Once set up, most solo consultants can realistically save 3 to 8 hours in a 40- to 45-hour week, mainly on proposals, pre-call research, meeting notes, email and first drafts of reports. Expect under 2 hours if your work is mostly live workshops and calls, and more than 8 only if you produce a lot of written deliverables. Setup takes 10 to 20 hours.
Those figures are estimates built task by task, with checking time included, not survey results. The best controlled evidence points the same way and adds a warning. In a 2023 field experiment by Harvard Business School and BCG with 758 consultants using GPT-4, those with AI completed 12.2% more tasks, 25.1% faster and at over 40% higher quality on tasks within the AI's capabilities. On a task deliberately chosen to sit outside them, consultants using AI were 19 percentage points less likely to reach the correct answer than those without it. AI saves time on some work and quietly makes other work worse, and knowing which is which is most of the skill.
Where the hours come from: a task-by-task estimate
This assumes a solo consultant with about 15 hours of client delivery a week and the rest spent on everything around it. The savings are net of the time spent checking AI output. All figures are illustrative.
| Task | Hours a week without AI | Realistic saving | What pushes it up or down |
|---|---|---|---|
| Proposals and scoping | 3 | 0.5 to 1.5 | Higher if you reuse a strong template and past proposals |
| Research before calls | 2 | 0.5 to 1 | Lower if your clients are small firms with little online |
| Meeting notes and follow-ups | 2.5 | 0.5 to 1.25 | Higher with a note taker; zero if clients won't be recorded |
| 5 | 0.5 to 1 | Most time is deciding, not typing | |
| First drafts of reports and decks | 6 | 1 to 1.75 | Higher for structured reports; lower for original analysis |
| Data analysis | 3 | 0 to 1 | Can go negative if errors slip through (see below) |
| Invoicing and admin | 2 | 0 to 0.25 | Mostly handled by accounting software already |
| Marketing posts and newsletters | 2 | 0.25 to 0.5 | Your own stories still take your own time |
| Total | 25.5 | About 3 to 8 |
Notice what isn't on the list: client delivery itself. The workshops, the conversations and the judgement you're hired for mostly stay the same length. That's why consultants whose week is heavy on delivery see small numbers.
Three consultants, three different weeks
A hospitality consultant who writes a lot: about 7 hours
This consultant advises boutique hotels and guest houses on operations and revenue. Each engagement ends in a 20- to 30-page report with recommendations, and she writes two or three proposals a week. Her writing-heavy week has about 14 hours of drafting and analysis. With a reference file of past reports (anonymised), a structured outline prompt and a note taker for interviews, she saves roughly 7 hours. The largest single saving is turning interview transcripts into the "findings" section, which used to take a full day per report.
A facilitator for tour operators: about 2 hours
This consultant runs strategy away-days and planning workshops for small tour operators. Most of his week is in the room or preparing to be. AI helps with workshop materials, the follow-up summary and proposals, and saves him around 2 hours. He tried using it to design workshop exercises and found the results generic; he now uses it only to reformat his own designs into handouts.
An operations consultant for letting agencies: about 4.5 hours
This consultant audits processes: how an agency handles tenant enquiries, maintenance jobs and renewals. Her work mixes interviews, spreadsheets of job logs and a short recommendations deck. AI saves time on interview notes, the deck and research on software options, about 4.5 hours in total, but she stopped using it for the job-log analysis after the mistake described below.
If you'd like to size your own week this way before buying anything, how much time AI can really save a small business each week has the same method for teams.
What the savings look like task by task
A proposal first draft from call notes
The hospitality consultant pastes her discovery-call notes and her proposal template and asks for a first draft. Sample prompt and output extract (illustrative):
Prompt:
Using my template and these call notes, draft the scope, approach and
timeline sections of a proposal for a 12-room boutique hotel. Keep my
headings. Only include work mentioned in the notes. Flag anything
you're unsure about with [CHECK].
Output extract:
"Scope: a review of the hotel's pricing, channel mix and direct
booking performance, plus a half-day staff training session on
upselling at check-in.
Timeline: the review will be completed within two weeks of the
kick-off meeting."Two fixes. The staff training session was never discussed; the model added it because it often appears in hospitality proposals, and she doesn't offer training. And two weeks is optimistic, because the hotel's channel manager data takes a week to extract; she changes it to four. The rest of the draft is sound and saves her about 45 minutes. The full method is in how consultants use AI to write proposals in under an hour.
Meeting notes into a follow-up, before and after
Before (her own notes, typed after the call):
"- occ down Q3, owner thinks OTA fees
- wants direct bookings up
- send data request by Thurs, check PMS export poss?"
After (AI draft from the transcript, edited, sent):
"Thanks for your time today. To confirm what we agreed:
1. I'll send the data request by Thursday, including the booking
export from your property management system.
2. You'll check whether your system can export bookings by channel.
3. Our first review call is pencilled for 14 October.
You mentioned that occupancy dipped in the third quarter and you
suspect commission costs are part of it; that'll be the first thing
we look at."The follow-up used to take 20 minutes and often slipped to the next day. Now it takes five and goes out within the hour, which clients notice. She still checks every date and commitment against her own memory of the call.
Some clients won't be recorded, and in a few sectors that's most of them: a family-owned business discussing succession, or anyone mid-dispute. The saving doesn't disappear, it shrinks. Straight after the call, walk to the kettle and record a three-minute voice memo on your phone: who was there, what was agreed, what you promised and by when. Paste the transcript of your own memo into the assistant with the same follow-up prompt. The draft is thinner than one built from a full transcript, because it only knows what you remembered, but it still turns 20 minutes of typing into about eight minutes of talking and editing. Nothing the client said verbatim ever leaves the room.
Pre-call research on a prospect
The letting-agency consultant has a discovery call tomorrow with a three-branch agency. She asks an assistant with web search to summarise the business: services, size, recent changes, what tenants and landlords say in public reviews, and any obvious operational pain points. An extract from an illustrative answer:
"Founded in 2009, the agency operates three branches and manages
around 900 properties. It recently launched an online maintenance
reporting portal. Reviews frequently mention slow responses to
repair requests, with several tenants describing waits of weeks.
Likely pain points: maintenance coordination, communication
during repairs, staff capacity at the busiest branch."Three things to check before the call. The founding year and "around 900 properties" came with no source link, and the agency's own website says nothing about either, so she drops both rather than quote them back to the owner. The maintenance portal is real, but the page announcing it is two years old, which changes the question from "how's the new portal going?" to "is the portal being used?". The review theme holds up when she reads a dozen reviews herself. Twenty minutes of research became about eight, and the one insight worth taking into the call, repairs, was one she verified.
Email: a few minutes back, not an hour
The table allows only 0.5 to 1 hour a week for email because most of the time goes on deciding what to say. The saving shows up on long, many-part emails. A client writes 400 words mixing four questions, a change of meeting date and a worry about the budget. Asking the assistant to "list every question and request in this email, then draft a reply that answers each one in order, leaving [DECIDE] where I need to choose" turns ten minutes of rereading into a checklist and a skeleton reply. The budget worry gets a [DECIDE] and she writes that paragraph herself. A short, delicate email, such as telling a client their project is running over, saves nothing: the thinking is the whole job, and a drafted version is harder to make sound like her than a blank page.
The analysis mistake that cost more time than it saved
The letting-agency consultant asked an assistant to summarise twelve months of an agency's maintenance-job logs: average days to close a job, by month. The output looked tidy and showed a clear improvement over the year. It was wrong: the model had averaged the monthly averages, giving a quiet month with 11 jobs the same weight as a busy month with 140. Recalculated properly, there was barely any improvement at all. She caught it only because the headline didn't match what staff had told her in interviews. Finding and fixing it took longer than doing the analysis herself would have, and she nearly presented a false conclusion to a client.
This is the "outside the frontier" problem from the study, at small-business scale. Her rule now: AI can write the formulas or code for an analysis, but she runs it herself in a spreadsheet and checks the totals against the raw data.
For a solo consultant, the tasks most likely to sit outside the frontier are predictable enough to list:
- Calculations on real data, especially weighted averages, percentages of percentages and anything across several sheets.
- Current or niche facts about a client's market, software prices or regulations, which the model may state from out-of-date training.
- Judgements about people and politics inside a client's business, where the transcript doesn't capture who holds the real influence.
- Genuinely new recommendations. Models are good at the usual answer; clients often hire a consultant for the unusual one.
Use AI around these tasks, to structure, draft and check wording, but do the core of them yourself.
The hidden costs that eat into the saving
- Checking. Every draft needs reading with care. The estimates above include this; many claims online don't.
- Prompt fiddling. The first month is slower, not faster, while you work out what to ask. Budget 10 to 20 hours for setup: templates, reference files, a note-taker routine.
- Tool sprawl. Three assistants, two note takers and a research tool means time spent switching and copying. Most solo consultants need one chat assistant and one note taker; the AI tool stack a one-person consultancy actually needs sets out a lean version.
- Subscriptions. ChatGPT Plus or Claude Pro is $20 a month. Business plans that don't train on your content by default need at least two seats, so a sole trader pays $40 to $50 a month for one, or stays on an individual plan with training switched off in privacy settings and keeps client-identifying details out. A note taker such as Otter Pro adds about $8.33 a month on annual billing.
- Client data. Anonymising interview notes or data before it goes in takes minutes per engagement. Skipping it is a false saving.
Put the setup cost against the saving and it looks small. The hospitality consultant's setup, spread over her first three weeks, went roughly like this:
Setup log (illustrative)
Week 1 Proposal template tidied, 3 past proposals anonymised
and added to a project 3h 30m
Week 1 Note taker installed, consent line added to
booking emails, tested on an internal call 1h 30m
Week 2 Report outline prompt written and tested on
two old engagements 4h 00m
Week 2 Anonymised findings sections added as examples 2h 00m
Week 3 Follow-up prompt, research prompt, fixes 3h 00m
Total 14h 00mAt about 7 hours saved a week once it was working, 14 hours of setup paid for itself in the third week of normal use. For the tour-operator facilitator, saving 2 hours a week, the same investment would take seven weeks, which is why he set up only the follow-up summary and the proposal template, about four hours in all.
Log two weeks to find your own hours-saved figure
Estimates are a starting point; your own figure is what matters for pricing and planning. Keep a simple log for one normal week without changing anything, then a week using AI on your three most promising tasks. Record the time for each instance, including checking. A filled-in extract from the hospitality consultant's second week (illustrative):
| Task | Week 1 (no AI) | Week 2 (with AI, incl. checking) | Notes |
|---|---|---|---|
| Proposal, guest house (8 rooms) | 2h 40m | 1h 35m | Removed an invented deliverable |
| Interview write-up (3 interviews) | 3h 30m | 1h 20m | Transcripts from note taker |
| Follow-up emails (6) | 1h 50m | 40m | Checked every date |
| Occupancy analysis | 1h 30m | 1h 45m | Slower: re-ran the numbers herself |
The last row matters as much as the others. A log that shows where AI costs time is how you learn which tasks to keep it away from. The same method, for a small team, is in how to measure time saved after rolling out AI, and how consultants use AI for client research before discovery calls covers one of the more reliable savings in detail.
What to do with the reclaimed hours
Decide before the hours appear, or they'll dissolve into email. Three sensible uses, in rough order of value:
- Business development. One more discovery call or proposal a week is worth more to most solo consultants than anything else those hours could buy.
- Better delivery. More time thinking about a client's problem, which is what the fee is for.
- A shorter week. A legitimate choice for a one-person business, and one that tends to improve the other two.
A rough sum shows why business development tops the list. Take the letting-agency consultant saving 4.5 hours a week. If she puts two of those hours into one extra discovery call and proposal each fortnight, that's about 23 more proposals over a 46-week year. Suppose one in four turns into work, as her past record suggests, at an average engagement of $5,000: that's five or six extra projects, roughly $25,000 to $30,000 of work, from two hours a week. Your own conversion rate and fee will differ, and the sum only holds if the pipeline of prospects exists. The point is the scale: two hours a week spent winning work can be worth more than every other use of the saved time put together.
If you bill by the day or the hour, faster drafting means smaller invoices unless you price by the project. How to handle that without short-changing yourself or your clients is covered in whether you should charge clients less when AI speeds up your work. Most consultants who move to fixed project fees find the saved hours finally show up where they should: in the margin, and in the diary.
Further reads
- Reusing Past Client Work With AI Without Leaking Client Data — Reuse past deliverables for speed without leaking client data.
- How to Turn Meeting Notes Into Tasks Automatically With AI — Automate the step after the meeting notes.
- How to Catch Made-Up Figures in AI-Drafted Proposals — The checking habit that protects the time you save.
- Claude vs ChatGPT for Business Writing, Proposals and Reports — Which assistant suits report and proposal writing.
- How to Keep Up With AI in 30 Minutes a Week — Stay current without the learning time eating the saving.
- Turning Hours Saved by AI Into Advisory Work Clients Pay For — Turning saved hours into paid work, from the accounting side.
- How Business Coaches Use AI Without Losing the Personal Touch — Where AI belongs in a coaching practice and where it doesn't, with a client memory file, recap before-and-afters and voice rules.
- Is My Business Too Small for AI? What Works for Teams of 1 to 10 — Why headcount is the wrong test for AI, what plans cost for one, three or ten people, and three small teams worked through with numbers.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Dell'Acqua et al., 'Navigating the Jagged Technological Frontier' (Harvard Business School and BCG field experiment with 758 consultants, 2023); OpenAI and Anthropic plan pages; Otter pricing. Weekly estimates and examples are illustrative.