How I am using AI for my consultancy job in 2026

AI is now part of how I work as a consultant every day. Not as a magic layer that replaces thinking, and not as a shiny demo I show clients for five minutes. I use it as a practical operating system for research, discovery, writing, automation, decision support, and delivery.

The important part is this: I do not use AI to skip the work. I use it to compress the slow parts of the work so I can spend more time on judgment, context, and client outcomes.

That distinction matters. In 2026, almost every consultant has access to the same models. The advantage is no longer “I use ChatGPT.” The advantage is knowing where AI actually improves a consulting workflow, where it creates risk, and how to turn it into repeatable systems instead of random prompts.

This is how I am using AI for my consultancy job in 2026, what is working, what I avoid, and the practical tips I would give to any consultant who wants to use AI without sounding like they outsourced their brain.

Key takeaways

• I use AI mostly for leverage: faster research, clearer synthesis, better discovery notes, reusable templates, and small internal automations.
• The best results come from giving AI context, constraints, examples, and a clear definition of what “good” looks like.
• I avoid using AI as the final decision-maker. It can challenge my thinking, but it does not own the recommendation.
• The most valuable AI work for clients is not a prompt. It is a mapped workflow, clean inputs, sensible guardrails, and a measurable business outcome.
• The consultants who win with AI in 2026 will be the ones who turn it into a repeatable delivery system, not the ones who collect the most tools.

Why AI matters for consultants now

AI adoption is no longer theoretical. McKinsey reported in its 2025 State of AI research that 88% of organizations use AI in at least one business function, while only a small minority have scaled it across the whole organization. Stanford’s 2025 AI Index also shows how quickly business adoption accelerated, with corporate AI investment reaching $252.3 billion in 2024.

That gap is where consultants can create real value.

Most companies are not asking, “Can AI do something impressive?” They are asking more practical questions:

• Which process should we automate first?
• What data do we need?
• How do we avoid exposing sensitive information?
• What does a useful AI pilot look like?
• How do we turn experimentation into measurable ROI?

Those are consulting questions before they are technical questions. AI makes the work faster, but it does not remove the need for diagnosis, prioritization, stakeholder alignment, and delivery discipline.

For my own work, the biggest shift has been moving from “AI as a tool I open sometimes” to “AI as a layer inside my consulting process.” I use it before calls, during analysis, after workshops, and when turning messy notes into clear next steps.

My principle: AI is a junior team, not a senior consultant

The easiest way to make AI useful is to give it the right job title.

I do not treat AI as the expert. I treat it as a fast junior team: a researcher, analyst, note-taker, editor, QA assistant, and automation builder. It can produce a lot quickly, but it needs direction, review, and correction.

That mindset keeps the quality high. If I ask AI to “write a strategy,” I usually get generic advice. If I ask it to compare three workflow options against specific constraints, identify assumptions, and draft questions for the client, the output becomes much more useful.

Here is the rule I use:

AI can accelerate the work, but I am still accountable for the recommendation.

That means I check the logic, verify claims, remove generic language, and add the parts AI cannot know: business context, client politics, implementation constraints, budget reality, and the difference between what looks good in a document and what will survive Monday morning.

1. I use AI before client calls to build context fast

Before a discovery call, I use AI to prepare. Not to pretend I know the client’s business better than they do, but to avoid wasting time on surface-level questions.

My pre-call workflow usually looks like this:

1. I collect the company website, LinkedIn profile, product pages, public case studies, and any notes from previous conversations.
2. I ask AI to summarize the business model, likely customer segments, visible operational bottlenecks, and possible AI use cases.
3. I ask it to generate hypotheses, not conclusions.
4. I turn those hypotheses into sharper discovery questions.

For example, instead of asking, “Do you want to use AI in customer support?” I might ask:

• “Where does the support team lose the most time: triage, response drafting, knowledge retrieval, or escalation?”
• “How many requests are repetitive enough that a well-designed assistant could handle the first pass?”
• “What would make an AI support workflow unacceptable: wrong tone, wrong answer, slow approval, compliance risk?”

That is a much better conversation.

AI helps me arrive prepared, but the call is still about listening. The client will always reveal context that no model could infer from public information.

Tip: Ask AI for “questions that would disprove this hypothesis.” This prevents you from walking into a call already in love with your own idea.

[ADD YOUR EXAMPLE: describe a recent discovery call where AI helped you prepare better questions.]

2. I use AI to turn messy discovery into usable strategy

Consulting work produces a lot of messy information: call transcripts, workshop notes, stakeholder comments, process screenshots, spreadsheets, and half-finished requirements.

AI is excellent at helping me organize that mess.

After a workshop or discovery session, I use AI to extract:

• Pain points mentioned by the client
• Repeated themes across stakeholders
• Open questions
• Decisions already made
• Assumptions that still need validation
• Potential quick wins
• Risks and dependencies

The goal is not to create a beautiful summary. The goal is to preserve the truth of the conversation and convert it into action.

One of my favorite prompts is:

“Separate facts, interpretations, assumptions, and recommended next steps. Do not mix them.”

That simple instruction improves the output immediately. A lot of consulting mistakes happen when assumptions are written as facts. AI can make that worse if you are not careful, because it tends to sound confident even when the input is incomplete.

So I force separation. Facts are things the client actually said or documents actually show. Interpretations are my reading of the situation. Assumptions are things we need to check. Next steps are actions.

This makes client communication clearer and reduces the risk of overpromising.

3. I use AI to map workflows before proposing automation

A lot of companies come to AI with a tool-first mindset. They ask for an agent, a chatbot, or an automation before the workflow is clear.

I usually do the opposite. I use AI to help map the current process first.

A typical workflow mapping exercise includes:

• Trigger: what starts the process?
• Inputs: what information is needed?
• Steps: what happens in order?
• Decisions: where does a human choose between options?
• Exceptions: what breaks the normal flow?
• Outputs: what should be produced?
• Systems: where does data live?
• Owners: who is responsible?
• Metrics: how do we know it improved?

Once that map exists, AI becomes much easier to apply. You can see which parts are suitable for automation, which require human review, and which should not be automated yet.

For example, an AI agent might be useful for drafting a first response to inbound requests. But if the company has no consistent categories, no up-to-date knowledge base, and no owner for approvals, the agent will only expose the chaos faster.

That is why I often tell clients: before building AI, map the work.

This connects directly to my AI readiness checklist. Readiness is not about buying the newest tool. It is about having enough clarity for AI to produce reliable outcomes.

Tip: If a team cannot explain a process clearly in plain English, do not automate it yet. Use AI to help document it first.

[ADD YOUR EXAMPLE: insert one process you often map for clients, such as lead qualification, document processing, support triage, reporting, or internal knowledge search.]

4. I use AI as a writing partner, not a ghostwriter

I write a lot: proposals, scopes of work, strategy documents, implementation plans, workshop recaps, emails, and articles like this one.

AI helps, but I rarely use a first draft as-is. The first draft usually sounds too smooth, too generic, or too much like everyone else on the internet. So I use AI in smaller, more controlled ways.

Here is what works better for me:

• Ask AI to create a structure before writing the content.
• Give it examples of my tone and tell it what to avoid.
• Ask for sharper transitions, not full rewrites.
• Use it to simplify dense paragraphs.
• Ask it to identify weak claims or unsupported statements.
• Ask it to create alternative headlines and meta descriptions.

When I want something to sound like me, I add constraints:

• Use short paragraphs.
• Avoid hype.
• Keep the tone direct and practical.
• Do not use phrases like “game changer,” “unlock potential,” or “revolutionize.”
• Include trade-offs and caveats.
• Prefer examples over abstract claims.

That last point is the most important. AI writing sounds like AI writing when it has no lived detail. A small real example is worth more than ten polished paragraphs.

Tip: After AI drafts something, add one specific moment only you could know: what the client said, what broke, what surprised you, what you changed, or what the result was.

[ADD YOUR EXAMPLE: add a short personal anecdote about a proposal, workshop, or automation project.]

5. I use AI to build small internal automations

Not every AI use case needs to become a big client-facing product. Some of the highest ROI comes from small automations inside the consulting workflow.

Examples I use or would consider using:

• Converting call transcripts into structured action items
• Turning workshop notes into a client recap
• Drafting a first version of a project plan from a discovery document
• Classifying inbound requests by priority and topic
• Creating reusable checklists for AI readiness, process mapping, or automation fit
• Generating test cases for an AI workflow before launch
• Monitoring recurring tasks and flagging missing information

These are not glamorous. That is why they work.

They remove friction from repeated tasks, and repeated tasks are where consultants lose time. Saving 20 minutes on one task does not sound impressive. Saving 20 minutes across 5 tasks per week for 40 working weeks is more than 66 hours per year. Multiply that across a small team, and suddenly the automation is not small anymore.

The key is to start with a task you already understand. If you cannot manually do it well, AI will not magically make it reliable.

Tip: Keep a “friction log” for one week. Every time you repeat a task, copy-paste the same information, rewrite the same email, or reformat the same document, write it down. That list becomes your AI automation backlog.

6. I use AI to improve client deliverables before sending them

Before I send an important deliverable, I use AI as a reviewer.

I do not ask, “Is this good?” That question is too vague. I ask targeted review questions:

• “What claims are unsupported?”
• “Where would a CFO push back?”
• “Which recommendation is unclear?”
• “What would be hard for the client to implement next week?”
• “What assumptions should be made explicit?”
• “Which section feels generic?”
• “What would a skeptical operations manager ask?”

This is where AI is genuinely useful. It can simulate different reader perspectives and help me find gaps before the client does.

But again, I make the final call. Sometimes AI suggests changes that would make the document more complete but less useful. A 40-page strategy deck can look impressive and still fail if nobody knows what to do next.

My preferred deliverables are practical: clear diagnosis, prioritized opportunities, estimated effort, expected impact, risks, and next steps.

In other words: less theater, more action.

7. I use AI to test ideas before recommending them

One of the most useful things AI can do is argue with me.

When I have a recommendation, I ask AI to challenge it:

• “What are three reasons this recommendation might fail?”
• “What would make this too expensive?”
• “What data would change the decision?”
• “What is the simplest version of this idea?”
• “What is the riskiest assumption?”
• “What would I recommend if the client had half the budget?”

This helps me avoid falling in love with the most elegant solution. Consulting is not about elegance. It is about fit.

Sometimes the best AI project is not an agent. It might be a better intake form, a cleaner knowledge base, a workflow redesign, or a small automation that removes two hours of admin work per week.

AI helps me explore those options faster, but the decision still depends on business context.

Tip: Always ask for the “boring version” of your AI idea. If the boring version creates value, the advanced version might be worth exploring. If the boring version creates no value, the advanced version is probably just decoration.

8. I use AI carefully with confidential client data

This is one area where I try to be conservative.

Client data is not prompt fuel. Before putting any information into an AI system, I think about confidentiality, consent, data processing, and whether the input is actually needed.

My default rules are:

• Do not paste sensitive client data into tools without approval.
• Anonymize examples where possible.
• Use synthetic or redacted data for early exploration.
• Keep human review in workflows that affect customers, money, legal obligations, or reputation.
• Document what the AI system can access and what it cannot.

This is not just compliance. It is trust.

The OpenAI 2025 enterprise AI report shows how quickly AI usage is growing inside businesses, with workplace usage and structured workflows increasing rapidly. That growth makes governance more important, not less.

If a client wants to use AI seriously, we need to answer basic questions:

• What data is being processed?
• Where is it stored?
• Who can access the output?
• What happens when the AI is wrong?
• Who approves the final decision?

Those questions are not blockers. They are what make AI usable in real businesses.

My practical AI stack for consulting work

The specific tools change, but the categories stay fairly stable.

I usually think in terms of capabilities:

• Research and synthesis: tools for summarizing documents, comparing sources, and extracting themes.
• Writing and editing: tools for outlines, rewrites, proposals, emails, and content drafts.
• Workflow automation: tools that connect forms, CRMs, spreadsheets, email, Slack, Notion, or internal databases.
• AI agents: systems that can follow multi-step processes with tool access and guardrails.
• Knowledge management: searchable internal notes, project history, reusable templates, and decision logs.
• Development support: coding assistants for prototypes, scripts, integrations, and validation.

[ADD YOUR TOOL STACK: list the actual tools you use, for example ChatGPT, Claude, Cursor, Notion, Make, Zapier, Airtable, Google Workspace, Slack, HubSpot, custom scripts, or internal agents.]

My advice is not to obsess over the stack. Tools matter, but workflows matter more. A simple setup used every day beats a complex AI stack nobody trusts.

A simple AI workflow any consultant can copy

If you want to start using AI in consulting work, do not begin with 20 tools. Start with one repeatable workflow.

Here is a practical one:

1. Before the call: ask AI to create a short briefing and discovery questions from public information and previous notes.
2. During the call: capture notes or a transcript, with consent where required.
3. After the call: ask AI to extract facts, assumptions, risks, decisions, and next steps.
4. Before sending: use AI to review the recap from the perspective of the client.
5. After delivery: save the useful prompts, templates, and lessons into your own knowledge base.

Do this for 10 client interactions and you will start seeing patterns. Those patterns become your consulting system.

That is the real benefit of AI: not one impressive output, but compounding improvement in how you prepare, think, write, and deliver.

Common mistakes I see consultants make with AI

The first mistake is using AI only for content generation. Writing is useful, but it is not the whole game. AI is often more valuable for analysis, preparation, review, and workflow design.

The second mistake is accepting generic output. If the answer could apply to any company, it is not ready. Add context, constraints, examples, and your own judgment.

The third mistake is skipping verification. AI can sound confident while being wrong. For client work, especially around numbers, compliance, strategy, or technical feasibility, verification is not optional.

The fourth mistake is automating before understanding. If the process is unclear, AI will not clarify it for you unless you explicitly make process mapping part of the work.

The fifth mistake is hiding AI usage instead of creating a professional standard around it. Clients do not need every prompt, but they do deserve quality, confidentiality, and accountability.

FAQ: using AI as a consultant in 2026

Will AI replace consultants?

AI will replace some low-value consulting tasks, especially generic research, basic summaries, and template-heavy documents. But it will not replace consultants who can diagnose problems, manage stakeholders, design workflows, and take responsibility for outcomes. The consultant’s value moves from producing information to creating clarity and execution.

What is the best AI use case for consultants?

The best first use case is usually discovery support: preparing better questions, summarizing calls, extracting action items, and turning messy notes into clear next steps. It is low risk, immediately useful, and improves the quality of client communication without requiring a complex technical setup.

How can consultants use AI without sounding generic?

Use AI for structure, options, critique, and editing, but add lived details yourself. Include real examples, constraints, client language, numbers, trade-offs, and personal observations. Generic AI content usually lacks specific context. Your job is to add the context AI cannot know.

Should consultants tell clients they use AI?

It depends on the work and the agreement, but I prefer transparency around process and strictness around data. Clients care less about whether AI helped draft a recap and more about whether their data is protected, the output is accurate, and the recommendation is accountable.

What should consultants avoid putting into AI tools?

Avoid sensitive client information, personal data, confidential strategy, financial records, credentials, proprietary documents, or anything you are not authorized to process. When in doubt, anonymize the input, use synthetic examples, or ask for approval before using an AI tool.

Final thought: AI makes good consultants sharper

AI has changed how I work as a consultant, but not in the way people often describe.

It has not removed the need to think. It has made weak thinking easier to spot. It has not made every process worth automating. It has made process clarity more important. It has not replaced client trust. It has made trust, judgment, and accountability more valuable.

In 2026, using AI for consultancy work is not about having the perfect prompt library. It is about building a better way to prepare, analyze, communicate, and deliver.

My rule is simple: if AI helps me give the client a clearer answer, faster, with better evidence and less busywork, I use it. If it makes the work vaguer, riskier, or more generic, I do not.

That is the line I try to hold.

If you are exploring how AI could improve your own consulting workflow or business operations, start with the basics: map the process, identify the bottleneck, check the data, and define success. Or start with a structured AI Discovery Sprint and turn the idea into a practical plan.

Link recommendations

Internal links used:

• AI readiness checklist - supports the point that AI projects need workflow and data readiness.
• AI Discovery Sprint - connects the article to a practical next step for prospective clients.

External sources used:

• McKinsey: The State of AI in 2025
• Stanford HAI: 2025 AI Index Report
• OpenAI: The state of enterprise AI 2025

Notes for Davide before publishing

• Replace the placeholders marked [ADD YOUR EXAMPLE] with 2-4 sentence real examples.
• Add your actual AI tool stack in the tool stack section.
• If you have numbers, add them: hours saved, proposals reviewed, workflows mapped, automations shipped, or client outcomes.
• Consider adding one original screenshot or diagram, for example “my AI consulting workflow” or “from messy notes to client action plan.”