Chapter 27 Key Takeaways: Business Communication: Email, Reports, and Documents

Core Principles

AI is most valuable for the emails you dread writing. Simple, routine emails don't need AI. The emails that require careful thought about tone, framing, and consequences — difficult news, sensitive follow-ups, escalations — are where AI assistance delivers the most value.
AI builds the architecture; you furnish it. AI can generate strong structures for emails, reports, and proposals. The specific details, relationship context, and personal voice that make business communication effective must come from you.
AI writing assistance shifts work from drafting to editing. With AI, less time goes into first drafts and more time goes into review, correction, and personalization. This is a skill shift — effective AI writing assistance requires good editing judgment.
Read every AI draft aloud before sending. The sentences you'd stumble over reading to a colleague are candidates for rewriting. Your natural speech register is the quality standard AI-generated business writing should be measured against.

The "difficult email" prompt pattern is where AI delivers disproportionate value. Emails that require careful tone calibration — declining, delivering bad news, escalating, acknowledging mistakes — are exactly where AI helps most by providing a professional starting structure that you then personalize.
Don't start emails with "I hope this email finds you well." Or any variant. Get to the point in the first sentence. This applies to AI-generated drafts (which will include these openers by default) and to your own writing.
Tone calibration prompting produces two versions for comparison. Generating both a more direct and a more diplomatic version of the same email allows deliberate tone selection rather than binary intuition. Different relationships require different registers.
Follow-up emails should never begin with "Just following up." This construction minimizes the sender's urgency and treats their request as an inconvenience. Effective follow-ups restate the ask clearly, specify what's needed and by when, and acknowledge the likely reason for delay without assuming bad intent.
Your personal email template library — built in your voice — is a compounding productivity asset. Templates built once and personalized to your communication style pay dividends every time you use them. Templates built by AI and not fully edited are someone else's voice.

The "brief me" technique is the most underused AI productivity technique in business writing. Pasting rough notes into a structured document prompt converts messy thinking into professional communication in minutes. The technique works because you provide the knowledge; AI provides the structure.
The standalone test is the right measure for executive summaries. A busy executive who reads only the summary and can state the main finding, key recommendation, and any required decisions has received what they need. A summary that fails this test is not yet an executive summary — it's a chapter list.
Progress reports should lead with status and decisions, not activities. "What completed" is less important to most stakeholders than "what's at risk, what you need from me, and whether we're on track." Restructure progress reports around decisions and actions, not accomplishments.
Lessons learned documents must be actionable. Every lesson should end with "therefore, in the future..." If a lesson doesn't have a behavior change implication, it's an observation, not a lesson.

Lead with the client's problem, not your credentials. This is the most common AI-generated proposal failure and the most common human proposal failure. Demonstrating that you understand the problem before presenting your solution signals that you've listened — which is often the most differentiated thing a proposal can show.
The competitive positioning section must be specific to what this client values. Generic positioning statements ("we bring deep expertise and a client-focused approach") communicate nothing. Effective competitive positioning identifies what this specific client cares most about and explains why you win on those dimensions.
Anchor price to value before stating the number. The business case for the investment — what it costs, what the current situation costs, what return looks like — should precede the price point. Confidence in pricing is conveyed by specificity (itemized, not round numbers) and by framing the investment relative to the value it delivers.
Each section of a proposal needs something AI couldn't have written. A proposal that reads like a template signals that you're treating the client like a template. At minimum, each major section should contain one reference specific enough to this client that it couldn't have come from a prompt — something from the discovery conversation, a specific concern they raised, a piece of their language.

Job descriptions should open with what makes the role compelling, not with job title and reporting structure. The opening of a job description is an advertisement. Most job descriptions are organized as administrative documents. AI will produce the administrative structure; you need to add the compelling reason to apply.
Meeting minutes should capture decisions and action items above all else. The comprehensive discussion summary is rarely consulted. The decisions and action items are checked repeatedly. Minutes organized around decisions > discussions makes them more useful.
Policies and SOPs should specify what happens when they're not followed. Documents that only describe the process and not the consequence for deviation are wishful thinking, not governance.

The same facts need different packaging for different audiences. Technical status updates that serve engineers don't serve executives. Not because executives are less capable, but because they need different information — business impact, decisions required, and plain-language risks — not technical root causes.
"What does this mean for our clients?" is the essential translation question. For every technical item, this question converts internal-facing content into stakeholder-relevant content.
Translation workflows improve the source documents over time. When engineers know their technical reports will be translated for business audiences, they become more disciplined about including business context in the original — creating a virtuous feedback loop.

Anonymize before pasting. Client names, financial figures, and identifying details should be replaced with generic placeholders before using external AI tools. The effort is small; the protection is significant.
Personnel information should not be processed through external AI tools. Performance reviews, compensation discussions, disciplinary matters, and HR-related communications involve legal and ethical protections that make external AI processing inappropriate.
Know your organization's AI usage policy. Enterprise AI tools (Microsoft Copilot in Microsoft 365, Google Workspace Gemini) typically operate under data handling agreements that provide different protections than consumer AI tools. Know which tools your organization has contracted and what their terms cover.
The person most likely to have a confidentiality incident is someone trying to do good work faster, not someone being careless. Build privacy habits proactively — before you need them — rather than reactively after an incident.