---
spec_name: "Industry Intel — AI Industry Monitoring Playbook"
spec_version: "1.0"
last_updated: 2026-05-07
target_audience: "Designers, creative directors, and agency owners who need to stay current on AI industry shifts without getting buried"
expected_outcome: "A repeatable weekly news-scan workflow that converts AI headlines into client trust signals, retainer-pricing decisions, and tool-stack bets — in under 30 minutes a week"
ai_tools_supported: [claude_desktop, claude_code, google_antigravity, cursor]
based_on: https://mikekwal.com/blog/industry-intel
author: Mike Kwal
---

# Industry Intel — AI Industry Monitoring Playbook

## Instructions for the AI

This spec is **not** a code-implementation playbook. It's a **monitoring playbook** — a reusable system for a designer/agency owner to scan AI industry news weekly, filter signal from noise, and convert it into client conversations, pricing decisions, and tool-stack bets.

Your job:
1. Set up the recurring scan (Claude Desktop with WebSearch is primary).
2. Apply the **"what does this mean for my agency?"** filter to inputs.
3. Generate trust-signal lines, client emails, and durability-log entries from raw news.

Keep the user's design/agency context (not the AI lab's PR cycle) at the center of every output.

---

## Step 1: Stand up your weekly scan

**What it does.** A 20-30 min weekly ritual where Claude Desktop pulls only the AI news that matters for your agency.

**Setup.** Recurring calendar block (mine: Friday 8am). Claude Desktop with WebSearch on. One rolling note: `industry-intel-log.md`.

**Scan template — paste into Claude Desktop every Friday:**

```
You're my AI industry analyst. I'm a designer/agency owner. WebSearch the past 7 days on:

1. Frontier model releases (Anthropic, OpenAI, Google, Meta, xAI)
2. Partnership / funding announcements ($1B+)
3. Custom AI silicon (chips, inference cost movements)
4. Enterprise deployments (Big 4, security vendors, universities)
5. SaaS-tool disruption (any design/dev/marketing tool losing ground to a model)
6. Regulatory / safety shifts that change what I can promise clients

For each item:
- One-line factual summary
- "What this means for a one-person design agency" in 2 sentences
- Trust-signal line for a sales call (only if news supports one)
- Durability flag: GREEN / YELLOW / RED

Skip hype, leaderboard chatter, AI-art Twitter drama.
```

**Why it works.** Forces Claude to filter through *your* lens, not the AI lab's. You don't care that GPT-5.5 scored 0.3 higher on a benchmark — you care whether your tool stack just moved.

---

## Step 2: The "what does this mean for my agency?" filter

Most AI news is irrelevant to a 1-5 person agency. Run every headline through this 4-question filter:

1. **Does this change what I can promise clients?** (new capability unlocked, or feature I was selling just got killed)
2. **Does this change my pricing?** (cheaper compute = more AI in retainers; vendor down = refund/rebuild)
3. **Does this change my tool stack durability?** (SaaS got disrupted, or my model just got $25B in funding)
4. **Does it give me a trust-signal line for a sales call?** ("the same model running Microsoft Security…")

- **0/4 — skip.** No YouTube breakdown, no HN thread.
- **1-2/4 — log it** in the durability log (Step 5).
- **3-4/4 — act this week.** Draft the email or trust-signal line immediately while it's fresh.

---

## Step 3: Trust-signal lines from raw news

The most underused move in agency sales: **convert AI news into a one-line trust signal in a discovery call.** Pattern:

> "[High-status entity] uses [model/tool]. The site I'm shipping for you is built using the same [model/tool]."

**Generation prompt — paste into Claude Desktop with the news:**

```
News item: [PASTE HEADLINE + 2-3 SENTENCE SUMMARY]

I'm a designer running a small agency. Generate 3 trust-signal lines for a discovery call with a [INDUSTRY] prospect. Each line:
- One sentence, max 25 words
- Names a Fortune 500 / Ivy / Big 4 / Big Tech entity
- Connects that entity's choice to my work
- Sounds natural in a sales call, not a press release

Flag any line where the connection feels like a stretch.
```

**Banked examples (from my May 2026 scan):**

- "The same Claude Opus 4.7 running Microsoft Security and PwC's audit stack is reading every line of code on this build."
- "Harvard just ditched ChatGPT for Claude — that's the model I'm shipping your site with."
- "Gemini is in 4 million GM cars. The AEO structure I'm putting on your site is what makes you the answer when a driver asks for a [niche] near them."

**Storage.** Pin the best 5-10 lines at the top of `industry-intel-log.md`. Retire any line where the entity has rotated off the news cycle.

---

## Step 4: Client emails when a major change drops

When a major change hits — model release, vendor outage, SaaS disruption — retainer clients should hear from you *first*, not from a tech newsletter. Cheapest retention move in agency work.

**Email prompt — paste with the news:**

```
News: [PASTE HEADLINE + KEY DETAILS]

Draft a 150-word email for my retainer clients. Tone: calm, expert, not breathless. Structure:
1. One sentence: what changed.
2. Two sentences: what it means for THEIR site (assume [Webflow / WordPress / Shopify]).
3. One sentence: what I'm doing about it.
4. CTA: reply or wait for our monthly call.

End with my name. No subject line — I'll write that.
```

**When to send:**
- Model in their stack changes pricing/capability — same day
- SaaS in their stack drops materially — same day
- Regulatory/safety shift changes what you can do for them — within 48 hours
- Generic AI hype — do not send

**Cadence.** 1-2 emails per quarter, not per week. Scarcity is the trust move.

---

## Step 5: The tool-stack durability log

A single rolling doc — `industry-intel-log.md` — tracking every tool, model, and vendor in your stack against the news cycle. Early-warning system for vendor lock-in disasters.

**Format (one row per tool, updated weekly):**

| Tool / Vendor | Role | Last news event | Rating | Exit plan if RED |
|---|---|---|---|---|
| Claude Opus 4.7 | Primary model | $25B Amazon round, Apr 2026 | GREEN | n/a |
| Figma | Design source of truth | Stock dropped on Claude Design | YELLOW | Tokens-first design system; Figma replaceable in 18mo |
| Webflow | Client CMS | AI builder shipping, stable | GREEN | n/a |
| ChatGPT | Drafting / image gen | Lost Harvard, May 2026 | YELLOW | Claude already running in parallel |
| Netlify | Deploys + Frontend Skill | New Frontend-Design Skill | GREEN | n/a |

**Ratings:**
- **GREEN** — $1B+ runway, recent positive news, no displacement signals. Bet on it for 12+ months.
- **YELLOW** — Negative news, displacement risk visible, workflow not yet broken. Watch monthly. Don't deepen dependence.
- **RED** — Tool is being replaced by a direct LLM call OR parent company is in distress. Exit plan within 30 days.

**Cadence.** Update any row that appeared in the week's scan. Full audit quarterly.

---

## Quick-start prompts for Claude Desktop / Claude Code

### Prompt 1 — Set up the weekly scan
```
Read this spec. Help me set up the weekly scan: pick a recurring time, create industry-intel-log.md in my notes folder, run the first scan using this week's news.
```

### Prompt 2 — Filter a headline
```
Apply the 4-question agency filter (Step 2) to: [PASTE HEADLINE]. Score 0-4. Skip, log, or act?
```

### Prompt 3 — Generate trust-signal lines
```
Step 3. Generate 3 trust-signal lines for a discovery call with a [INDUSTRY] prospect, based on: [PASTE NEWS]. Flag any that feel forced.
```

### Prompt 4 — Draft a client email
```
Step 4. Draft a retainer email about: [PASTE NEWS]. Clients on [PLATFORM]. 150 words, calm.
```

### Prompt 5 — Update the durability log
```
Read my industry-intel-log.md plus this week's scan. Update affected ratings. Flag YELLOW/RED moves; draft exit plans for RED rows.
```

---

## Verification (after first 2 weekly scans)

| Check | How |
|---|---|
| Ritual is sticky | Calendar block exists, 2 consecutive Fridays completed |
| Filter is working | 50%+ of headlines get rejected — if everything scores 3+/4, you're too generous |
| Trust-signal bank exists | 3-5 lines pinned in `industry-intel-log.md`, tied to recent events |
| Emails sent appropriately | 0-1 in the first month; no retainer spam |
| Durability log current | Every stack tool has a row + rating updated in the last 30 days |

---

## Need help?

- **Free** — [Talk-to-Build community](https://www.skool.com/talktobuild). Designers running this playbook, comparing notes weekly.
- **Done-for-you** — [MK-Way](https://mk-way.com) ships AEO-ready websites and AI-native systems.
- **Quick question** — DM Mike on [Instagram](https://www.instagram.com/mikekwal). He reads every message.
- **B2B / strategy** — [LinkedIn](https://www.linkedin.com/in/mikekwal/).

---

*Companion to the Industry Intel hub at [https://mikekwal.com/blog/industry-intel](https://mikekwal.com/blog/industry-intel). The hub is my monthly read; this spec is your weekly system.*

*Last updated: 2026-05-07*