AI can take repetitive admin off your plate and speed up your consulting work. However, it's not the answer to everything. Use it to draft, sort, and summarise, but keep decisions with strategic, financial, or reputational weight for yourself.
The consulting tasks made for AI are transactional, repetitive or lack a competitive advantage. These tools can only play back ideas, concepts, or results that have already been tried and implemented.
All the tool does is pattern matching. There's no room for new or untested ideas.
That gap is a consultant's biggest value-add.
A client isn't paying you for a generic report. You're being paid for your judgement, expertise and observations applied to the client's specific problem, and no amount of prompting AI will get you a solution built for that problem.
Is it a good idea to feed AI client data?
Before we look at the consulting tasks you shouldn't give to AI, let's go back a few steps.
Most consulting agreements have a standard confidentiality clause. If you paste a client's financials, an unreleased product roadmap, or interview transcripts into a chat tool like Claude or ChatGPT, you risk breaching it.
Before you paste anything in, check two things. First, check your contract, since some clients restrict which tools you can use. Second, read the tool's data terms to find out whether it keeps what you paste and trains its models on it.
The same applies to any software with AI built in, including Kiaro. Kiaro only sends data to an AI provider when you use an AI feature, and strips card numbers, bank details and tax IDs first. Its providers work under API terms that prohibit training on your data, as set out in our privacy policy.
Read more: AI Micro-Automations for Freelance Admin: 10 Low-Risk Use Cases
Which consulting tasks shouldn't you hand over to AI?
Here are the four consulting tasks to keep for yourself, and the risks of relying on AI too heavily.
1. Strategic final calls
An AI model like Claude or ChatGPT can do the following consulting tasks in about five minutes:
- Model three pricing scenarios
- Run a competitor analysis scan
- Draft a five-year growth plan
While it helps speed up the grunt work, it's not the end product. Your clients are hiring you for your strategic insight. If you leave that up to AI, you're selling a generic output that carries none of the expertise your client is paying for.
If AI makes a strategic decision that ends up being wrong, you're the one who will be held accountable.
2. Contextual diagnostics
Think of AI as a painter only using broad strokes to create a picture. From a distance, it looks like a painting of a forest, but on closer inspection there's no detail. Only different shades of green and brown.
That's the main drawback of relying on AI to diagnose a client's problem. You're setting yourself up for textbook plausible reasons, yet none will capture the detail or nuance that comes from reading the room.
Kiaro founder Collen Kriel, who also works as a freelance data analyst, ran into this on a dashboard project:
A client asked me for a dashboard that tracked their thought leaders. Nobody had defined what a thought leader was. You only find that out by asking.
AI can't do the observing for you. It can only organise what you've observed.
3. Pricing and rate decisions
AI is good for drafting a rough ballpark figure of what to charge or how to price your consulting services. It's useful when you're starting out as a consultant and need some guidance.
However, AI models draw on job listings, salary surveys, and articles. Their answers aren't based on your skill set, what you need to earn to cover your living expenses, or how many billable hours you have available.
The data may also not be relevant to your location. For example, a consultant operating in South Africa and one in the United States might charge different rates, based on what clients are willing to pay and local service benchmarks.
Collen's own rates show how much depends on context:
I'm South African, I live in the Netherlands, and most of my clients pay in US dollars. This year I've quoted between $50 and $75 an hour. The rate moved with the work, from dashboards at the low end to data engineering at the top.
Your consulting rate is personal. Take the time to run the calculations for your own circumstances, instead of taking AI's response as gospel.
Read more: How to Close Your Effective Rate Gap as a Consultant
4. Fact-checking your analysis
AI loves to hallucinate.
Give it real client data, and it will produce a plausible-looking answer, correct or not, and present it with the same air of confidence.
The flaws surface when you introduce any ambiguity into the data set. As soon as there are missing fields, conflicting sources or other inconsistencies, you can't bank on AI giving you a correct answer.
That's a problem for consultants, because client data is usually messy, conflicting and incomplete.
To avoid handing in a factually incorrect deliverable with your name on it, treat any AI-generated number, trend or data summary with caution. Verify it first, and trace it back to the source data before it goes anywhere near a client.
Let AI draft the admin. Keep the final say.
Kiaro's AI drafts the invoices, projects and follow-up emails, and nothing is saved or sent until you review it and click confirm. Free, no credit card.
Where should you draw the line with AI and consulting tasks?
AI is great at regurgitating data and reacting to the information you feed it.
What it lacks is the human ability to observe and spot things that don't fit into a neatly packaged pattern or have never been documented before.
Let it take over the boring, repetitive administrative tasks slowing you down. Don't use it to make decisions that need your strategic oversight and have a direct impact on your money and reputation.
Frequently asked questions
Can AI help price a consulting project?
Yes and no. AI can pull up comparable rates, but you can't rely on it to dictate what you should charge for a consulting project. You make the final call, and it should reflect your real numbers. For a realistic number, use a rate calculator that takes your expenses and available billable hours and works out the rate you need to charge.
Is it safe to let AI draft emails to clients?
You can use AI to draft emails up to a point. It helps you get past the blank page and generate templates for invoice reminders and other routine messages. You should still adjust the tone, because most of us have read enough AI output to recognise the machine cadence, which comes across as impersonal.
Can I rely on AI to review a contract before I sign?
Yes, you can use AI to help you find anything you may have missed in a contract. For example, Kiaro runs an AI risk analysis on contracts to identify missing clauses and highlight red flags. It's a good way to double-check yourself, but you should still read the contract in full. AI isn't infallible and might miss something you want to push back on.