Otto

A Consultant's AI Assistant: What It Should Actually Do

A Consultant’s AI Assistant: What It Should Actually Do

Consultant recording a client follow-up

A good AI assistant for a consultant does not write your deliverables. It keeps track of everything you have told clients you would do, across email, calendar and calls, and makes sure none of it slips. That is where independent consultants and small agency owners lose money and trust: not in the analysis, but in the follow-up that never went out.

This guide explains what to expect from an AI assistant in consulting work, where general chatbots and single-purpose tools fall short, how to judge one, and how to test it in a week.

Otto
ottohq.app
Keep client follow-ups in view
Otto brings meeting, email and calendar commitments into one shared memory, then drafts follow-ups for your approval.
Explore Otto

TL;DR

  • Consulting runs on promises made in meetings, emails and hallway chats. The assistant that matters is the one that captures all of them in one place.
  • A chatbot that forgets everything between sessions is a research tool, not an assistant. Look for persistent memory across your inbox, calendar and calls.
  • Client confidentiality comes first. Check how data is handled before you connect anything.
  • Nothing should go to a client without your approval. Drafts, not autopilot.
  • Start with one client and one week, and judge the tool by how many follow-ups it caught that you would have missed.

Table of Contents

Why consultants need an assistant, not another chatbot

Most consultants already use AI for writing and research. That helps with the deliverable, but it leaves the real operational problem untouched: you are running five engagements, each with its own thread of emails, calls and small commitments.

Think about a normal Tuesday. You tell a client on a call that you will send the revised scope by Thursday. A different client emails asking for an intro to your contact at a supplier. Your calendar shows a workshop tomorrow that needs a pre-read. None of these live in the same tool. So you carry them in your head, and you become what we call the human glue between your own apps.

Business Insider’s interviews with AI leaders at top consulting firms point the same way. They describe the shift as moving away from using AI as a search engine for one-off lookups and toward persistent agents that keep ongoing work moving. The idea is that AI handles the routine administrative bulk so the consultant can spend time on judgment and strategy.

That applies even more to a solo practitioner. A large firm has coordinators, project managers and assistants. You have an inbox.

Why consultants need an assistant, not another chatbot — overview diagram

What the big firms are doing, and what it means for you

The large firms have already built this into daily work. A 2025 report says that McKinsey’s internal assistant is used by more than 70% of staff, at roughly 17 queries a week, for summarizing documents, research and drafting proposals. Bloomberg reported that McKinsey consultants are drafting proposals and slides with the firm’s own platform.

Two lessons carry over to a small practice.

First, the assistant augments the consultant. The same coverage describes time saved being reinvested in client advisory and complex analysis, not in cutting the consultant out.

Second, those firms insist on controlled platforms for confidential material. Reporting on the firm’s approach says the point of using a proprietary system is that client data is not used to train third-party models. You do not have a proprietary platform, but you can ask the same questions of any tool you adopt.

The five jobs a consultant’s AI assistant should cover

Ignore feature lists for a moment. Judge any tool against the work you actually do.

1. Capturing commitments wherever they are made

This is the core job. A promise made on a call, in an email or in a calendar invite description should land in one list, with the client, the date and the context attached. If the tool only reads email, it never sees what you said aloud. If it only transcribes meetings, it never sees the email that changed the deadline.

We built Otto around this gap. It listens in meetings, reads your email and calendar, and records each promise, whether you made it or someone made it to you, in a single ledger that it keeps chasing until the item is closed. For how this differs from a basic to-do list, see our piece on reminders versus tasks for founders.

2. Follow-ups that actually go out

Consultants rarely lose clients because of the work. They lose them because a proposal sat for nine days, or a promised summary arrived late. An assistant should notice when a thread has gone quiet, when a deadline is close, and when someone owes you a reply. Our guide to capturing, queueing and acting on follow-ups walks through the routine.

3. Email drafting in your voice, with you in control

Drafting is useful only if it saves you the context switch, and only if nothing is sent without you. A draft that references the right meeting, the right deliverable and the right date is worth ten generic replies. More on this in AI email drafting for busy professionals.

4. Meeting summaries that become tasks

A summary nobody acts on is just a longer note. The useful version maps what was said to owners and due dates. We cover how in meeting summaries that map notes to tasks.

5. A morning answer to “what should I focus on”

With several clients, priorities shift daily. An assistant that sees your inbox, calendar and open commitments together can rank them. One that sees only one of those cannot.

Where single-purpose tools fall short

Most AI tools are clever inside their own box:

  • Meeting notetakers have the call but have never seen your inbox. They capture a promise but not the email that revised it.
  • Email assistants have your inbox but do not know what you agreed out loud.
  • Contact managers know your clients but not your schedule.
  • General chat assistants are strong for brainstorming and drafting but start every session cold unless you paste in context.

The result is that you do the integration yourself, copying context between tools. That is the job to remove. If you use an AI notetaker and still lose follow-ups, this is almost certainly why: the notes exist, but they are not connected to anything that chases them.

A quick comparison by type of tool

Type of tool Strength Blind spot Good fit for
General chat assistant Research, drafting, analysis No memory of your work unless you paste it in Thinking and writing
Suite-bundled assistant Deep ties to the documents and mail in one ecosystem Weak on anything outside that suite, including calls Firms standardized on one suite
Meeting notetaker Accurate transcripts and summaries Does not see email or calendar changes Interview-heavy work
Email assistant Triage and drafting Does not know what you promised on calls High-volume inboxes
Shared-memory chief of staff One ledger across email, calendar and meetings Needs access to several sources, so review security Anyone juggling multiple clients

Pricing for the first two categories is fairly uniform. According to one 2026 comparison, flagship individual plans cluster around $20 a month, and bundled suite plans such as the $19.99 tiers add storage and office apps. Choosing between them is mostly a question of where your files live. That is a different decision from choosing the tool that tracks your commitments, and many consultants end up with one of each.

Confidentiality: the question to ask first

Client work is confidential, often under a signed NDA. Before connecting any assistant to your inbox, answer these:

  1. Is your data used to train models? Get this in writing from the vendor.
  2. What can the tool do on its own? Anything that sends, deletes or schedules should need your approval.
  3. Can you see and correct what it stored about a client?
  4. Does your client contract restrict where their information can be processed?

Read the vendor’s security page, not only the marketing. Ours is at Otto security. If a client contract is strict, ask the client before connecting their threads, or keep that engagement out of the tool.

The approval point matters beyond security. We wrote about when an assistant should ask first, and our view is simple: a tool that sends on your behalf without asking will eventually send the wrong thing to the wrong client.

Our perspective: remove the glue job

We think the most expensive thing in a consulting practice is not software. It is the consultant acting as the integration layer between tools that cannot see each other.

When we started building Otto, the pattern was the same in every conversation with founders, operators and chiefs of staff. They had a notetaker, an email assistant, a task app and a CRM. Each worked. Yet they still missed things, because nothing held the whole picture. The notetaker had the promise, the inbox had the changed deadline, the calendar had the conflict, and the person was the only place all three met.

Our perspective: remove the glue job — overview diagram

So we made one shared memory the foundation, not a feature. Email, calendar, meetings and commitments sit in the same place, so nothing has to be re-explained. Every promise goes into one ledger, and the ledger is chased until closed. And because it sees everything at once, it can answer the morning question: what should I focus on today.

We also hold a firm line on control. Otto drafts; you decide. Nothing is sent or finalized without your explicit approval. For consultants, whose reputation is the product, that is not a limitation. It is the design.

— Eddie

A one-week test you can run with any tool

Do not evaluate an assistant by its demo. Evaluate it against your real week.

  1. Pick one active client. Connect only that engagement’s email and calendar first.
  2. Run it through three meetings. After each, check whether every promise you made was captured, including the small ones.
  3. Log misses. Write down any commitment you noticed that the tool did not. Misses reveal the blind spots.
  4. Check the drafts. Do they cite the right context, or do you have to rewrite them from scratch?
  5. Count follow-ups caught. At the end of the week, how many items would you have forgotten or sent late?
  6. Decide on one number. If it caught even a couple of things that would have cost you a client’s confidence, it has paid for itself.

If you want to test memory specifically, our two-minute second-brain test shows what a tool actually remembers.

Setting it up for several clients

Consultants and agency owners need structure on top of the capture. A few practices that work:

  • Organize by contact. Tasks tied to the person who requested them are easier to review before a call. See tasks by contact for a five-step approach.
  • Keep your task app in sync. If you plan in Notion, decide which system is the source of truth. Our note on syncing tasks to Notion covers the tradeoffs.
  • Make cross-app workflows safe to repeat. If an automation runs twice, it should not create duplicates or send twice. Read why idempotency comes before AI agents.
  • Review weekly. Spend fifteen minutes on Friday going through open commitments by client. Close what is done and renegotiate what is not.

For executives and chiefs of staff

The same logic holds if you track other people’s commitments. An executive assistant or chief of staff needs to see what the executive promised, what was promised to them, and what is overdue, without reading every thread. The ledger approach gives you one place to look.

For founders and small business owners

If you run a small company, you are your own consultant, project manager and account lead. The priority is the same: capture promises once, chase them automatically, and keep a human in charge of anything outbound.

Trying Otto

Otto is an AI chief of staff that works from one shared memory of your email, calendar, meetings and commitments. It integrates with over 130 tools, including Gmail, Google Calendar, Slack, Notion, GitHub and Todoist.

Plans are published on our site: Free at $0 per month, Upgraded at $20 per month and Max at $50 per month. Higher plans add higher usage limits, integrations and advanced features. Start on Free with one client and run the week-long test above.

FAQ

What is the best AI assistant for a solo consultant?

The best one is the one that sees your whole workflow: email, calendar and calls. A chatbot helps with writing and research, but it will not track what you promised a client. Look for a tool with shared memory, a commitment ledger and approval before anything is sent.

Is there a free AI assistant I can try?

Yes. Many assistants have a free tier, including ours at $0 per month. A free plan is enough to run a one-client, one-week test. Upgrade only if you hit usage limits or need more integrations.

Is it safe to connect client email to an AI assistant?

It can be, if you check the vendor’s security practices first. Confirm how data is stored, whether it is used for training, and what actions need your approval. If an NDA or client contract restricts processing, ask the client or leave that engagement out.

Will an AI assistant replace a consultant?

No. The firms leading on AI describe it as augmenting consultants, with saved time going back into advisory and analysis. An assistant handles tracking and administration so you spend more time on judgment.

Sources