A few years ago, a new prospect filled out a website form, the form emailed a paralegal, the paralegal added the person to a spreadsheet, and, if things went well, someone ran a conflicts search the next morning and pinged an attorney by lunch. Today, in firms that have wired agentic AI into the same pipeline, the form submission triggers software that reads the answers, extracts party names, checks them against the firm’s client and adverse-party records, opens a preliminary matter file, and books the intake call before anyone on staff has read the email.
That change matters because intake is where firms lose money without noticing and take on risk without pricing it. A slow response loses the client; a sloppy conflicts check loses the firm. Software that can act, not just answer, moves the needle on both. The question every managing partner is now asking is a practical one: what does this do inside the workflow, and where do the guardrails go?
What Makes an Agentic System Different From the AI Firms Already Use?
Most lawyers have used generative AI in a chat window. You type a question, it answers, you copy the useful part. An agentic system works differently.
It’s given a goal and a set of tools (the firm’s CRM, its document management system, the conflicts database, a calendar, an email account) and it decides which tool to reach for and in what order. Bloomberg Law describes agentic systems as software that can autonomously execute multistep workflows rather than sitting inside a single chat interface. The practical difference is that the software takes action. It doesn’t draft a conflicts memo for a human to run; it runs the check, records the result, and escalates the exceptions.
Where Does an Agent Actually Plug Into Intake?
The clearest wins tend to sit at the seams between systems, the places where information used to sit in an inbox until someone had time to move it. A custom agent built for a firm’s specific process usually earns its keep in four places:
- First-touch response. The agent acknowledges the inquiry, asks the follow-up questions a human intake coordinator would ask, and captures the answers as structured data. Every hour a lead sits untouched, the odds of signing them drop.
- Preliminary conflicts screen. Party names, spouses, related entities, and opposing counsel get normalized and run against the firm’s records. Clean hits close automatically; anything ambiguous is routed to a human with the evidence attached.
- Matter-open packet. Once conflicts clear, the agent creates the matter in the practice management system, pulls the right engagement letter template, populates it from the intake data, and sets up the client folder.
- Handoff to the attorney. The lawyer walks into the consultation with a short brief: what the prospect said, what the conflicts search found, and what’s still missing.
Does This Cross the Line on Conflicts and Supervision?
Conflicts screening is not a place where firms get to be casual. The professional conduct rules impose an affirmative duty on the firm to make reasonable efforts to identify and address conflicts, and “the software did it” is not a defense when a screen misses a related party. The right framing for an agent inside conflicts is preliminary work, not final clearance.
The current ethics guidance on generative AI leans in the same direction. A recent analysis of ABA Formal Opinion 512 walks through the duties that carry over when a lawyer delegates work to an AI system, including competence, confidentiality, supervision, and honest client communication.
The opinion doesn’t ban delegation. It insists the lawyer own the output. Applied to intake, a human signs off on the conflicts result before an engagement letter goes out, and someone in the firm actually understands what the agent looked at and what it skipped.
What Do You Have to Build Yourself, and What Comes Off the Shelf?
Off-the-shelf legal tech vendors are shipping agentic features fast. For a lot of firms, those features are enough for the generic parts of intake, like the form, the acknowledgment, and the calendar booking. The trouble starts where the firm’s process stops being generic: a specialty practice with an unusual conflicts rubric, an intake questionnaire that changes by matter type, or integrations with a legacy billing system.
That’s where custom development enters the picture. A firm can commission a purpose-built agent that talks to its actual systems, follows its actual intake logic, and enforces its actual supervision points, the kind of work engineering shops like DEV.co handle for law firms building on top of their existing case management and document stacks. The build isn’t glamorous. It’s connectors, permissioning, audit logs, and a clear map of which decisions the agent can close and which it has to escalate.
Drawing the Guardrails Before the First Prospect Hits the Form
Intake used to be the part of the firm nobody wanted to modernize because touching it meant touching every other system. Agentic software makes the modernization possible without a rip-and-replace.
Done well, the client gets an answer in minutes, the conflicts search actually gets run the same day it should, and the lawyer walks into the first call already knowing what the matter is about. Done poorly, the firm has automated a compliance problem. The difference is largely in how the guardrails were drawn before the first prospect ever hit the form.
Our dedicated team gathers information from all the reliable sources to make the law accessible and understandable for everyone. We provide the latest legal news stories from across the country, delivered straight to you.