow stage. The agent reads the document text alongside the surrounding business metadata. Crucially, the agent respects dynamic permissions: it can only look at information it has explicit clearance to view, and it can only modify properties it has been given authorized permission to change.
3. Automated Advancement and the Audit Trail
The agent performs the required analysis, updates the approved properties on the document’s metadata card, and automatically advances the document to the next workflow phase.
To bridge the accountability gap often associated with AI, the Custom Agent writes out its entire reasoning and sources directly onto the document’s metadata card. It doesn’t just change a status; it notes why it changed that status, creating a fully transparent, explainable, and auditable trail for human compliance officers.
The Possibilities are Endless with M-Files Custom Agents
The introduction of Custom Agents means that your business can build an AI agent that’s just right for your context. Here are two examples of what a company could build with M-Files Custom Agents:
1. Smart Invoice and Financial Validation
A Custom Agent can be deployed to automatically compare incoming vendor invoices against active contract agreements stored in M-Files. It can independently validate line items, verify tax details, and route the invoice directly to the proper ledger or approver based on transaction thresholds. And because the custom agent leaves a complete record of its decisions, your audit trail will be safely documented.
2. Accelerated Contract Compliance Reviews
A Custom Agent can go beyond the pre-built Contracts Agent to scan the text of a contract for non-standard indemnification clauses or unfavorable payment terms, note the deviations on the metadata card, and route the contract back to council with localized annotations. The agent can help route a contract through its entire approval process, automatically sending it on to the appropriate reviewer at each stage.
Scaling Enterprise AI with Confidence
The true limitation of modern corporate AI hasn’t been a lack of intelligence; it has been a lack of AI governance and context. Tools that operate in a vacuum risk hallucinating, leaking sensitive data, or violating compliance regulations.
M-Files ensures that its AI agents—both specialized and custom—operate within a securely governed framework. Because every decision made by a Custom Agent is grounded in your business parameters and left with an immutable audit trail, your company can finally scale automation safely.
By turning passive, scattered repositories into active, context-aware participants in knowledge work, M-Files Custom Agents allow teams to stop spending time managing documents and start putting their content to work.
Need to Strategize? Let’s Talk!
If you’re excited about the possibilities of M-Files Custom Agents but don’t quite know where to start—or if you’re looking to implement M-Files for the first time—let’s talk! Here at Laminin, we love tailoring software to meet our clients’ unique business needs and goals. Contact us today to get the conversation started!
