Generative Messaging — Business Overview
Last updated: February 2026
Feature Owners
| Role | Owners |
|---|---|
| Product managers | Ivan Rakikj, Jussi Kuusisto |
| IT project manager | Martina Todorovska |
| Lead developers | Marija Vitorovikj, Gabriela Vanevska |
| Stakeholders | Joe, Dragica, Kristijan Todorov |
Table of Contents
- What Is Generative Messaging?
- Why It Matters
- How It Works — The Big Picture
- The Email Generation Process
- Quality Assurance — The AI Check
- Multi-Provider AI
- Follow-Up Chasers
- What Users See
- Glossary
1. What Is Generative Messaging?
Generative Messaging is Sopro's AI-powered email writing system — internally called AIM (AI Messaging). Instead of sending the same template to every prospect, AIM creates unique, personalised emails for each recipient using artificial intelligence. The AI considers the prospect's role, their company, what the campaign is selling, and why that company would be a good fit.
In Simple Terms
"AIM is like having a team of expert copywriters — each one researches a prospect and writes them a custom email explaining why the client's product matters to their specific company."
2. Why It Matters
| Generic Templates | AI-Generated Messaging |
|---|---|
| Same email to every prospect | Unique email tailored to each recipient |
| No mention of the prospect's specific company context | References the prospect's industry, role, and company |
| Low engagement — feels impersonal | Higher open and reply rates — feels personal |
| Time-consuming to write variations manually | Thousands of unique emails generated automatically |
| No quality checks on content | AI quality agents review every message before sending |
3. How It Works — The Big Picture
The Five Steps
- Campaign Brief — The client describes what they're selling and who their ideal customer is
- Company Research — The system gathers data about each prospect's company
- AI Generation — An AI model writes a personalised email for each prospect
- Quality Check — A separate AI reviews the email for quality, accuracy, and tone
- Delivery — The approved email is stored and sent through the email sending system
4. The Email Generation Process
What the AI Receives
For each prospect, the AI is given a rich package of information to write from:
What Makes Each Email Unique
| Element | How It's Personalised |
|---|---|
| Opening line | References something specific about the prospect's company |
| Value proposition | Explains why the client's product matters to this specific company |
| Relevance | Connects the prospect's role to the client's offering |
| Call to action | Tailored to the prospect's seniority level |
5. Quality Assurance — The AI Check
Every generated email is reviewed by a separate AI quality agent before it's approved for sending. This acts as a "second pair of eyes."
What the QA Agent Checks
Why Two AIs?
- The writing AI optimises for persuasion and personalisation
- The QA AI optimises for accuracy and compliance
- Having separate agents prevents the writer from "marking its own homework"
6. Multi-Provider AI
AIM doesn't rely on a single AI provider. It can use multiple large language models:
| Provider | Strengths |
|---|---|
| OpenAI (GPT) | Strong general-purpose writing, reliable |
| Anthropic (Claude) | Excellent reasoning, detailed responses |
| DeepSeek | Cost-effective for high volume |
| Grok | Fast generation with good quality |
Why Multiple Providers?
- Resilience — if one provider goes down, others continue working
- Cost optimisation — different providers have different pricing
- Quality comparison — Sopro can test which provider works best for each use case
- No vendor lock-in — freedom to switch providers based on performance
7. Follow-Up Chasers
If a prospect doesn't respond to the initial email, AIM generates follow-up messages called chasers. Each chaser is also AI-generated and personalised.
Chaser Strategy
Each chaser:
- References the previous email
- Takes a different angle (new benefit, different value prop)
- Is shorter than the previous one
- Maintains the same professional tone
8. What Users See
Sodastream (Internal CRM)
| Feature | Description |
|---|---|
| Generated Emails | View AI-generated messages before sending |
| Quality Score | See the QA agent's assessment of each email |
| Regeneration | Request a new version if the generated email isn't suitable |
| Template Configuration | Set up the campaign brief and style guidelines for AIM |
Data Admin
| Feature | Description |
|---|---|
| Campaign Details | View which AI provider was used for generation |
| Message Preview | Preview generated emails for prospects |
9. Glossary
| Term | Definition |
|---|---|
| AIM | AI Messaging — Sopro's AI-powered email generation service. |
| Generative Messaging | The process of using AI to create unique, personalised email content for each prospect. |
| Campaign Brief | A document describing what the client sells, who their ideal customer is, and the key messaging points. |
| Company Suitability | An AI-generated score and explanation of why a prospect's company would benefit from the client's product. |
| QA Agent | A separate AI that reviews generated emails for quality, accuracy, and compliance before approving them. |
| Chaser | A follow-up email generated by AI when a prospect doesn't respond to the initial message. |
| Prompt | The instructions and context given to the AI model to generate the email content. |
| AI Provider | A company offering large language model access (OpenAI, Anthropic, DeepSeek, Grok). |
| Personalisation | Tailoring email content to reference specific details about the prospect and their company. |
| Regeneration | Requesting the AI to write a new version of an email that didn't meet quality standards. |