How to Draft Market Commentary with AI (Without Sounding Like a Vendor)

Most market commentary advisors send sounds like vendor marketing copy. There's a reason — most of it is repackaged from market-data vendors with minor tweaks. AI lets you actually write it yourself in a fraction of the time, in your own voice, with your own takes.

Here is the operator process we deploy at advisor firms.

Setup: the inputs you need

To produce good commentary, you need four data sources ready:

• Weekly market data — Major indices, sector performance, key macro indicators (CPI, unemployment, Fed positioning). Pull from Bloomberg, YCharts, FactSet, or even free sources like Yahoo Finance.
• A point of view — What you think clients are worrying about this week. This comes from your client meetings, not the data.
• Your firm voice template — A markdown document with examples of your prior commentary, key phrases you use, key phrases you avoid.
• Compliance guardrails — A list of words/phrases your compliance team has flagged before (e.g., "guaranteed," "always," "best-in-class," forward-looking language without disclosure).

The workflow

Step 1 — Define the week's lens (5 minutes)

What is THE topic this week? Pick one. Examples:

• The Fed signaling potential cuts and what it means for fixed-income allocations
• Tech earnings missing on AI capex and the read-through
• A specific geopolitical event your clients are asking about

Pick one. Don't try to cover everything. Generic round-ups are exactly what AI produces by default — your job is to give it focus.

Step 2 — Build the prompt scaffold

In Claude or ChatGPT Team (enterprise plan only — never consumer):

`` Draft a market commentary for our advisory firm's clients.

This week's topic: [YOUR TOPIC] This week's data points: [3-5 specific numbers, e.g., "S&P down 1.8%, 10-year yield to 4.2%, CPI at 2.4%"] Audience: high-net-worth households, ages 55-75, primarily focused on retirement income and estate planning

Voice and style:

• Calm, operator-led
• Specific numbers, not vague claims
• Acknowledge uncertainty without being alarmist
• 400-500 words
• One section: market recap
• One section: what we're watching
• One section: what it means for your portfolio

Compliance constraints:

• No forward-looking statements without disclaimer
• No "guaranteed" or "always"
• Reference specific data, not generalizations
• Include "Past performance is not a guarantee of future results" at the close

Examples of our prior commentary: [paste 2-3 paragraphs of your prior writing] ``

The example paragraphs are the critical piece. They steer voice better than any instruction.

Step 3 — Generate first draft (2 minutes)

Run the prompt. Read the output. It will be 70-80% there if the inputs were good.

Step 4 — Edit for voice (10-15 minutes)

Three edits to make every time:

• Cut any sentence that sounds like a vendor wrote it
• Add one specific client conversation you had this week ("Several of you asked about...") — this is what makes it feel like real advisor voice
• Tighten the close — most AI drafts run 50-100 words too long

Step 5 — Compliance pass (5 minutes)

Run through your firm's flagged-phrase list. Check disclaimers. Check forward-looking language is properly hedged.

Step 6 — Send

Through your normal email or CRM channels.

Total time

A weekly commentary that used to take 90-120 minutes drops to 25-35 minutes once the workflow is built. For a firm with 3+ advisors all sending commentary, that's 6-9 hours/week recovered.

The mistakes to avoid

Mistake 1: Generic prompts. "Draft a market commentary for this week." This produces vendor copy. Always anchor with a specific topic and specific data points.

Mistake 2: No voice examples. Without examples, the output sounds like every other AI commentary. Always paste 2-3 paragraphs of your actual prior writing.

Mistake 3: Trusting the data. AI can hallucinate market data. Always verify numbers from a primary source before sending.

Mistake 4: Sending without editing. Even a great first draft needs the human voice layer. Skip the edit step and clients can tell.

Mistake 5: Same commentary every week. AI drafts toward statistical averages. Without you steering the topic, every week's commentary looks the same. Steer the topic.

A worked example

Topic: Fed signaling potential cuts. Audience: pre-retirement clients worried about income.

Bad prompt: "Write market commentary about the Fed."

Good prompt: Include this week's actual rate signal, the 10-year yield move, the question you've heard from clients ("does this mean I should lock in fixed income now?"), and the specific portfolio framing you'd offer ("for clients in the distribution phase, we're maintaining duration in the 4-6 year range because...").

Output from a good prompt: ~450 words of usable commentary that sounds like you wrote it.

Compliance considerations

Market commentary is supervised communication under SEC and FINRA rules. AI-drafted commentary is no different — you still own the compliance review:

• Supervisor reviews and approves
• Documented in the marketing review system
• Filed under the firm's records retention policy
• Disclaimers attached per firm template

AI doesn't change the review process. It changes the draft generation step.

Scaling commentary to multiple advisors

For firms with multiple advisors who each send their own commentary, the workflow gets layered:

• One operations lead drafts the firm-wide weekly using the workflow above
• Each advisor receives the firm-wide draft + a prompt to personalize it for their specific client segment
• Advisor personalizes in 5-10 minutes and sends from their account

This compresses 5 advisors × 90 minutes/week (7.5 hours) down to 1 hour firm-wide + 5 × 10 minutes (~1.8 hours total). Material time recovery.

Bottom line

AI-drafted market commentary works if you steer the topic, paste your voice, verify the data, and edit for the human layer. The firms doing this well sound like themselves. The firms doing it badly sound like every other AI commentary in the inbox. The difference is operator discipline, not technology.