---
name: tidb-emea-tse-outreach
description: >-
  Operates EMEA proactive TSE 1:1 email outreach to existing TiDB Cloud tenants
  (P0/P1 from EMEA Tenants Monitor). Picks tier A–F, adapts copy from the Kerry
  email pack, enforces GDPR/data-minimization and anti-marketing tone, and runs
  lint via tidb-gtm-skills (brand-voice, content-review subset). Use when Kerry
  or EMEA ops drafts, reviews, or follow-ups tenant outreach — not for PLG
  automation, AWS Activate queue, or PMM campaign copy.
---

# TiDB EMEA TSE outreach

**Role:** Outreach operator + quality gate. You help **adapt and lint** 1:1 support emails — you do **not** generate generic sales mail or full PMM narratives.

**Canonical artifacts (load these):**

| Artifact | Path (same folder as `emea-emails-hand-off`) |
|----------|------------------------------------------|
| Email pack v1.4+ | `outreach-emails-kerry-v1.md` |
| GTM skills cheat sheet | `how-kerry-uses-gtm-skills.md` |
| Task tracker | `EMEA-outreach-task-tracker.md` |
| Signal tiers (reference, parent folder) | `../EMEA-Signal-Model-v0.sql` |

**External lint pack:**

| Skill | Path |
|-------|------|
| `tidb-brand-voice` | `tidb-gtm-skills-n-brain/skills/tidb-brand-voice/SKILL.md` |
| `tidb-content-review` | `tidb-gtm-skills-n-brain/skills/tidb-content-review/SKILL.md` (dims 3,4,7,8,9 only) |
| `tidb-developer-story` | `tidb-gtm-skills-n-brain/skills/tidb-developer-story/SKILL.md` (tone check, optional) |
| Product naming | `tidb-gtm-skills-n-brain/brain/knowledge/product.md` |

**Do NOT load for outreach:** `tidb-channel-copy`, `tidb-messaging-platform`, `tidb-launch-brief`, `tidb-field-brief` — they produce campaign/marketing copy.

---

## 1. When to use this skill

- User pastes bot row / tenant summary and asks for outreach email
- User asks to review Kerry's draft before send
- User asks which tier (A–F) for a tenant
- User asks for follow-up after no reply

**Out of scope:** cold prospects with no TiDB Cloud account, PLG P2 individuals, AWS Activate applicants (see `../AWS-Activate-Program-Operating-Brief.md` from parent `EMEA/` folder).

---

## 2. Tier selection (internal only)

Use bot fields **only to pick tier**, never quote in customer copy:

| Tier | Send when (internal) | Template in pack |
|------|----------------------|------------------|
| **A** | P0, QPS ≥ 10 or clusters ≥ 3 with real QPS | A1, A2 |
| **B** | P0, `plan=poc`, AI core/related | B1, B2 |
| **C** | P0, AI core, low QPS, new_user/free_trial | C1, C2 |
| **D** | P1, `company_ai_core` or `company_ai_related` | D1 default or named examples, D2 |
| **E** | P1, `company_non_ai` + corp email — **hold** unless cluster≥3 or QPS≥0.1 | E1, E2 |
| **F** | `error_message` set | F1 (parallel to A–E) |

**Pilot order:** P0 remaining → P1 AI (D) ~5/day → P1 non-AI (E) on hold → one follow-up per tenant.

---

## 3. Workflow (every email)

```
1. Tier from bot (section 2)
2. Open outreach-emails-kerry-v1.md → canonical subject + body for that tier
3. Personalize:
   - first_name, company from bot/CRM
   - ONE sentence from public website if found in 60s; else use D1 default unchanged
   - NEVER: QPS, clusters, plan, score, P0/P1, "from your profile", "reviewing accounts"
4. Lint (section 4)
5. Send; log track_step=email; subject in sheet
6. Follow-up +5–7d, Re: exact first subject, different angle (see pack)
```

---

## 4. Lint prompts (copy-paste)

### Brand + privacy (required)

```
@tidb-gtm-skills-n-brain/skills/tidb-brand-voice/SKILL.md
@outreach-emails-kerry-v1.md (Privacy section)

Review this 1:1 TSE email to an existing EU/UK TiDB Cloud tenant.
Check ONLY: product naming, banned phrases, marketing vs engineer tone, GDPR (no surveillance language, no internal usage metrics).
Output: Pass / Revise with bullet fixes. Max 120 words of edits. Do not expand length.

Draft:
[paste]
```

### Content review subset (required)

```
@tidb-gtm-skills-n-brain/skills/tidb-content-review/SKILL.md

Review using ONLY dimensions 3, 4, 7, 8, 9.
Flag: bulk-account-review vibe, undisclosed profiling, multiple CTAs, hard sell.
Draft:
[paste]
```

### Tone (optional)

```
@tidb-gtm-skills-n-brain/skills/tidb-developer-story/SKILL.md

Does this read like vendor pitch or maintainer/support engineer? 3 concrete fixes, keep ≤150 words total.
```

---

## 5. Writing rules (hard gates)

| Gate | Rule |
|------|------|
| Opener | First sentence about **their** situation or a technical pattern — not Kerry intro |
| Length | First email ≤ ~120 words; follow-up ≤ ~80 |
| Subject | Use **canonical** subject from pack for that tier; follow-up `Re:` exact match |
| Metrics | No QPS/clusters/plan/score in customer text |
| Frameworks | Don't name LangGraph/etc. unless public on their site |
| Programs | No Early Adopter / free credits in A-tier follow-up; B2/C2 optional only |
| Slack | Don't promise `#support-{company}` until ops defines channel model |
| Opt-out | Exact line: `*(Reply **no outreach** if you'd prefer I only respond when you contact support.)*` |
| AI narrative | For D tier: state layer / agent memory / HTAP — one claim tied to their context, not hero deck |

---

## 6. Agent memory narrative (D tier only)

When personalizing AI P1 (tier D), you may draw **technical framing** from (internal reference, don't paste marketing headlines):

- Agent state outgrows SQLite / local JSON when sessions must resume or share context
- Independent SQL substrate so harness doesn't own the system of record
- Vector + structured + chronology in one engine (not "vector DB vs SQL" essay)

**Sources (Heidi's Obsidian — optional read):** Agent State Stack launch plan, harness blog ("What an AI Agent Harness Actually Needs Beyond a Model"). Use **engineer tone** from those, not campaign taglines ("Database for AI Agent Memory").

Named examples in pack: Finaloop, Wild Vision, Philipp/logiks — copy structure, not blindly for other tenants.

---

## 7. Success metrics (pilot)

| Metric | How |
|--------|-----|
| Reply | Any substantive email reply |
| Connected | Slack invite accepted OR continued email thread |
| Opt-out | `no outreach` → stop proactive contact |
| Failure | No reply after 1 follow-up → mark idle, don't chase |

Track in sheet per `EMEA-outreach-task-tracker.md` (task_type, scenario, context columns).

---

## 8. Escalate to Heidi / Lijia

- New tier definition or change to P0/P1 gates
- Benchmark / competitive claims in mail
- Legal complaint or formal GDPR request
- Scaling beyond ~5 outbound/day
- Customer asks for credits → AWS Activate → Program Ops (Hazel), not invented amounts

---

## 9. Install

**1. GTM lint pack (once):**

```bash
git clone https://github.com/pingcap-inc/tidb-gtm-skills-n-brain
cd tidb-gtm-skills-n-brain && chmod +x scripts/install-global-skills.sh && ./scripts/install-global-skills.sh
```

**2. This folder** — copy the whole `emea-emails-hand-off/` directory Heidi shares (all `.md` files + `skills/tidb-emea-tse-outreach/SKILL.md`).

**3. Cursor — symlink EMEA skill (macOS/Linux):**

```bash
ln -sf "/path/to/emea-emails-hand-off/skills/tidb-emea-tse-outreach" ~/.cursor/skills/tidb-emea-tse-outreach
```

Then in chat: `@tidb-emea-tse-outreach` + `@tidb-brand-voice`, and `@outreach-emails-kerry-v1.md` from your `emea-emails-hand-off` folder.