The problem: choosing wrong is very expensive
Real scenario: A company hires "AI experts." Pays €15,000 upfront. 3 months later: the system doesn't work as promised. Money lost + time lost + frustration.
Another scenario: They set up a generic Tidio/ManyChat chatbot. It doesn't understand complex queries. Frustrated customers. The company turns off the chatbot.
Solution: Know what to ask before hiring.
How Should You Choose an AI Automation Company?
When selecting an AI partner, demand documented case studies with real metrics, transparency regarding their technology (e.g., custom integrations vs. generic templates), and clear pricing models. A trustworthy provider, like Drishtech, often offers a "pay-for-results" structure with no lock-in contracts, ensuring the AI solution genuinely benefits your business, whether in hospitality, gastronomy, retail, or real estate.
10 questions you MUST ask before hiring
1. Do they have documented real case studies?
Red flag: "We work with many companies but can't share names."
Good sign: They show you case studies with data:
- Company X: increased conversions by Y%
- Company Z: saved €X/month
- Verifiable data (links, real testimonials)
What to ask for: At least 2-3 documented case studies with real metrics (not vague promises).
2. How do they charge? (Pricing model)
Red flags:
- Everything upfront with no guarantees
- Long lock-in contracts (24 months)
- No cancellation option
Good sign:
- Pay for results (you only pay if it works)
- Or a hybrid model: small setup fee + % of results
- No mandatory lock-in period
What to ask for: A clear contract with cancellation terms. What happens if it doesn't work (refund / no charge).
3. What technology do they use?
Red flags:
- "We use advanced AI" (without specifying)
- Template chatbots (ManyChat, Tidio without customization)
- No integration with your systems
Good sign:
- They explain which AI models they use (GPT-4, Claude, etc.)
- Full customization (not generic templates)
- Integration with your current systems (CRM, PMS, Shopify, etc.)
What to ask for: A working demo (not just slides). A test with real cases from your business.
4. How long does implementation take?
Red flags:
- "It's ready in 2 days" (impossible if it's customized)
- "It takes 6 months" (too long for something simple)
Good sign:
- 1-4 weeks depending on complexity
- Clear timeline with milestones
What to ask for: A step-by-step implementation plan with realistic dates.
5. How do they handle cases the AI doesn't understand?
Red flags:
- "The AI understands everything" (a lie, no AI is perfect)
- Customer is left without a response when it fails
Good sign: Clear explanation of what happens when the AI doesn't know:
- Escalates to a human
- Logs the query
- Learns for next time
What to ask for: A real example of how they handle edge cases.
If you want to dive deeper into common mistakes when implementing AI, read our article on mistakes when implementing AI in companies.
6. Can I see the conversations and data?
Red flags:
- "Everything is automatic, you don't need to see anything"
- No dashboard
- No access to metrics
Good sign:
- Full dashboard
- Access to all conversations
- Real-time metrics (conversions, response rate, etc.)
What to ask for: A demo of the dashboard. Which metrics are tracked.
7. Who handles maintenance?
Red flags:
- "After setup, you don't need anything" (a lie, there are always adjustments)
- Extra charges for every small change
Good sign:
- Maintenance included in the price
- Automatic updates
- Clear technical support
What to ask for: What's included in maintenance. How to request a change.
8. Does it integrate with my current systems?
Red flags:
- "You need to change your entire system"
- Doesn't integrate with your CRM/PMS/Shopify
Good sign:
- Connects with what you already use
- Open API for future integrations
What to ask for: List of supported integrations. What happens if you use a non-standard system.
9. What happens with my data?
Red flags:
- No clear privacy policy
- They use your data to train AI (without permission)
Good sign:
- GDPR compliance
- Encrypted data
- They don't share or sell data
What to ask for: Written privacy policy. Where data is stored (EU / US / etc.).
10. Can I cancel whenever I want?
Red flags:
- 24-month lock-in contract
- Cancellation penalties
- "Once paid, no refunds"
Good sign:
- Cancel anytime
- No penalties
- You own your data if you cancel
What to ask for: Written cancellation terms.
Talk to a company that meets all these criteriaBusiness model comparison
| Model | Pros | Cons | When to choose |
|---|---|---|---|
| Upfront payment | Known fixed price | High risk if it doesn't work | Only if they have proven cases |
| Fixed monthly fee | Predictable | You pay even if it doesn't deliver results | If you've already validated it works |
| Pay for results | Zero risk | Usually a higher % | Ideal for risk-free validation |
| Hybrid model | Lower setup + % of results | Requires initial investment | Good risk/cost balance |
Recommendation: If you don't have proven cases, start with pay for results.
Serious red flags (run if you see these)
- No real case studies: "We're new but really good"
- Everything upfront with no guarantees
- Impossible promises: "You'll increase sales 300% in 1 month"
- Can't explain how the technology works (just buzzwords)
- No clear technical support
- Long lock-in contracts with no exit
Final checklist before hiring
- They have at least 2 documented real case studies
- Clear pricing model (preferably pay for results)
- Working demo (not just slides)
- Integration with your systems
- Dashboard with metrics
- Technical support included
- Clear privacy policy
- Cancellation without penalties
- Realistic timeline (1-4 weeks)
- They explain what happens when the AI fails
To dive deeper into how intelligent automation works in real companies, check out our guide on intelligent automation for companies.
Next step
If you want AI automation with zero risk, we can help.
You only pay for results. No conversions, no charge.
Request a free consultation and see why we meet every criteria on this list