Short answer: the best predictor of whether an AI agency will deliver is whether they use what they sell in their own operations. Ask to see their internal automations, demand a fixed-price scoped pilot (2-4 weeks) instead of an open-ended discovery phase, and check that they talk about your workflows — not about models. Below: the 7 questions that expose the difference in one call.
Everyone is an "AI agency" in 2026. The difference between the ones that ship and the ones that sell decks is easy to detect if you ask the right things — here is the checklist we would use ourselves as buyers.
The 7 questions
1. "Show me what AI runs inside YOUR business." The killer question. An agency that automates its own lead handling, reporting and support will show you live systems in two minutes. One that cannot is selling theory. (Our own operations run on the same AI agents and n8n workflows we deploy for clients — ask us and we will show you.)
2. "What exactly will be working after the pilot, and what does it cost?" Good answer: a concrete deliverable ("an agent answering your support email, trained on your docs, escalating to your team") with a fixed price and a 2-4 week timeline. Bad answer: a discovery phase, then a roadmap, then pricing.
3. "Which model will you use and why?" You want to hear it depends on the use case with specifics (Claude for long-context and reliability-sensitive work, GPT where it fits, small models where latency matters). Model-agnostic with reasons beats fanboys of a single vendor.
4. "Who owns the prompts, workflows and infrastructure?" The only acceptable answer: you do, from day one. Lock-in through hosted black boxes is the most common trap in AI services right now.
5. "What happens when the agent doesn't know?" Mature shops design the escalation path first — handoff to humans with full context, logged and measurable. If hallucination handling sounds improvised, the production incidents will be too.
6. "How do you measure success?" Resolved-without-human rate, lead response time, hours saved per week — numbers tied to your P&L. "Engagement" is not a metric, it is a dodge.
7. "What does maintenance look like after launch?" Models drift, your business changes, prompts need tuning. A real answer includes monitoring, a feedback loop with real conversations, and a defined support arrangement.
Red flags that predict failed projects
- Pricing that starts with a five-figure "AI strategy audit" before any working software.
- Demos that only work on their data, never on yours.
- Promises of "fully autonomous" anything in week one.
- No answer to the ownership question.
- A team that talks about AGI more than about your CRM.
The boring truth about pricing
Fair AI agency pricing in 2026 is unglamorous: a fixed quote per scoped deliverable, integration work priced by systems touched, and a maintenance arrangement you can cancel. That is how we price AI agent projects — quote within 24 hours of a discovery call, pilot in weeks, you own everything. If you want to test the checklist above on us, book the call — question 1 included.