Hiring an AI implementation consultant: what a good engagement looks like
Most AI projects in small businesses don't fail because the technology is bad. They fail in implementation: the gap between a promising demo and something the team relies on every day.
That gap is the whole job of an AI implementation consultant. Here is what a good engagement looks like, what an AI agent development project should include, and the questions I'd ask before hiring anyone, including me.
What a good engagement looks like
Whether the project is an automation or a full AI agent, a sound implementation runs in six stages. If a proposal skips one, ask why.
| Stage | What happens | What you should receive |
|---|---|---|
| 1. Job description | Define what the AI may see, do and never touch | A written scope, like hiring a person |
| 2. Connections | Link it to the phone, calendar, CRM, inbox or store | A list of every system it can reach, and with what permissions |
| 3. Guardrails | Rules for when it stops and hands off to a person | The hand-off rules in writing |
| 4. Testing | Run it on real past cases before customers see it | Test results, including the failures |
| 5. Launch and training | Go live in a limited way; show the team how to work with it | A short guide your staff will actually read |
| 6. Monitoring | Review what it did, where it struggled and what to adjust | A monthly report against the agreed numbers |
Stage one is the one I care about most. I write every agent a job description before any code. What may it see? What may it do on its own? What must it never touch? If you can't answer those for a person you're hiring, you shouldn't give the job to software either.
What AI agent development services should include
An AI agent is different from a chatbot. A chatbot answers questions. An agent takes steps: it answers the call, checks the calendar, books the appointment, updates the CRM, and hands the conversation back to you when it should. I explain the difference in more detail on AI agents.
Because an agent acts, development has to include more than the conversation. Ask for these explicitly:
Tool permissions. Exactly which actions it can take, and which it can only suggest.
Disclosure. It should tell people they're talking to AI. Customers forgive a machine; they don't forgive being tricked.
Data handling. What it stores, for how long, and who can see it. If the job doesn't need a piece of data, it shouldn't be stored.
A breach plan. Who gets told, by whom, and how fast, if something goes wrong.
Ongoing monitoring. An agent isn't finished at launch. It needs reviewing, especially in the first weeks.
Questions to ask before you hire
Can I see one you've built that's running now? Not a demo. Something live.
What did your last project get wrong, and what did you change? Everyone has an answer to this. The honest ones are useful.
What happens if I stop paying? Find out what you own.
Who reviews the agent's work, and how often? "Nobody, it's automatic" is the wrong answer.
What number should move, and by when?
How I price it
A done-for-you AI agent is $3,500 to set up and $1,000 a month to run and monitor. Custom builds beyond that get a written fixed-price scope first; see custom AI software. If you'd rather start by seeing what an agent could do, you can build a free preview in Agent Studio in a few minutes, then bring it to a free 30-minute call. Every price is on pricing.
Common questions
What does an AI implementation consultant do?
An AI implementation consultant turns an AI plan into working systems: choosing the tools, connecting them to existing software, setting guardrails and hand-off rules, testing on real cases, training the team and monitoring results each month. Implementation is where most AI projects succeed or fail.
What do AI agent development services include?
They should include a written job description for the agent, connections to your tools with clear permissions, guardrails and hand-off rules, testing on real conversations, disclosure that customers are talking to AI, data-handling rules, launch support and ongoing monitoring.
How much does it cost to build an AI agent for a small business?
It depends on scope. My done-for-you AI agent is $3,500 to set up plus $1,000 a month to run and monitor, and larger custom builds are quoted as a fixed price after scoping. You can build a free preview in Agent Studio first.
How long does AI implementation take?
A focused first project, such as answering and routing enquiries, can usually go live in weeks rather than months, followed by a closely watched first month. Larger projects take longer, which is why starting small is the better plan.
About the author. Siamak Kalhor is an AI marketing and LLMO consultant in Los Angeles with more than forty years of experience in marketing strategy. He advises professionals and e-commerce leaders on becoming the answer AI assistants give. Get in touch, or connect on LinkedIn.