# Thunderbird + ThunderAI + Regolo: AI email summaries & replies with an OpenAI‑compatible endpoint

👉[Try ThunderAI on Regolo for free](https://dashboard.regolo.ai)

Email workflows waste time because **“AI-in-email”** setups usually break on one of these: vendor lock‑in, fragile integrations, unclear data flows for GDPR reviews, and hidden operational costs when scaling beyond a single mailbox.
If your team handles customer tickets, legal threads, sales negotiations, or HR requests, even a simple “summarize + draft reply” feature becomes risky when you can’t control where data goes, how it’s logged, and how quickly you can swap models.

**Run ThunderAI with [Regolo ](https://regolo.ai/)in less than 10 minutes. Ready to scale, OpenAI‑compatible, and built for privacy‑first deployments.**

## **Outcome**

- 3-field setup in **ThunderAI** (Host, Model, Key) to get AI actions directly inside **Thunderbird** without custom code.
- 2 daily use cases in one click: summarize long threads and draft replies while staying in the email client.
- 1 OpenAI‑compatible backend that keeps your integration portable across models (start with Llama‑3.1‑8B‑Instruct, change later without rewriting flows).

## **Prerequisites**

- [Regolo API Key](https://dashboard.regolo.ai/) (store it as **REGOLO\_API\_KEY** or paste it into ThunderAI once).
- Thunderbird installed.
- ThunderAI add-on installed.
- Model name ready (example used here: [Llama-3.1-8B-Instruct](https://regolo.ai/models-library/)).

## Step-by-step Guide

### 1) **Create your Regolo API Key**

*Generate a dedicated key for email usage (recommended: least-privilege, separate from other apps).*

```
REGOLO_API_KEY=your_key_hereCode language: Python (python)
```

**Expected output:** you have a key to paste into ThunderAI.

### 2) Install ThunderAI in Thunderbird

*Install the ThunderAI add-on and open its settings.*

**Expected output:** ThunderAI options page is available inside Thunderbird.

### 3) Configure ThunderAI to use Regolo (OpenAI‑compatible)

*In ThunderAI settings, set the OpenAI-comptible provider fields.*

```
Host Address: https://api.regolo.ai/
Compatible OpenAI models: Llama-3.1-8B-Instruct
OpenAI Compatible key: <your Regolo API key>Code language: JavaScript (javascript)
```

**Expected output:** ThunderAI can call Regolo as its backend.

### 4) Test a default prompt (summary)

![](https://regolo.ai/wp-content/uploads/2025/07/image.png)*Open any email thread, click the AI button, and run “Summarize”.*
**Expected output:** a compact summary of the email content appears in the ThunderAI output panel.

### 5) Enable high-signal custom prompts with placeholders

*Create a custom prompt that uses placeholders so only the needed parts are sent (data minimization by design).*

Example prompt text:

```
Summarize this email in 5 bullet points.
Highlight action items and deadlines.
Email subject: {%mail_subject%}
Email body: {%mail_text_body%}Code language: JavaScript (javascript)
```

**Expected output:** repeatable summaries across the team, with consistent formatting and less prompt drift.

### 6) Draft a reply that stays “human-safe”

*Use a reply prompt that asks for a structured draft + a “things to verify” checklist before sending.*
Example prompt text:

```
Write a professional reply.
Keep it concise, friendly, and GDPR-aware.
Ask clarifying questions if needed.
Return:
1) Draft reply
2) Checklist (facts to verify before sending)
Context: {%mail_text_body%}Code language: PHP (php)
```

**Expected output:** a usable draft plus a safety checklist to reduce accidental misinformation or over-sharing.

## **Step‑by‑step with working code**

Goal here is to replicate the same behavior outside Thunderbird (for QA, audits, or future migrations) using an **OpenAI-compatible** call pattern.
**Working example (server-side summary endpoint):**

```
import os, requests

REGOLO_API_KEY = os.getenv("REGOLO_API_KEY")
BASE_URL = "https://api.regolo.ai/v1"
MODEL = "Llama-3.1-8B-Instruct"

def summarize_email(subject: str, body: str) -> str:
    prompt = (
        "Summarize this email in 5 bullet points.\n"
        "Highlight action items and deadlines.\n\n"
        f"Subject: {subject}\n\n"
        f"Body:\n{body}\n"
    )

    r = requests.post(
        f"{BASE_URL}/chat/completions",
        headers={"Authorization": f"Bearer {REGOLO_API_KEY}"},
        json={
            "model": MODEL,
            "messages": [
                {"role": "system", "content": "You are a helpful email assistant for a European company."},
                {"role": "user", "content": prompt},
            ],
            "temperature": 0.2
        },
        timeout=30
    )
    r.raise_for_status()
    return r.json()["choices"][0]["message"]["content"]

# Example usage:
# print(summarize_email("Contract amendment", long_email_text))Code language: Python (python)
```

**Expected output:** deterministic-ish summaries (low temperature), reproducible for testing, and easy to wrap with logging, redaction, or retention controls.

## **Benchmarks &amp; costs**

Email workloads vary widely (thread size, language, attachments), so the most useful “benchmark” is a pilot in your own mailbox set with p50/p95 latency and cost per 1,000 emails.

| **Option** | **Typical deployment footprint** | **Latency (p50 target)** | **Pricing unit** | **Data residency posture** |
|---|---|---|---|---|
| Regolo (OpenAI‑compatible) | Single endpoint, swappable models | Track in pilot | Per-token or per-request depending on model | EU-focused integration patterns (recommended for European orgs) |
| US-based OpenAI-style API | External provider endpoint | Track in pilot | Usually per-token | Often US processing / cross-border considerations |

👉[Try Implement ThunderAI on Regolo for free](https://dashboard.regolo.ai)

---

## **Resources &amp; Community**

**Official Documentation:**

- [Regolo Platform](https://regolo.ai) - European LLM provider, Zero Data-Retention and 100% Green
- [ThunderAI Github](https://github.com/micz/ThunderAI) – It’s a Thunderbird Addon that uses the capabilities of ChatGPT

**Related Guides:**

- [Boost Your Workflows with Regolo AI on n8n](https://regolo.ai/boost-your-workflows-with-regolo-ai-on-n8n/)
- [Build Multi-Agent Workflows with crewAI Teams](https://regolo.ai/build-multi-agent-workflows-with-crewai-teams/)

**Join the Community:**

- [Regolo Discord](https://discord.gg/ZzZvuR2y) - Share your automation builds
- [CheshireCat GitHub](https://github.com/cheshire-cat-ai) - Contribute plugins
- Follow Us on X[ @regolo\_ai](https://x.com/regolo_ai) - Show your integrations!
- Open discussion on our[ Subreddit Community](https://www.reddit.com/r/regolo_ai/)

---

## **🚀 Ready to Deploy?**

## [**Get Free Regolo Credits →**](https://dashboard.regolo.ai)

---

> *Built with ❤️ by the Regolo team. Questions? support@regolo.ai*