AI Strategy? Start with One Process (Not a 50-Page Plan)
Erwin Berkouwer
Every month you wait costs your team 8+ hours. Start with one AI process, go live in 4 weeks and learn what really works — without a 50-page roadmap.
Every month you delay with AI, your competitors are saving an average of 8 hours per employee per week. With a team of 5, that's 40 hours — an entire work week that evaporates. Not through grand AI strategies, but through one simple automation. The question is no longer whether you start, but how long you wait.
The best AI strategy for SMBs is no strategy. It's one concrete process, live in 4 weeks.
Why big AI plans fail (and what it costs you)
Recognize this? A consultant creates an "AI Roadmap 2025–2027". 50 pages. A team of five. A budget of €50,000. And six months later: nothing. The document sits in a drawer.
We see this pattern at 70% of the SMBs we talk to. The reason is always the same: too much strategy, too little action.
What it really costs:
- While you analyze, your competitor is already automating his quote process (saving: 6 hours/week)
- While you compare budgets, a competitor is already processing his purchase invoices automatically (saving: €12,000/year)
- While you meet about the roadmap, a competitor is already answering his FAQs automatically (saving: 3 hours/day)
"We'd talked about AI for half a year. In the first 4 weeks of actually starting, we saved more time than in all those meetings combined." — Director, Logistics Company, 45 employees
The only AI strategy that works for SMBs: one process
The solution is so simple it almost feels like cheating: choose one process, automate it in 4 weeks, and learn from the results.
No roadmap. No strategy document. Just start.
Here's how it works:
- Choose one painful, repetitive process (see the selection framework below)
- Build a working prototype in weeks 2–3
- Go live in week 4 — with real data, real results
- Scale only after proof — based on what you've learned
The result? You don't need a strategy anymore. You have experience.
How do you choose the right first process?
This is where most companies stumble: they choose a process that's too complex, has too little data, or delivers too little return. Use this framework:
| Criterion | Good for AI | Not suitable for AI |
|---|---|---|
| Repetition | Same steps, over and over | Different every time |
| Time Investment | 5+ hours/week per employee | Sporadic, less than 1 hour/week |
| Available Data | 50+ examples per month | Fewer than 10 per month |
| Human Judgment | Objective, based on rules | Emotional, situational |
Examples that always work:
- Processing purchase invoices → automatically read and record
- Customer questions via email → automatically answer or forward
- Quote requests → automatic triage and draft response
- Compiling reports → automatic from multiple data sources
Examples that don't work as well:
- Complaint handling with emotional context
- Strategic decisions about new markets
- Creative work where originality is essential
Doubt? Take the free AI scan — in 10 minutes you know which processes in your business are most suitable.
What a 4-week implementation looks like (practical example)
A transport company with 40 employees received 60–80 emails daily with track & trace questions. Their administration spent 18 hours per week answering them — 936 hours per year, or €28,000 in labor costs.
Week 1 — Analysis (2 hours your time)
We analyze the process: how many emails, what variations, which systems are involved. Result: a concrete plan of 2 pages.
Weeks 2–3 — Building
The system reads incoming emails, recognizes track & trace numbers, checks the carrier, and automatically sends an answer. You test it on real emails.
Week 4 — Live
The system goes live. Your team gets one hour of training. We monitor closely the first week.
Result after 6 weeks:
- From 18 hours/week to 2 hours/week (-89%)
- Response time from 4 hours to 12 minutes
- Investment: €4,200 one-time
- ROI: fully repaid within 9 weeks
For more examples, check our complete AI implementation guide or discover the 5 processes SMBs most automate.
What comes after the first success?
Once the first process is live, you have something much more valuable than a roadmap: proof. You now know what AI can do in your specific context, how your team works with it, and which processes are next in line.
Typical pattern we see:
- Months 1–2: First process live (saving: 5–15 hours/week)
- Months 3–4: Second process automated (cumulative saving: 15–30 hours/week)
- Month 6+: Multiple AI agents running independently — team focuses on work with real value
An accounting firm that started this way with invoice processing (€3,500 investment) had their entire administration automated within 9 months. Total investment: €6,000. Total savings: €65,000 per year.
When do you actually need a bigger strategy?
Honest answer: almost never, if you're an SMB with fewer than 250 employees.
An extensive AI strategy makes sense if:
- You have more than 1,000 employees and complex IT infrastructure
- AI becomes your core product (tech startup)
- Compliance is heavy (banking, healthcare, government)
For most SMBs: every week you wait for the perfect strategy is a week your competitor uses.
Ready to start?
You don't need a consultant. You need one well-chosen process.
In 30 minutes we can determine together which process in your business brings the most time savings. Use our ROI calculator to see what automation brings your business, or schedule a free process analysis — no strings attached.
Frequently Asked Questions
How much time does it take me to start with AI?
The first week asks 2–3 hours of you: a conversation about the process to automate and sending examples. After that, Unify AI builds the system — you test in between. Total time investment for you: 6–8 hours in the first month.
Does my business need to be data-ready first?
Not necessarily. For most starting processes (email, invoices, quotes) you already have existing data available. We check this in the first conversation. Only if you have fewer than 50 examples per month do we recommend a different starting process.
What if the first AI project fails?
That's rare if you choose the right process, but it can happen. That's why we always start small: maximum investment of €5,000, results in 4 weeks. That way you limit your risk while learning what works.
How much does a first AI implementation cost for an SMB?
A typical starting project costs €2,500–€6,000 one-time, depending on complexity. Most customers repay this within 8–12 weeks through time savings. We always work with a fixed price — no open-ended costs.
Does my team need technical knowledge?
No. You and your team use the system like normal software. The technical side is entirely our responsibility. We provide training for daily use — you don't need to program or manage anything.
