What is Propense Hatfield?

Propense Hatfield is an agentic AI 'client service partner' built specifically for accounting and legal professionals. Unlike internal productivity tools, Hatfield is designed to communicate directly with clients on the firm's behalf — delivering personalized service touchpoints, follow-ups, status updates, and responses without adding to staff workload. It launched in limited availability in February 2026, targeting firms that want to improve client responsiveness without hiring additional client service staff.

How is Hatfield different from other AI tools for professional services firms?

Most AI tools for professional services help staff work faster internally — drafting documents, summarizing research, generating reports. Hatfield is designed to work externally, toward clients. It represents a new category: AI that manages ongoing client communication at a personalized level. Rather than a staff member following up with 30 clients about outstanding documents, Hatfield handles the outreach on the firm's behalf, tracking engagement and flagging situations requiring human attention.

What types of client communication is AI appropriate for?

AI is most appropriate for routine, low-stakes client communication: appointment reminders, document request follow-ups, status updates on clearly defined deliverables, and satisfaction check-ins after routine engagements. AI is not appropriate for sensitive communications — delivering bad news, discussing fee disputes, explaining adverse outcomes, or navigating complex relationship dynamics. The question to ask before automating any client communication is: if a client responded to this message with frustration or confusion, would it be acceptable for an AI to handle the follow-up?

What is a realistic 90-day rollout for client-facing AI at a small accounting or law firm?

Month 1: Define scope. List every recurring client touchpoint in your firm and categorize each as routine (document requests, reminders, status updates) vs. sensitive (adverse news, fee discussions, relationship management). Only routine touchpoints are candidates for AI. Month 2: Run a pilot on one communication type with 10-15 clients who have long-standing relationships with your firm and would be forgiving of a rough edge. Review every AI-generated message before it sends. Month 3: Based on pilot feedback, refine the message templates and expand to a second communication type. Establish internal policy on which categories require human review and which can send automatically.

Propense Hatfield: Client-Facing AI Review

Q: Are there professional responsibility concerns with client-facing AI for law or accounting firms?

Yes. For law firms, the ABA Model Rules require that communications with clients be competent and that lawyers supervise non-lawyer assistance. AI-generated client communications must be reviewed and supervised in a way that satisfies this standard. Additionally, ABA Formal Opinion 512 requires that lawyers maintain reasonable understanding of AI capabilities and limitations and verify AI-generated output. For accounting firms, AICPA standards on client communication require accuracy and professionalism — AI-generated communications must meet the same standard as human-drafted communications. Both professions should have written policies defining which communication types require human review before sending.

Published: March 14, 2026 | By: The Crossing Report | 7 min read

Summary

In February 2026, Propense launched Hatfield — an agentic AI described as a "client service partner" for accounting and legal professionals. Unlike AI tools that help your staff work faster internally, Hatfield is designed to communicate with your clients on your firm's behalf. This is a meaningful category shift: AI moving from back-office productivity to front-of-house client relationship management. Here's what that means for a 5-25 person professional services firm, where it helps, and the guardrails you need in place before turning it on.

What Hatfield Actually Does

Most AI tools for professional services help someone inside your firm do something faster — draft a document, summarize a research memo, generate a financial model. The value stays internal.

Hatfield is built for the external layer. It delivers personalized service interactions, follow-ups, and status communications to your clients on the firm's behalf. In a 20-person accounting firm where the senior partners are billing time and the admin team is managing 200+ client relationships, the problem Hatfield addresses is real: clients don't get timely follow-up, document requests go unnoticed for days, and engagement touchpoints get missed because no one has capacity to manage them.

According to CPA Practice Advisor's February 2026 coverage of the launch, Hatfield manages ongoing client communication touchpoints without adding to staff workload and surfaces situations requiring human attention rather than handling them autonomously. It is not a fully autonomous client agent — it is a managed communication workflow that replaces the manual follow-up burden most firms handle poorly or not at all.

The Category Shift This Represents

This is worth naming clearly, because it represents something new in how AI is being positioned for professional services.

For the past two years, every AI vendor pitch to accounting and law firms has been essentially the same: "use AI to work faster internally." Faster document review. Faster tax prep. Faster research. The value proposition was efficiency — the same billable work in less time.

Hatfield's pitch is different. It is not helping your staff do existing work faster. It is replacing a category of relationship management work that your staff was either not doing at all (because there wasn't time) or doing poorly (because it required more capacity than you had).

That is both its appeal and the risk. When you deploy AI to do internal work faster, the failure mode is a bad internal document that a human catches before it goes out. When you deploy AI to communicate with clients, the failure mode is a confused or frustrated client interaction that damages a relationship before a human knows it happened.

When Client-Facing AI Is Appropriate

Not all client communication is the same. Before any professional services firm deploys a tool like Hatfield, the single most important exercise is categorizing your client communications by stakes and sensitivity.

Low stakes, appropriate for AI assistance:

Document request follow-ups ("We're still waiting on your Q4 statements — can you send by Friday?")
Appointment reminders and scheduling confirmations
Status updates on clearly defined, in-progress deliverables ("Your tax return is in final review — we expect to have it to you by [date]")
Post-engagement satisfaction check-ins for routine matters
Deadline reminders for client-side obligations

Higher stakes, require human handling:

Delivering adverse outcomes or unwelcome news
Explaining fee changes or discussing invoices
Responding to client complaints or expressions of frustration
Any communication that requires nuanced judgment about relationship dynamics
Communications where the client's response may require legal or professional advice

The test I use: if a client responded to this message with frustration or a question you didn't anticipate, would it be acceptable for an AI to handle the follow-up? If no, a human should have sent the original message too.

The Professional Responsibility Layer

For law firms, there is a specific compliance consideration that requires addressing before deploying any client-facing AI.

The ABA Model Rules require that client communications be competent and that lawyers supervise non-lawyer assistance. Under ABA Formal Opinion 512 (now governing in all ABA-member states), lawyers must maintain "reasonable understanding" of AI capabilities and limitations and verify AI-generated output. A law firm deploying Hatfield — or any AI client communication tool — must have a written policy defining which communication types require attorney review before sending and how that review is documented.

This is not an argument against using Hatfield. It is an argument for deploying it with a review protocol rather than treating it as a set-and-forget automation.

For accounting firms, the standard is analogous: AICPA communication standards require accuracy and professionalism. AI-generated client communications must meet the same standard as human-drafted communications and should be reviewed with the same care.

The practical implication: the responsible deployment of a tool like Hatfield starts with a policy that answers two questions. Which communication categories require human review before sending? And who is responsible for that review?

A Realistic 90-Day Rollout

For a 10-20 person accounting or law firm evaluating Hatfield or a comparable client communication tool, here is a realistic pilot structure:

Month 1 — Define scope. List every recurring client touchpoint in your firm: document requests, deadline reminders, status updates, post-engagement check-ins, scheduling. Categorize each as routine or sensitive. Build a short policy document (one page is sufficient) that defines which categories require human review before sending and which can be automated.

Month 2 — Pilot with established clients. Run the pilot on one communication type — document request follow-ups are usually the best starting point — with 10 to 15 long-standing clients who have strong relationships with your firm and would be forgiving of an occasional rough edge. Review every AI-generated message before it sends in the first month. Collect feedback from clients informally ("Did you find our follow-ups timely and clear?").

Month 3 — Expand based on evidence. If the pilot communications landed well, expand to a second communication type. Adjust the message templates based on what you learned. Update your internal policy to reflect which categories have been validated for lower-oversight automation. Document the process so that when a new staff member joins, the policy is clear.

The Honest Bottom Line

Client-facing AI is genuinely useful for a specific, bounded set of communication tasks that most professional services firms handle poorly or not at all due to capacity constraints. The administrative overhead of managing 150 client follow-up conversations while doing billable work is a real problem, and a tool that removes that burden has legitimate value.

The risk is not that the AI is incompetent. The risk is deploying it in contexts where AI is not appropriate — where a human relationship, a human judgment call, or a professional responsibility consideration requires a human in the loop. That distinction requires a policy, not a product feature.

Propense Hatfield is worth evaluating for any firm that has identified client communication follow-through as a gap. Evaluate it with your review protocol designed before you turn it on, not after.

An AI Just Launched That Talks to Your Clients for You. Is That Actually a Good Idea?