What are AI consulting services, and what do you get from ours?
AI consulting services help a company decide where AI and automation should do work, which tools to use, and what rules apply. Ours end in a written plan and, if you want it, a working system that we build on your accounts.
Some AI consulting ends with a slide deck. Ours has two parts, and you can stop after the first one:
- Free audit, then paid consulting and a written roadmap. The strategy call and first audit are free. The consulting work, quoted for your business by size and scope, looks at how work moves through your company today, score each candidate process for AI readiness, and write down what to build first, which tools and models fit, what data has to be cleaned up, and which compliance questions need an answer before launch. Where you need one, we also write your AI policy, fitted to your brand, your customers and the tools your teams use. We do this for small firms and large companies alike.
- Implementation. We build the plan on your own accounts: AI agents for voice, chat and reception work, workflows on n8n, Zapier or Make, a GoHighLevel CRM backend, or custom software where off-the-shelf tools fall short.
The roadmap is the main thing this page is about. If you already know what you want built and need a team to connect CRM, workflows and agents into one system, our AI automation services page covers that build work in detail.
What does the written roadmap contain?
- A list of candidate processes with a readiness score for each (the method is below).
- The first two or three builds, in order, with the hours or response time each should save, worked out from your own volumes.
- The tool and model choice for each build, and why.
- Data gaps to fix first: missing fields, files that live in inboxes, duplicate contacts.
- Compliance notes: where a person must approve, what customers must be told, and which rules from the table further down apply to you.
- A fixed price for each build.
Where should a business start with AI?
Start with a process that happens often, follows clear rules, uses data you can already reach, and costs little when a mistake slips through. Score every candidate on those points and build the highest scorer first.
Readiness scores in the worked example
We score five things from 1 to 3. The total is out of 15.
| Factor | 3 points | 2 points | 1 point |
|---|---|---|---|
| Volume | 500 or more a month | 100 to 499 a month | Under 100 a month |
| Rule clarity | Written rules or past examples cover most cases | Rules exist, exceptions are common | Mostly judgment calls |
| Data access | Data sits in systems with an API or export | Partly in email, PDFs or spreadsheets | In people's heads or on paper |
| Error cost | A mistake is cheap and easy to catch | Costs money but can be reversed | Harms a customer, or money or legal exposure is hard to undo |
| Compliance exposure | No personal or regulated data | Personal data under normal privacy rules | Regulated data or decisions about people (health, credit, housing, hiring) |
How we read the total: 12 to 15 means build it first. 9 to 11 means build it with a person reviewing the output. Under 9 means fix the data or the rules first. A score of 1 on error cost or compliance means a person approves every output, whatever the total.
Worked example: a property management company (illustrative)
This is an invented 25-person property management firm, used to show the method. The volumes, minutes and the $30 hourly staff cost are assumptions, not client data.
| Process | Volume | Rules | Data | Error cost | Compliance | Total | Hours a month |
|---|---|---|---|---|---|---|---|
| Answering after-hours and missed inquiries | 3 | 2 | 3 | 2 | 2 | 12 | 600 x 6 min = 60 h |
| Typing supplier invoices into accounting | 2 | 3 | 2 | 2 | 3 | 12 | 400 x 5 min = 33.3 h |
| Monthly owner reports | 1 | 3 | 2 | 2 | 3 | 11 | 40 x 30 min = 20 h |
| Approving rental applicants | 1 | 2 | 2 | 1 | 1 | 7 | 80 x 20 min = 26.7 h |
At the assumed $30 an hour, the first three are worth about $1,800, $1,000 and $600 of staff time a month (60 x 30, 33.3 x 30, 20 x 30). The roadmap would put inquiries first (it ties with invoices on score and saves the most hours), invoices second, reports third. Applicant approval stays with a person: the plan would only automate collecting documents and chasing missing ones. Put your own numbers into our automation ROI calculator before you trust any estimate, ours included.
How much do AI consulting services cost?
The strategy call and first audit are free. AI consulting, meaning the scored roadmap, tool and model choice and compliance review, is paid and quoted for each business by size and scope. Builds are a fixed price: a small build of two or three automations is about $1,000 at the time of writing.
What you pay us, at the time of writing:
- Strategy call and first audit: free.
- AI consulting and written roadmap: paid, quoted for your business before work starts. The quote depends on how many teams, processes and systems are in scope, so a small company and a large one get different numbers.
- Small build (two or three workflows or automations): about $1,000.
- Larger systems (several agents, a CRM backend, custom software): a fixed price after the audit. You see the number before anything starts.
- Fixes to anything we built: free for 6 months. Optional monthly support from $200 after that.
You also pay the tools directly: model usage, Zapier or n8n hosting, CRM seats, phone numbers. Our automation project cost breakdown lists what moves a quote.
What do other AI consultants charge?
Published rates vary a lot by who you hire. These are figures we read on October 30, 2026:
| Source | What it states |
|---|---|
| Clutch AI consulting pricing guide (updated March 25, 2025) | Average rate of AI consulting firms: $200 to $500 per hour |
| Clutch AI consulting directory, first eight firms listed | Hourly rates from $25 to $49 up to $150 to $199. Minimum project sizes: five at $10,000+, two at $25,000+, one at $50,000+ |
| New Origins AI on Clutch (2 to 9 people) | $150 to $199 per hour, minimum project $10,000+ |
Many firms on those lists also build software, so check whether a minimum covers the build or only the advice. When you compare quotes, ask whether the price includes the build, who owns the accounts afterwards, and what fixes cost after launch.
How does an AI consulting and implementation project work?
We talk for 30 minutes, audit your current tools and processes, send a written roadmap with fixed prices, then build, test and hand over. A small build takes days; larger systems take weeks.
From free call to handover
- Free strategy call. You tell us where time or leads are lost. See what happens on the call.
- Audit. With read access, we look at your CRM, inboxes, forms, phone setup and spreadsheets, and count real volumes.
- Roadmap. Scored processes, build order, tools, data fixes, compliance notes, an AI policy written for your brand if you need one, and a fixed price per build.
- Build. On your accounts, one process at a time, with logging and error alerts from day one.
- Test. Real past examples go through the system before customers do, including the awkward ones.
- Handover. Written notes on what each part does, plus 6 months of free fixes.
Timelines from our own published work: small builds of two or three automations take days, and our published custom software builds took 5 to 10 weeks. Our build timeline guide explains what slows a project down.
Which AI tools and models should a business choose?
Choose the simplest tool that does the job. Use plain rules where the steps are fixed, add one AI step where language must be read or written, and use an agent only where a conversation or multi-step decision is needed.
| The job | What we usually use | Why |
|---|---|---|
| Fixed steps: move data, send a reminder, update a record | CRM built-in workflows, Zapier, Make or n8n | No model cost, easy to test, behaves the same every time |
| Read or write language inside a workflow: sort an email, pull fields from a PDF, summarise a call | One AI step inside the workflow (see AI workflow automation) | The model does one narrow task and the workflow checks its output |
| Hold a conversation: answer inquiries, qualify, book | A voice or chat agent connected to the CRM (see AI customer service agents) | Needs memory, tools and rules for when to hand over to a person |
| Multi-step work across your own systems | A custom agent with defined tools (see AI agent development) | Each tool it may call is listed and logged |
| Back-office paperwork and approvals with no AI needed | Business process automation | Rules alone do the work |
For the model itself, we pick per task on accuracy against your real examples, cost per run, and data terms. One data point to check with any provider: OpenAI's API documentation states that data sent to the API is not used to train its models unless you opt in, and that abuse-monitoring logs are kept for up to 30 days unless the law requires longer (OpenAI, your data, read October 30, 2026). Read the equivalent page for whichever provider the roadmap names.
What rules apply when a business starts using AI?
In the US, existing consumer protection law applies to AI claims and AI use. If your AI's output is used in the EU, the EU AI Act can apply even to a non-EU company. NIST's framework is voluntary but useful.
EU AI Act dates a business should know
| Framework | What it says | What it means for an adopting business |
|---|---|---|
| NIST AI Risk Management Framework | Released January 26, 2023, "intended for voluntary use". Four functions: Govern, Map, Measure, Manage. A Generative AI Profile (NIST AI 600-1) followed on July 26, 2024. NIST says the framework is being revised. | A free checklist structure. Our roadmap's risk notes follow the same four headings: who is responsible, what can go wrong, how you measure it, what you do about it. |
| EU AI Act, Article 2 | Covers providers and deployers outside the EU when "the output produced by the AI system is used in the Union". | A US firm with EU customers can be in scope. |
| EU AI Act, Article 50 | AI systems that interact directly with people must be designed so those people are told they are dealing with an AI, unless that is obvious. | Chat and voice agents serving EU users should say they are AI. |
| EU AI Act, Article 4 | Providers and deployers "shall take measures to support the development of AI literacy" of the people who run AI systems for them. | Train the staff who use the system. Our handover notes explain what each part does. |
| FTC, Operation AI Comply (September 25, 2024) | Five actions over deceptive AI claims, including DoNotPay's "robot lawyer" ($193,000) and Rytr's AI review writer. The FTC Chair said "there is no AI exemption from the laws on the books". | Do not claim your AI does more than it does, and do not use AI to write fake reviews. |
Dates from the European Commission's AI Act page, read October 30, 2026: the Act entered into force on August 1, 2024; prohibited practices and AI literacy applied from February 2, 2025; general-purpose AI model rules from August 2, 2025; the Act became applicable on August 2, 2026 with exceptions, and transparency rules take effect in August 2026. After the AI Omnibus, high-risk rules for Annex III uses (such as employment and education) apply from December 2, 2027, and for Annex I products from August 2, 2028.
Sector rules sit on top of these. For health data, read our guide to HIPAA-compliant automation; for AI phone calls, see whether AI calling is legal. This page is not legal advice, and a roadmap that touches regulated decisions should be read by your lawyer.
When is AI consulting the right call, and when is it not?
It is worth it when you have several possible AI projects and no clear first step, or when data and compliance questions block you. It is not needed when the task is obvious and small enough to just build.
| Your situation | What fits |
|---|---|
| You know the one thing you want, such as missed calls answered | Skip the roadmap. Ask for a quote on that build. |
| Many ideas, no order, limited budget | A scored roadmap, then build the top one or two. |
| Data spread across inboxes, spreadsheets and old tools | Roadmap first. The data clean-up is usually the first project. |
| Company-wide change programme, board-level AI strategy, model training on your own data | A large consulting firm or a specialist data science team is often a better match than us. |
| You have developers and want advice only | A paid advisor billed by the hour may suit you better than a build team. |
| Decisions about people (credit, housing, hiring, health) | Roadmap with legal review. Automate the paperwork around the decision, and keep a person making it. |
Why do AI projects fail, and how do you prevent it?
In the projects we are asked to fix, the cause is usually scope, data or missing checks, seldom the model. A clear first process, clean data, a named owner and error alerts prevent most of it.
Our research on AI automation success rates traces the common failure statistics to their sources, and our AI agent adoption statistics cover how many businesses run agents today. The failures we are asked to fix look like this:
| Symptom | Usual cause | Fix |
|---|---|---|
| Pilot worked in the demo, stalled in daily use | Nobody owns the result; it sits outside the tools staff already use | Name an owner; write outputs into the CRM or inbox the team opens every day |
| AI gives confident wrong answers | It reads outdated or conflicting documents | Clean the source files first; limit answers to approved sources; send unknowns to a person |
| Costs climb month on month | AI used for steps that plain rules could do; no usage limits | Move fixed steps to rules; set spending caps and alerts |
| Automation stopped weeks ago and nobody noticed | No error alerts or logging | Error workflow, alert to a named person, weekly run report |
| Customers complain the bot will not let them reach a person | No handover rule | Clear handover triggers and a fallback to a human on every channel |
| Legal or brand trouble over what the AI said | No limits on what it may promise | Approved answer list, disclosure that it is AI, human approval for high-risk replies |
Who owns the AI roadmap, accounts and data?
You do. The written roadmap is yours to keep and use with any provider, and everything we build runs on your accounts, with your API keys and your data, so you can change providers without starting over.
- Model provider accounts, automation tools and the CRM are opened in your company's name and billed to you.
- API keys sit in your accounts. We use access you grant and you can remove it at any time.
- Prompts, workflow exports and documentation are handed over with the build.
- If you take the roadmap to another team, nothing in it is locked to us.
Read next:
- whether the EU AI Act applies to US companies
- 30-point AI readiness checklist
- how Autoesta writes and builds in your AI policy
- what Autoesta charges for AI consulting
More questions about AI consulting
What does an AI consultant do?
An AI consultant studies how work moves through a company, finds the tasks where AI or automation would save time or money, and recommends tools, models and safeguards. Some stop at advice. Others, like us, also build what they recommend. Ask any consultant which of the two you are paying for before you sign, and who maintains the system after launch.
What is AI implementation consulting?
AI implementation consulting covers the step after strategy: choosing the exact tools, connecting them to your existing systems, testing with real examples, training staff and handing over. It is where most of the risk sits, because data problems and missing handover rules only show up once the system meets real customers and real records.
How much does an AI consultant cost per hour?
Clutch's pricing guide puts the average for AI consulting firms at $200 to $500 per hour, and the firms we checked in its directory listed $25 to $199 per hour, on October 30, 2026. We do not bill hourly: the strategy call and first audit are free, consulting is quoted for your business before it starts, and builds get a fixed price after the audit.
Is AI consulting worth it for a small business?
It is when you have more ideas than budget and need to know which one pays back first. A roadmap that costs more than the first build rarely makes sense for a small company, so we size the consulting scope to the business. If you already know the one task you want automated, skip consulting and get a build quote.
What is an AI readiness assessment?
An AI readiness assessment checks whether a process, and the data behind it, can support AI today. Ours scores each process on volume, rule clarity, data access, error cost and compliance exposure, out of 15. A low score usually points to a data or rules fix that should happen before any AI is added.
Does the EU AI Act apply to a US company?
It can. Article 2 of the Act covers providers and deployers outside the EU when the output of the AI system is used in the EU. If you sell to EU customers with an AI chat or voice agent, check the transparency duty in Article 50 and the AI literacy duty in Article 4, and get legal advice for your case.
Will AI replace my staff?
In the roadmaps we write, AI takes over repetitive steps such as first replies, data entry and reminders, and people keep the decisions and difficult conversations. Our roadmap covers what changes for a team in practice, which tasks move to the system first, and how to plan the change with the people who do the work today.
Can you work with the AI tools we already pay for?
Yes. The audit starts with what you already have, and the roadmap reuses it where it fits: an existing CRM, a Zapier or Make account, a model provider you have a contract with. Replacing a tool only makes the plan when the current one cannot do the job or costs more than an alternative at your volume.
What do we need to prepare for the audit?
Bring rough monthly volumes for the tasks that take the most time, a list of the tools you pay for, and two or three real examples of the work, such as an inquiry, an invoice or a report. Read-only access to the CRM helps. Nothing needs to be cleaned up before the call.
Do you only work with small businesses?
No. We work with small firms and large ones, and the method is the same: score the processes, build the highest scorer first, keep a person on anything risky. A larger company usually has more processes to score and more systems to connect, so the roadmap is longer and the builds are quoted separately.
How do I get started?
Book a free strategy call, or send us a short description of your situation. We confirm whether a roadmap is worth doing for you or whether a single build is the faster route, and either way you get a written answer with a fixed price for any build we recommend.
Sources and further reading
- NIST, AI Risk Management Framework (read October 30, 2026)
- NIST AI Resource Center, AI RMF Core: Govern, Map, Measure, Manage
- European Commission, AI Act regulatory framework and timeline
- AI Act Service Desk, Article 2 (scope)
- AI Act Service Desk, Article 4 (AI literacy)
- AI Act Service Desk, Article 50 (transparency)
- FTC, Operation AI Comply press release, September 25, 2024
- OpenAI API documentation, your data
- Clutch, AI consulting pricing guide and AI consulting directory
