AI Readiness
The staff accountant used an AI tool to draft a client email summarizing the quarterly financial review. The draft was professional, well-organized, and accurate in its data presentation. It also recommended that the client "consider accelerating depreciation on recent equipment purchases for tax optimization." The recommendation was reasonable in general — but inappropriate for this specific client, who was in the middle of a loan application where accelerated depreciation would reduce the net income the lender was evaluating. The AI did not know about the loan application. The staff accountant who reviewed the draft did know — and caught the recommendation before sending. Context that AI cannot see is what makes human oversight non-negotiable for client communications.
AI drafts client communications efficiently but cannot account for relationship context, client-specific circumstances, strategic timing, or the professional judgment that determines what should and should not be communicated. Human oversight before any AI-drafted communication reaches a client is non-negotiable — because the communication represents the firm's professional voice, carries implicit advice, and shapes the client relationship in ways AI cannot evaluate.
Why AI-drafted client communications require human review and what the review should evaluate before sending.
Client relationship managers, team leads, and anyone using AI to draft client-facing communications.
A single poorly calibrated communication can damage a relationship that took years to build. AI efficiency does not offset relationship damage.
AI generates professional text but cannot calibrate tone to relationship dynamics. A long-standing client who communicates informally receives a stiff, formal email. A new client who expects professional distance receives an overly casual message. A client going through a difficult business period receives a cheerful email about optimization opportunities. Tone carries relationship information that AI cannot read.
AI-drafted communications may include recommendations, suggestions, or observations that constitute professional advice. If the advice is inaccurate — based on outdated rules, wrong assumptions, or incomplete information — the firm has delivered incorrect guidance to a client who may act on it. The liability exposure extends to any advice in any communication, not just formal deliverables.
AI tools may inadvertently include information from one client's context in another client's communication. If the AI tool has been used across multiple client engagements, there is a risk of cross-contamination where details, figures, or recommendations from one client appear in another's communication. This is a data flow problem with immediate relationship and legal consequences.
Clients can detect when communications are formulaic rather than tailored. AI-generated communications often have a recognizable quality: perfectly structured, appropriately detailed, and subtly generic. When every client receives communications that read similarly, the firm signals that it is processing clients rather than serving them. The efficiency gain creates a relationship cost.
Routine status updates. "Your quarterly financial statements are ready for review" emails based on structured data. AI drafts the update, human verifies accuracy and adds client-specific notes before sending.
Meeting preparation. AI generates agenda drafts from engagement notes, prior meeting summaries, and open items. The professional reviews, adjusts for current priorities, and adds context-specific items.
First drafts of standard documents. Engagement letter drafts from standard terms, proposal templates from client information, and report cover letters from deliverable summaries. Each requires human review for accuracy and appropriateness before sending.
Information summarization. Condensing complex analysis into client-friendly language. AI excels at translating technical accounting into accessible prose. The professional verifies that the simplification does not sacrifice accuracy or create misleading impressions.
Accuracy check. Does the communication contain any factual statements, numbers, or advice? If yes, verify every factual element against source data. An email that says "your revenue increased 12% year-over-year" must be verified against actual financial data. Inaccurate numbers in a client email erode confidence immediately.
Tone check. Does the tone match the client relationship? Is it appropriately formal or informal? Does it account for the client's current situation? A tone check takes seconds for someone who knows the client and is impossible for AI that does not.
Context check. Does the communication account for what is happening in the client's business that the AI may not know? Open transactions, pending decisions, ongoing negotiations, sensitive situations — context that affects what should be communicated and how.
Confidentiality check. Does the communication include any information that should not be shared? Could any content have been influenced by another client's data? This check is critical when AI tools are used across multiple client engagements.
Standards check. Does the communication meet the firm's quality and branding standards? Is the format consistent? Are required disclaimers included? Does it represent the firm appropriately?
AI transcription and summarization tools produce meeting summaries efficiently. These summaries are increasingly used as records that clients rely on. Three review requirements apply:
Accuracy of substance. Did the summary correctly capture what was discussed? AI transcription can misattribute statements, misinterpret technical terminology, and miss nuance. The professional who attended the meeting must verify the substance.
Accuracy of commitments. Did the summary correctly capture action items and commitments? An AI-generated action item that the firm did not actually commit to creates an obligation the firm may not intend. Every action item and commitment must be verified by the professional who made (or did not make) the commitment.
Appropriateness of inclusion. Should everything discussed be included in the written summary? Some meeting discussions are preliminary, off-the-record, or sensitive. The professional determines what belongs in the formal summary and what does not.
They establish review requirements by communication type. Routine status updates: self-review sufficient. Communications containing advice: manager review required. Communications to new or sensitive clients: partner review required. The review level matches the risk level of the communication type.
They maintain client-specific communication notes. Tone preferences, communication frequency expectations, topics to avoid, preferred formats — these notes guide AI drafting and human review. They ensure that AI-assisted communications reflect client knowledge rather than generic professional standards.
They use AI for internal communications freely. The oversight requirement applies to external, client-facing communications. Internal communications — team updates, workflow notifications, project status — can use AI assistance with lighter oversight because the audience is colleagues, not clients.
They treat communication quality as a service quality indicator. The quality of client communications reflects the quality of client service. Strong firms measure communication quality alongside deliverable quality, recognizing that clients evaluate the firm through every interaction, not just through formal deliverables.
Client communications are the daily expression of the client relationship. They are more frequent than deliverables, more personal than reports, and more immediate than engagements. When AI assists with communications, it touches the most relationship-sensitive element of the firm's service.
Human oversight of AI-drafted communications is not a quality control measure — it is a relationship protection measure. The review takes minutes. The relationship took years to build. The calculus is straightforward.
Firms working with Mayank Wadhera through DigiComply Solutions Private Limited or, where relevant, CA4CPA Global LLC, develop AI communication frameworks that capture efficiency while protecting the client relationships that define the firm's value.
AI drafts efficiently but cannot read relationship context. Human oversight ensures every communication reflects the firm's professional judgment and client knowledge.
Sending AI-drafted communications after only checking format and grammar. Tone, context, and accuracy require professional review that AI format-checking cannot provide.
They calibrate review requirements to communication type, maintain client-specific notes, and treat communication quality as a service quality indicator.
The review takes minutes. The relationship took years. Human oversight of AI communications is relationship protection, not bureaucracy.
AI cannot account for relationship context, client circumstances, or strategic timing. Communications carry the firm's voice and implicit advice.
Routine status updates, meeting preparation, first drafts of standard documents, and information summarization — all with human review before sending.
Tone misalignment, inaccurate advice, information disclosure errors from cross-client contamination, and generic signals that damage relationship perception.
Review framework: verify accuracy, tone, context, confidentiality, and firm standards. Calibrate review level to communication risk.
Yes, with review. Verify substance accuracy, commitment accuracy, and appropriateness of inclusion before distributing to clients.
Transparency builds trust. Be prepared to discuss AI usage. Disclose through engagement letters for substantive deliverables.
Communication oversight is a specific application of the principle that all AI output entering external deliverables requires human review.
Concise insights on workflow design, AI readiness, and firm economics. No fluff. Unsubscribe anytime.
Not ready to engage? Take a free self-assessment or download a guide instead.