The case study
What actually
happened.
A dental laboratory in Hungary needed new clients in Austria. The traditional path: hire a sales rep at €3,000–5,000/month, wait 3–6 months for results, accept that they work 8 hours a day, take holidays, and forget to follow up.
Instead, we built an autonomous outreach agent. Here is exactly what it does every day.
Dental Lab — Austrian Expansion
Autonomous lead generation & outreach agent
Live system
08:00
Agent starts. Searches Austrian dental directories, Google Maps, job boards, and professional registries for practices matching a precise qualification profile — newly opened, showing expansion signals, no visible lab partnership.
08:30
47 practices found. Each scored on 8 qualification criteria on a 10-point scale. Practices below 6 filtered out. The ones that pass get their websites fetched and analysed for personalisation signals — a new piece of equipment, a recent award, a hiring post that signals growth.
09:15
12 personalised German-language emails drafted. Not templates with names swapped in. Each opening line references something specific and real about that practice. Tone and formality calibrated to Austrian dental industry norms.
10:00
Owner receives all drafts for review. Two words to send any of them. Skip one that doesn't feel right. Send everything at once. The approval step takes under 3 minutes.
Next day 09:00
Agent checks Gmail for replies. Identifies which leads responded and where each conversation stands. Drafts follow-up emails calibrated to each situation — a cold lead after 7 days gets a different angle than one that opened but didn't reply.
Always
Two agents watch each other. If one crashes, the other restarts it. The business owner's only job: approve emails and respond to interested prospects.
5–20
personalised emails per day
365
days a year, no holidays
~3 min
owner time per day
"This is not a chatbot. This is not automation in the Zapier sense. This is an autonomous agent with judgment — it makes decisions, handles exceptions, recovers from failures, and operates continuously without supervision."
Why this is different
Your Zapier stopped at the
Czech website.
Traditional automation is brittle. It handles the case you anticipated. When a dental practice website is in Czech instead of German, a workflow breaks. When a lead's email bounces, a sequence stops. When a new directory appears that wasn't in the original spec, the system misses it entirely.
An agent adapts. It reads the Czech website anyway and extracts what it needs. It flags the bounced email and tries to find an alternative contact. It discovers the new directory during a search and incorporates it without being told to.
This matters enormously for small businesses because their processes are inherently irregular. A 5-person company doesn't have clean CRM records or predictable workflows. Traditional automation requires you to formalise everything before you can automate it. Agents operate in the mess.
Traditional automation (Zapier, Make)
Breaks when input doesn't match expected format
Can't read a website in the wrong language
Stops when a lead email bounces
Misses sources not in the original spec
Sends identical templates to every prospect
Requires clean, structured data to function
AI Agent
Handles exceptions and adapts in real time
Reads any website, extracts what it needs
Flags the bounce, finds an alternative contact
Discovers new sources and incorporates them
Personalises every email with specific, real detail
Operates in messy, irregular real-world data
The economics
One new client pays for
everything. Permanently.
For a business with a €5,000 monthly value per new client, converting even one additional client from a system that costs €500/month is a 10× return. The math works at almost any realistic conversion rate.
Option
Monthly cost
What you get
Junior sales hire
€1,500–2,500
8 hrs/day, 5 days/week. Takes holidays, sick days, needs management. Leaves after 14 months on average.
AI SDR SaaS tools
€499–599
Monthly subscription per agent. €6,000–7,000/year for one function, on software you don't own and can't modify.
MesterAI agent
~€30–80 running costs
Runs 7 days a week. Searches, scores, personalises, follows up. Setup fee paid once — you own the system.
* Running costs = server + AI model usage paid directly to your providers. MesterAI setup fee is one-time. Optional monthly retainer covers ongoing maintenance and support.
Beyond lead generation
The same architecture.
Any business function.
The dental lab system is one implementation of a general architecture. The same stack — autonomous agent, approval workflow, outreach channel, watchdog — can be deployed across virtually any business operation that currently requires a human doing repetitive, information-heavy work.
🎯
Lead generation & outreach
Finds prospects in any directory or database, scores them against your ideal profile, drafts personalised outreach, manages follow-up sequences. Runs every morning without being asked.
🔍
Competitive & market intelligence
Monitors competitor websites, pricing pages, and job postings daily. Synthesises what changed and what it means for your business. Not a Google Alert — an agent that reads and reasons.
🔄
Client retention monitoring
Tracks engagement signals across your client base — payment patterns, communication frequency. Flags at-risk relationships and drafts re-engagement messages before churn happens.
📊
Financial operations
Daily P&L pulled from your accounting software, cash flow anomalies flagged, overdue invoices chased automatically. Your morning starts with a CFO-style summary, not a spreadsheet.
👥
Recruitment pipeline
Sources candidates from job boards, LinkedIn, and professional communities. Scores against your criteria. Drafts personalised outreach. You see warm leads, not raw CVs.
⚖️
Regulatory monitoring
Watches relevant regulatory bodies daily. Flags changes and summarises implications for your specific business. No more missing a rule change buried in a government PDF.
🤝
Partnership development
Identifies potential referral partners, distributors, or resellers. Researches their fit, drafts introduction emails, tracks conversations through to response.
📦
Procurement & supplier monitoring
Monitors supplier prices and availability. Flags when a preferred supplier raises prices or a better alternative appears. Drafts RFQ emails to shortlisted suppliers.
This works across industries. A few examples of who this is built for:
Recruitment agencies
Law firms
Accounting firms
Commercial cleaning
Industrial suppliers
Translation agencies
SaaS companies
Financial advisors
Healthcare providers
Real estate agencies
Logistics companies
Marketing agencies
How control works
The agent proposes.
You decide.
The most important design decision in every system we build is where the human stays in the loop. The agent does everything that benefits from speed, consistency, and scale: searching, qualifying, personalising, drafting. You do everything that requires judgment, relationship awareness, and accountability.
The approval interface is designed to make oversight frictionless rather than a bottleneck. Review 12 email drafts in 3 minutes. Send 9, skip 3. Done. The agent handles delivery, tracking, and follow-up.
As trust builds, the human step can be selectively removed. Follow-up emails where the template is proven can be sent automatically. Initial outreach to lower-priority leads can run without review. The architecture supports this progression without rebuilding anything.
"Péter knows his business, his reputation, and his relationships in ways the agent never will. The agent knows how to search Austrian dental directories at 8am every morning without being asked. Each does what it's best at."
Being honest
What these systems
cannot do.
Agentic systems are not magic and overselling them does a disservice. Here is what to expect honestly.
⚠️
They make mistakes
An agent will occasionally misqualify a lead, write an email with a subtle error, or fail to find a contact that exists. The human approval layer exists precisely because these errors happen. Quality improves over time as the system learns what works for your business.
🔧
They require setup and maintenance
The dental system took several sessions to build properly — the infrastructure, qualification criteria, email templates, approval flow, watchdog architecture. This is not a plug-and-play product. That is why there is a setup fee and an optional monthly retainer.
🧠
They amplify your thinking — in both directions
An agent with poorly written qualification criteria finds the wrong leads. An agent with weak email templates drafts weak emails. The AI amplifies the quality of the human thinking behind it. We work with you to get this right before launch.
🗺️
They are not a substitute for strategy
An agent can execute outreach at scale but it cannot invent a compelling value proposition, identify the right market, or recover a relationship that went badly. The strategic thinking still comes from you. What the agent does is execute that strategy consistently, at scale, without the overhead of hiring for it.
"What they are is a force multiplier for businesses that already know what they want to do and need help doing it consistently, at scale, and without the overhead of hiring for it. That is a very large market."
Why now
The gap between early movers
widens every month.
The businesses that adopt this first in any given industry gain a compounding advantage. More outreach means more data on what works. Better data means better qualification criteria. Better qualification means higher conversion rates. Higher conversion funds more sophisticated automation.
The gap between early adopters and laggards widens over time — not because the technology is unavailable to everyone, but because the operational learning doesn't transfer. Your competitor's 6 months of refined qualification criteria and tested email templates cannot be copied by someone starting today.
Five years ago, building this required a developer team, months of work, and significant ongoing maintenance. Today it can be provisioned in hours. The same system that required enterprise budgets in 2020 now costs less per month than a single business lunch to run.