Katie Stone helps tutoring and education businesses improve their systems, including CRM setup and audits, and teaches tutors how to use AI in practical, human-checked ways.
3.4K followersAI for tutorsBased on 20 posts · latest Sep 2026
Katie Stone is most useful for practical content and systems advice on AI and CRM for tutoring businesses.
Best for
An edtech vendor selling AI or systems software to tutoring businesses.
Not for
Mass-reach campaigns, scripted endorsements, or products that hide their prompts.
Key takeaways
A CRM audit found hundreds of lessons labelled lesson; a naming change can save hours.
Claude helped turn survey results into safeguarding reports, cutting 30 to 40 minutes to 15.
A custom GPT once invented notes for a meeting that never happened.
Check first: Ask for recent results from webinars and CRM audit posts, plus audience analytics.
Main ideas
Generic AI tools produce generic work unless you control the prompt · AI is a prediction machine, not a maths machine, so numbers need code or a check · Keep human judgement on the parts that carry context and consequence
Audience
3,424 LinkedIn followers · Nearly half of respondents chose "comfortable, show me advanced" Reported poll result on AI comfort, as described by the creator
Katie Stone helps tutoring and education businesses make their systems easier to use and teaches tutors practical ways to use AI.
Their LinkedIn posts show two strands of this work: hands-on CRM audits and builds, and advice on using AI with human judgment still in control.
Follow them for specific examples of changes in real businesses, from clearer lesson records to faster reports—and candid accounts of what AI gets wrong as well as what it can do.
Stone makes operational advice concrete.
In one post, they describe using Claude to help turn anonymised survey results into safeguarding reports: the work went from 30–40 minutes to 15, while they checked the figures, chose the participant quotes and edited the finished report.
In another, they explain how a CRM audit found hundreds of lessons all labelled “lesson”—a small naming change that could save a team hours of searching.
Their AI advice follows the same practical pattern: tutors need control over prompts, and a custom GPT once invented notes for a meeting that never happened.
The through-line is not automation for its own sake, but useful changes that keep people responsible for context and checking.
They also resist easy promises about business ownership, describing time away and business growth as choices with trade-offs, not effortless freedom.
Katie Stone’s recurring arguments are about control: control of the prompts a business uses, control over which parts of a task stay human, and control over systems so a small education business does not depend on one person’s memory.
Read more
Stone teaches tutors and tutoring-business owners to use AI on real work, treating it as a prediction machine that needs checking, not an authority.
Their other priorities are operational: audits that find small fixes rather than prescribe templates, systems built to connect tools a client already owns, and honest accounts of the trade-offs in running a business.
A reader should come away able to explain why Stone opposes hidden-prompt AI tools, how they decide what a human must still do, and why they think written-down systems let an owner step away.
Generic AI tools produce generic work unless you control the prompt
Stone argues that AI wrapper tools keep tutors dependent.
Read more
A wrapper, as they describe it, is a bolt-on to ChatGPT or Claude that runs a prompt written by someone else, which users cannot see or change.
They cannot tailor the tone, structure or detail to their teaching style, adapt the output for different students or learn what AI is good or bad at.
Stone says the wrapper itself is not the problem; it is a prompt built to work for everyone, producing output suited to no one in particular.
Their priority is for tutors to write, inspect and reuse their own prompts.
Their teaching covers what to include, what context a task needs and how to turn repeated tasks into projects.
Stone compares using a lesson-plan tool to learning to ride a bike with the stabilisers left on for life.
The tool asks for a few data points and returns a plan based on a prompt the tutor cannot see, so they cannot make the output sound like their own teaching.
They names TeachMate as an example, while saying the tool is not bad in itself, and points readers to their prompting webinar.
AI is a prediction machine, not a maths machine, so numbers need code or a check
Stone distinguishes AI's strengths from tasks that depend on calculation.
Read more
In an earlier attempt with ChatGPT, the tool could not read a spreadsheet properly, and the percentages in a school survey report were wrong.
They see this as expected: AI predicts text rather than calculating.
Their practical answer is to use a tool that runs code for the arithmetic and verify the outputs.
That distinction shapes how they teach AI use in education businesses: delegate drafting and structuring, but keep the calculation path inspectable.
For a safeguarding report, Stone downloaded Airtable survey results as a spreadsheet, anonymised the data and gave it to Claude with examples of earlier reports. Claude ran code to do the calculations, and every number was correct, unlike in an earlier ChatGPT attempt.
Keep human judgement on the parts that carry context and consequence
Stone's human-in-the-loop rule is not a blanket ban on automation but a deliberate division of labour.
Read more
They decide in advance which parts of a job depend on their knowledge of the business and the people involved, and keep those for themself.
They also read the full output and adjust anything that still sounds machine-written.
Their argument is that human judgement should not be buried under repetitive work, not that humans should do everything.
In the safeguarding reports, Stone let Claude handle percentages, written summaries of each section and short paragraphs about the learning, trainer and facilities.
They kept the participant quotes: the AI offered a shortlist, and Stone made the final choice, judging that their three years in the business and a decade in schools were not replaceable there.
They read and adjusted every word at the end.
Reports that took thirty to forty minutes now take fifteen.
Publishing AI failure stories is part of teaching people to use AI
Stone treats failures as useful learning, not embarrassment.
Read more
They describe what went wrong, rule out the obvious causes, identify the real one and give readers a rule they can apply.
In doing so, they model the checking behaviour they ask of others and make the point that AI is only as good as the model underneath it.
Sometimes the right response is to stop and try again later.
Stone's custom meeting-notes tool, nicknamed Marlene, produced detailed notes from a meeting that never happened, every time, using a perfectly good transcript.
They ruled out the transcript and the prompt, noticed the failure coincided with a model update, then waited a day.
The tool worked normally again.
Their three lessons: AI is fallible, humans must check at every step, and sometimes the fix is to wait.
Written-down systems are what let a small-business owner step away
Stone argues that the freedom advertised to business owners is a trade, not a default.
Read more
Growth and time off compete; to regain some freedom, owners need to put systems in writing and build a team that can cover.
They reject the idea that running your own business automatically means beaches and school holidays.
Their test of a system is whether the business still works when the owner is away.
Before driving to France, Stone wrote that clients had been sorted, either with work they could do while away or cover from another team member, and that colleague Lucy Oldershaw was available for emergencies and audit enquiries.
Earlier, on a family trip to Bath, they took the laptop and worked in quiet moments.
Six years in, Stone described the earnings ceiling as one they build themself, with every associate and every day off a choice between growth and freedom.
The Bath post frames the trade-off as a personal choice; the France post shows the same argument once systems and cover are in place, allowing the business to continue without Stone.
A fresh outside pair of eyes finds the simple fixes an owner cannot see
Stone's audit pitch rests on a modest claim: their team is not cleverer than the owner and does not know the business better, but has seen hundreds of other tutoring businesses and can spot what is hard to see from inside.
Read more
Audits are live and bespoke rather than templated, and owners are expected to push back on suggestions that will not work. They look for small fixes with large effects, not necessarily a large mess.
One audit found hundreds of lessons recorded with nothing but the word 'lesson' in the name, so anyone wanting to know what a lesson was had to open the record.
Stone called it the smallest fix they spotted: a naming convention that takes seconds to implement and saves the team hours of clicking.
The business itself was strong, but most knowledge lived in the owner's head and was hard for anyone else to find.
Audit clients should describe problems while they are actually having them
Stone's audit method does not expect owners to know everything on the spot.
Read more
Clients get a WhatsApp group for about a week before the audit, where they can send voice notes or messages about problems as they occur while using their systems.
Stone and a colleague read the notes beforehand, then hold a live session to test whether the stated problem is the real problem.
The written audit that follows is bespoke.
Before an audit, Stone wrote about fourteen unlistened voice notes in a client WhatsApp group.
They and Lucy Oldershaw went through them before a live call that afternoon.
Stone brought automation, AI and systems; Oldershaw brought safeguarding, deep TutorCruncher knowledge and compliance.
The client then said what would and would not work.
Build the connective layer between tools a business already owns
Stone's systems work does not assume clients should replace everything.
Read more
They often build an Airtable database and interface that links the tools already in use, bringing scheduling, documents and conversations together in one client view. They prefer systems that can grow as new needs appear and are candid that their favourite improvements are small and unglamorous rather than showy.
A CRM build for a tutoring business linked TutorCruncher, Google Drive and email, which had previously been checked separately.
Instead of staff copying a whole TutorCruncher URL into Airtable, they enter the TutorCruncher ID and a formula creates the link.
That reduces copying and mistakes; if the URL structure changes, the formula can be updated in one place rather than across every record.
The client was technical and wanted a system that could grow, and ongoing VA support ran through a shared Slack channel.
Teach from what tutors actually do, not what you think they should do
Stone polls the audience before designing training.
Read more
When the answers showed a shift in confidence, they changed the plan rather than repeating the beginner framing.
They ask what tutors are using AI for now and what they deliberately keep human-led.
Their priority is to build the next masterclass around those answers.
A poll asked tutors where they were with AI.
Nearly half answered “comfortable, show me advanced,” which Stone said was a change from the last time they asked.
They then asked what those tutors use AI for and what they keep human-led, and said they would build the next masterclass around the answers rather than what they think people should be doing.
Earlier material was written for people who had tried AI, got a poor result and stopped. A newer poll showed that beginner framing no longer described a large part of Stone’s audience, so they changed the teaching they planned to build.
There is no single right way to run a small business
Stone resists the idea that self-employment has one correct shape.
Read more
They argue that business owners who present only freedom and holidays set a dangerous precedent, and that anyone claiming there is one right way is selling something. Their own choice is a mix of family time and work in quiet moments, presented as one version rather than a model.
Writing from a short family trip to Bath, Stone contrasted holiday photos on LinkedIn with having a laptop in their bag. They described choosing two days away while fitting client tasks into quiet moments, and said their version would not be anyone else’s.
This is the clearest account of what Katie Stone's practical AI work looks like inside an education business, including where they draw the line between what the AI does and what they keep.
Read more
A walkthrough of a real client job: producing safeguarding training reports for a school or academy trust from feedback surveys.
They start with the manual process: around twenty results sections in Airtable, percentages to check, and open-ended answers to mine for the best quotes.
Each report took thirty to forty minutes.
Then they explain the change: downloading the results as a spreadsheet, anonymising them, and giving the file to Claude with examples of earlier reports.
An attempt a year earlier with ChatGPT failed because it could not read the CSV or keep the percentages straight.
Claude runs code to do the calculations, so the numbers came back correct.
They kept choosing participant quotes human: the AI offered options, but they made the final call.
They also read and adjusted the whole report where it sounded machine-written.
Reports now take fifteen minutes.
For brands planning a campaign
A credible on-camera demonstration: real input, real output, measured before-and-after results, and clear limits.
Partners can build content around a specific process improvement rather than general AI claims.
They set the guardrails themselves, so a partner script should not promise more automation than they do.
Their clearest statement of the teaching position they would bring to a collaboration, framed as an argument rather than a product pitch.
Read more
An argument against bolt-on AI tools that hide their prompts: they are built on someone else's prompt, which users cannot see or change, so the output cannot be tailored to a tutor's teaching style, tone, structure or individual students.
They say generic prompts produce generic output, and relying on these tools means never learning what AI is genuinely good or bad at.
They name one such tool as an example, but say it is not bad in itself — it is simply built to work for everyone rather than one person.
The post ends by pointing to their prompting webinar.
For brands planning a campaign
This is their most quotable position and a debate-ready point of view. It also narrows the partner fit: a vendor whose product hides its prompts would be a poor match, while content about prompt control and open workflows aligns with what they already teach.
It shows how they handle AI failure in public — a part of AI literacy most trainers skip and education businesses need.
Read more
They describe their custom meeting-notes GPT, nicknamed Marlene, producing detailed notes from a meeting that never took place, every time they tried it with a perfectly good transcript.
They ruled out the transcript and their prompt, noticed the failure coincided with a model update, waited a day and tried again.
The tool then worked normally.
They draw three lessons: AI is only as good as the model underneath it, humans still need to check outputs at every step, and sometimes the best fix is to wait.
The post ends with a reminder about their August webinar and invites readers to comment for a discount code.
For brands planning a campaign
Useful for trust-building content: they publish the failure, diagnosis and fix, not just the wins. The post also shows their standard call to action — comment for a code — which a partner can use to measure engagement.
Concrete examples of their hands-on systems work, detailed enough to show what they actually do rather than just what they sell.
Read more
A close look at a CRM build for a tutoring business that had been juggling TutorCruncher, Google Drive and email, with client information spread across all three.
They describe building an Airtable database and interface as a central hub linking them.
Their favourite detail is small: instead of pasting a whole TutorCruncher URL into Airtable, staff enter the TutorCruncher ID and a formula builds the link.
That means less copying, fewer mistakes and a single place to update if the URL structure changes.
They also explain that the client is fairly technical and wanted a system that could grow.
Ongoing VA support runs through a shared Slack channel, so the client is not left without help after handover.
For brands planning a campaign
A natural technical case study that shows the vocabulary and level of detail they bring to systems content. Because it involves a named tool and a client story, a partner should not imply that the client endorses anything.
Their audit method differs most from the templated audits they criticise elsewhere, and this post explains the whole sequence.
Read more
Instead of asking audit clients to fill in a form or talk through problems cold on a call, they give them a WhatsApp group for about a week.
Clients can voice-note or message about issues as they arise while using their systems.
They and their colleague Lucy Oldershaw, who brings compliance and TutorCruncher knowledge, read through the notes before a live hour with the client.
There, they explore ideas, test whether the stated problem is the real problem, and let the client say what will and will not work.
The write-up afterwards is bespoke, not templated.
The post is framed as a workday aside, not a sales pitch.
For brands planning a campaign
Shows a differentiated process worth explaining in service-comparison or “how it works” content. It also reveals the team structure — a two-person operation with complementary specialist skills — which matters when planning who appears in a collaboration.
A small, vivid example of the fixes they look for, and a good answer to what a systems audit actually finds.
Read more
An audit found a tutoring business had hundreds of lessons recorded with nothing but the word “lesson” in the name, so anyone wanting to know what a lesson was had to open the record.
They call it the smallest fix they spotted: a naming convention that takes seconds to implement and saves the owner and team hours of clicking.
They stress that the business itself was strong; most of the knowledge simply lived in the owner's head and was hard for anyone else to find, something difficult to see from the inside.
They end by noting that summer is a good time to book an audit.
For brands planning a campaign
A memorable, shareable example for content about data hygiene and system naming. It also shows their instinct to lead with the smallest useful change rather than recommend a rebuild.
It shows how they talk about running their own business and the promises they make about time and freedom.
Read more
A post written during a short family trip.
They contrast the holiday pictures that fill LinkedIn when school holidays arrive with the reality of self-employment: they brought the laptop.
Six years into running their business, they say the earnings ceiling did not disappear; it became one they build themselves.
Every associate they take on and every day off is a trade-off between growth and freedom.
They chose two days with their family and work in the quiet moments.
Their closing line: anyone claiming there is one right way to do this is selling something.
For brands planning a campaign
Honest, unpolished founder content rather than a product pitch. It works well in a “day in the life” or founder-story format and sets realistic expectations if a partner wants to frame their work as simplifying business ownership.
It connects their systems work to the outcome most small education businesses want: being able to step away.
Read more
An out-of-office post written before driving to France.
With processes written down and clients either covered by other team members or handled while they are away, they say the business carries on without them.
They direct urgent requests and audit enquiries to Lucy Oldershaw.
Their argument: extracting yourself from your own business is hard but doable when systems do not live only in one person's head and a team can cover.
They invite readers to talk about doing the same for themselves.
For brands planning a campaign
They sell the outcome — time and cover — rather than tools, a strong hook for content aimed at tutoring and education business owners. It also gives a partner a ready-made seasonal angle around school holidays.
Practical AI training for tutors and education business owners4
CRM audits for tutoring businesses4
Formats she works in3
Straight talk about running a small business3
Human in the loop as a working rule2
Building the connective layer between tools a business already uses2
Own your prompt instead of accepting a generic tool1
Designing training from what tutors actually do1
Practical AI training for tutors and education business owners
Katie Stone teaches tutors and small education businesses to use AI on real tasks, not as a novelty.
Read more
Her live sessions — a Basics of AI for Tutors webinar and a Claude for Beginners Masterclass — cover what AI is and isn’t good at, what belongs in a prompt, how to gather the context a task needs, how to turn repeat tasks into Projects, and how to keep personal and client data safe.
Attendees leave with a workbook, template prompts and a reusable prompt built for their own business.
Stone also uses her own failures as teaching material, including a custom GPT that produced detailed notes from a meeting that never happened.
For brands planning a campaign
A ready-made co-branded format: a sponsor can supply a product or data set for a hands-on prompt-building exercise, or take the teaching slot in an existing webinar. The workbook and template prompts give a partner durable materials to attach to the event.
Own your prompt instead of accepting a generic tool
Katie Stone argues that AI wrapper tools keep tutors on stabilisers.
Read more
She describes a wrapper as a bolt-on to ChatGPT or Claude that runs someone else’s prompt, which users cannot see or change.
That prevents them from shaping tone, structure or detail to match their teaching style, and from learning what AI does well or poorly.
She names TeachMate as an example, while noting that the tool itself is not the problem: a prompt designed to work for everyone produces generic output.
Tutors who want lessons, feedback or reports that sound like them, she argues, have to control the prompt.
For brands planning a campaign
A tool that hides or fixes its prompt would sit badly against Katie Stone's published position, and a collaboration built around one would likely be declined or undercut. Products that let tutors see, edit and reuse prompts are the natural fit.
For one safeguarding-report process, Stone anonymised school survey data and gave it to Claude with examples of past reports.
Because Claude runs code, it returned correct percentages, unlike an earlier ChatGPT attempt that could not read the spreadsheet.
Stone still made the call on participant quotes: the tool shortlisted candidates, and she chose and read every word, adjusting language that still felt AI-written.
Report time fell from thirty to forty minutes to fifteen.
In a separate post about a hallucinating meeting-notes GPT, she draws three lessons: AI is only as good as the model underneath, outputs need checking at each step, and sometimes the fix is to wait.
For brands planning a campaign
Content collaborations should include a visible verify-and-adjust step. Katie Stone is willing to publish failure stories, so an AI vendor seeking only success stories or resisting her showing the checking process is a poor match.
Katie Stone’s LinkedIn posts are first-person teaching stories, usually built around one client job or mistake and ending with a plain question or invitation.
Read more
Live 90-minute webinars and masterclasses deliver structured teaching, with replays, discount codes, workbooks and prompt templates.
Audit and CRM-build work is bespoke, delivered through live calls and written reports.
Stone also provides ongoing virtual-assistant support after a build is finished.
For brands planning a campaign
Partnerships can scale with the budget: a webinar or masterclass sponsorship is the lightest touch. Anything tied to audit or build work must respect client confidentiality and the fact that each deliverable is custom, not templated.
Auditing the systems behind tutoring businesses is a recurring service for owners whose setups have drifted through busy terms.
Read more
Katie Stone describes the report as bespoke: a state of play, what is working well, quick wins and longer-term changes.
She discusses it live so the owner can push back on suggestions that will not work.
Stone markets the service seasonally, positioning June and July as the window to fix things before September.
She distinguishes a plain tidy-up from a full audit for owners who suspect their CRM is not doing what they want.
Clients get a WhatsApp group for about a week before the audit, where they can voice-note and braindump problems as they use their systems.
Both auditors review the notes and prepare for a live call of up to 90 minutes.
They explore ideas, look for holes and check whether the stated problem is the real one.
Not every audit uncovers a large mess. In one case, the owner ran a strong tutoring business, but most of the knowledge lived in his head and was hard for anyone else to find. The smallest fix was renaming lessons that were all labelled "lesson".
For brands planning a campaign
A CRM or systems vendor could co-host or sponsor audit content, but the offer depends on live, client-specific problem-finding rather than a template. A campaign built around a generic checklist would misrepresent the service.
Building the connective layer between tools a business already uses
Beyond audits, Katie Stone builds systems, often Airtable databases and interfaces that bring together tools a client already owns.
Read more
One tutoring-business build linked TutorCruncher, Google Drive and email in a single client view.
It included a formula that generates a TutorCruncher client link from an ID, so nobody has to copy and paste URLs and the structure can be updated in one place.
The build was deliberately flexible: the client is technical and wanted something that could grow with new use cases.
Stone says her favourite parts are tiny, unglamorous fixes that make a system easier to use every day.
For brands planning a campaign
She works across multi-tool stacks, not single-vendor environments, so integration stories, migrations and interoperability material fit her content better than a pitch that assumes one product does everything.
A recurring theme in Katie Stone’s posts is that the freedom of running your own business is a trade, not a holiday.
Read more
She writes about taking a laptop on a family trip to Bath and hitting the ceiling of what she could earn alone, then choosing between growth and free time.
Her answer is systems and a team: when she drove to France, client cover was arranged, a team member was available for emergencies and audits, and the business kept running.
Stone names the messiness directly — front ends look calm while back ends are held together with sticky tape — and explains why an outside pair of eyes helps.
For brands planning a campaign
A campaign selling passive income or effortless freedom will ring false alongside Katie Stone’s posts. Brands that value honest accounts of operations, delegation and trade-offs can use her voice well.
Katie Stone polls her audience before building teaching.
Read more
In one poll, she asked where tutors stood on AI and, for the first time, nearly half answered "comfortable, show me advanced" rather than beginner. She then asked what those tutors use AI for and what they deliberately keep human-led, and said she would build the next masterclass around their answers rather than what she thinks people should be doing.
The audience’s stated readiness has shifted. Earlier material addressed people who tried AI, got a poor result and stopped; that no longer describes the group answering the newer poll.
For brands planning a campaign
An edtech brand can contribute a question, data set or feature to a poll-to-masterclass cycle, provided the content follows respondents’ answers. Katie Stone says she will not prescribe before collecting the evidence.
Katie Stone writes about two connected subjects: using AI on real work in a tutoring business, and getting the systems behind that business out of one person’s head.
Read more
A good place to start is their argument about prompts: when a tool runs on someone else’s hidden prompt, tutors stay on stabilisers.
Read the posts explaining why controlling the prompt makes output sound like you.
Then follow their worked examples, such as turning a school safeguarding survey into a report in fifteen minutes instead of forty and explaining which parts they kept for themselves.
For people who run or work in a tuition business, the audit and build posts offer practical guidance on naming conventions, linking tools already in use, and why an outside pair of eyes sees what an owner cannot.
Their posts about taking a laptop on a family trip and arranging cover before driving to France make the same argument about systems, told personally.
Their live webinars and masterclasses package the teaching with workbooks and reusable prompts.
Oldershaw works with the same audience—tuition businesses and agencies—on the systems behind the teaching.
Read more
She specialises in TutorCruncher, onboarding systems, safer recruitment and operational workflows.
She and Stone audit clients together, with Stone covering automation and AI and Oldershaw covering safeguarding, TutorCruncher and compliance.
Her post on the KCSIE updates offers a practical, system-level answer to a regulatory question: gather checks and records in a CRM, automate reminders, and use the tools properly rather than duplicating work.
Read her for the compliance side of the operational picture Stone writes about.
Oldershaw focuses on safeguarding, safer recruitment and compliance rather than AI and prompt training, and also runs a separate service supporting families directly. Her writing covers regulatory readiness and skills rosters; Stone’s focuses on AI workflows and tool architecture.
Abraham makes the same underlying argument as Stone about delegation and ownership: work that nobody owns silently consumes a business, and the fix is often to put someone in charge of what already exists rather than hire.
Read more
His posts ask what keeps coming back to the owner that should not, pointing to follow-up, CRM updates and reporting as tasks that quietly drain expensive people. That overlaps with Stone’s case for written-down systems and the point that a strong business can still depend on the owner’s memory.
Abraham frames the problem through organisational design and hiring for general small businesses and leadership teams, with no education-sector focus, AI teaching or software implementation. His posts are short advisory essays, unlike Stone’s accounts of specific client jobs.
Galajda writes about the unglamorous data work Stone highlights: field naming, records that no longer answer the questions people ask, and the long-term cost of restructuring a CRM.
Read more
His account of a data cleanup, where renamed and abandoned fields left gaps in the story, makes the same point as Stone’s naming convention fix: small conventions, applied consistently, make a system usable later. He also builds and explains a technical product, as Stone does with CRM builds.
Galajda approaches the subject through data integrity for mergers and acquisitions and his work running a consumer shopping app, not education businesses or AI training. His lens is what a buyer or acquirer can reconstruct from the record; Stone’s is what a tutoring team can use day to day.
Placement decision: this is a small, narrowly defined audience of tutoring and education-business owners, not a broad education-technology market.
Read more
A collaboration is best judged as a targeted niche placement on practical AI and CRM systems, rather than a reach buy.
What supports that read: the LinkedIn profile shows 3,424 followers, and the posts speak directly to tutors and tutoring-agency owners.
They ask what those audiences use AI for and what they deliberately keep human-led, offer CRM audits to tuition businesses, build Airtable-and-TutorCruncher systems for them, and sell live AI webinars for education business owners.
Buyers can reasonably expect to reach people who run or staff small tutoring businesses.
The UK framing—Bath, school-term language and UK spelling—comes from the content, not audience data.
The follower count is the only measured audience number stated.
Employer, job role and seniority—the attributes often decisive for an edtech buyer—are not reported.
One self-reported signal is that they describe a poll in which nearly half of respondents chose "comfortable, show me advanced" about AI, pointing to an engaged subset already using AI.
With 3,424 followers, the likely value lies in the direct channels described in the posts—comments-to-DM offers, webinars and an existing client base—rather than raw reach.
Next step: ask for recent results from comparable AI-for-tutors webinars and CRM audit posts, including impressions, comments and DM or link conversions.
Also request audience analytics or a survey covering job role, business size, country, age and gender, plus list size and open and click rates if email is part of the offer.
LinkedIn followers
3,424
LinkedIn profile katiestonepa · 2026-10-09
Reported poll result on AI comfort, as described by the creator
Nearly half of respondents chose "comfortable, show me advanced"
Katie Stone's LinkedIn post describing a poll she ran · 2026-04-07
The LinkedIn posts do not say whether Katie Stone has a sponsorship programme, rate card or brand-collaboration format.
Read more
Their promotional space is used for their own paid offers to tutoring and education businesses: CRM audits and Airtable CRM builds, VA support, a Basics of AI for Tutors webinar, an AI for Tutors masterclass, and a Claude for Beginners masterclass.
They sell directly through posts and personal messages (“poke me”), sometimes with limited places (“2 spots available in June”, “only one more spot”) and comment-to-DM discount codes.
The same recurring webinar link appears across several posts.
Named colleagues support their work, including Lucy Oldershaw for CRM audits and safeguarding expertise and Toby Stone; ask who will deliver any agreed work and how it will be scheduled.
Timing also affects a plan: Katie takes school-holiday breaks with client coverage arranged, was away in mid-July with calls resuming 3 August, and positions July–August as the quiet window for audit bookings.
Their LinkedIn following is roughly 3,400 accounts.
A vendor should treat any collaboration as bespoke outreach: ask about sponsorship policy, audience specifications, pricing and a sponsor-facing inquiry page.
The only visible contact channel is messaging Katie on LinkedIn.
Need: an education-technology company selling AI or systems software to tutoring businesses wants a content collaboration on practical AI and CRM work.
Read more
Katie Stone fits closely because the requested collaboration is work they already publish.
They run a Basics of AI for Tutors webinar and a Claude for Beginners Masterclass, teaching what AI is and is not good for, what belongs in a prompt, how to gather the context a task needs, and how to keep client data safe.
Attendees leave with a workbook and a set of template prompts.
They also build and audit the systems behind tutoring businesses: Airtable databases that join TutorCruncher, Google Drive and email into one client view, and bespoke CRM audits delivered through a live call after a week of client voice notes.
An edtech vendor is buying subject-matter alignment rather than a general influencer slot: the audience self-selects as people who run or staff small tutoring businesses, the buyer the vendor wants to reach.
Two conditions shape the fit.
On topics, Stone argues publicly that AI wrapper tools that hide their prompts keep tutors on stabilisers they never take off: users cannot tailor the output or learn what AI is good or bad at.
A vendor whose product hides its prompt sits badly here; one that lets tutors see, edit and reuse prompts is the natural partner.
On scale, the account shows 3,424 followers, so value lies in the direct channels described in the posts — comment-to-DM codes, webinars and an existing client base — rather than reach.
Judge a plan on conversions and list growth, not impressions.
Publishing failures is an asset for a vendor comfortable being shown honestly.
Stone has described a custom meeting-notes tool inventing a meeting that never happened, using the example to argue that outputs need checking at every step.
They have also described keeping participant quotes as a human decision in a report process where the tool only shortlisted options.
When to choose a specialist
Need: a national tutoring chain, recruitment brand or general small-business software company wants a mass-reach paid advocacy campaign: a scripted endorsement with an impression or reach target, presenting the product as a fully automated fix for a busy owner.
Read more
Why it does not work.
The audience is not a mass market: the profile shows roughly 3,400 LinkedIn followers, and the posts address tutors and tutoring-agency owners specifically, not small-business owners generally.
The offers are Stone's own services — CRM audits, CRM builds, virtual-assistant support, webinars and masterclasses — promoted through posts and personal messages, sometimes with limited places.
A campaign would need bespoke outreach, with terms negotiated rather than bought off the shelf.
The larger mismatch is the message.
A scripted “the software does it all” endorsement conflicts with the published position: automation needs a human in the loop at all times; judgement calls were deliberately kept in a real report workflow; and a personal AI tool's mistake was published rather than buried.
A brand that wants only success stories, no visible checking step and no room for the presenter's own wording would be asking Stone to contradict their teaching.
A vendor whose tool hides its prompt is an even poorer fit: that is the model tutors are told to avoid.
Scheduling is a smaller constraint.
The posts describe school-holiday breaks with client cover arranged and a July–August window for audit bookings.
Audit and build work is delivered bespoke through live calls, so agree timing and deliverables in advance.
Before committing, ask for the commercial terms, per-post performance data and a verified audience breakdown by business size or country.
The posts disclose none of these, so establish terms directly.
A worked campaign example
Angle: a co-hosted “own your prompt” workshop for tutoring and education-business owners, produced with an AI or CRM vendor.
Shape: reuse the live 90-minute format already in use.
The partner supplies a realistic tutoring-business scenario — a feedback report to write up, a client enquiry to handle or a CRM record to update — and Katie Stone teaches what the task needs in the prompt, what context to provide, where the model runs the calculation and where the person stays in the loop to choose and check.
Attendees build one reusable prompt for their own business and leave with a workbook and a set of template prompts.
Show the human-check step live, using the reasoning from the safeguarding-report walkthrough: the tool shortlisted participant quotes, the person made the final choice and read every word afterwards.
Why it works: it uses a format already established through webinars and masterclasses, with replays and discount codes; it carries the clearest published point of view, prompt control over hidden wrapper prompts; and it demonstrates rather than asserts the human-in-the-loop rule.
The audit method — a week of client voice notes and a live call testing whether the stated problem is the real problem — could provide a second session or case-study segment.
For a systems vendor, the natural worked example is the connective-layer build: an Airtable hub joining TutorCruncher, Google Drive and email, with a formula that generates a client link from an ID and replaces copy-and-paste.
Logistics to agree: who appears.
Lucy Oldershaw carries the CRM and compliance side of the audits, and Toby Stone is named in the travel post, so a session with a systems or safeguarding angle may need more than one presenter.
Avoid the school-holiday breaks described in the posts.
July and August are the busiest window for audit bookings and the quietest for tutoring-business owners, which may suit a teaching session.
For measurement, the comment-to-DM discount code is the visible conversion route in existing posts; a partner code or gated workbook download is the obvious way to count sign-ups.
Before committing, ask for rates or terms, dates and results from earlier AI-for-tutors webinars and CRM audit posts, and any audience breakdown by business size or country. These details are not disclosed in the posts.
Choose Lucy Oldershaw when the collaboration leans on compliance and the operational side of a CRM rather than on teaching AI.
Read more
Her posts offer two things a tutoring-systems vendor can use directly.
The first maps new KCSIE safeguarding expectations onto system behaviour: written confirmation of safeguarding checks and risk assessments, records of every address where a child might be taught, and regular placement reviews.
She turns each into a concrete feature, such as forms that collect DBS details and training certificates straight into a database, automated renewal reminders, and TutorCruncher tasks that prompt a review.
Her second post outlines her work: CRM systems like TutorCruncher, onboarding, safer recruitment and operational workflows for tuition businesses and agencies, plus a wider service for families.
If the vendor’s product is sold on making a school’s safeguarding easy rather than complicated, she can demonstrate it against guidance the buyer already has to follow.
Her posts also feature Katie Stone’s audit partner, who handles the compliance and TutorCruncher side, making a two-presenter session a good fit for a discussion of both.
You give up the AI teaching that anchors Katie Stone’s audience: the prompting method, what AI is and isn’t good for, the argument against tools that hide their prompt, and the worked example of a report where the model ran the calculation while a person chose the quotes and read the result.
Their posts are fewer and narrower in subject, so there is little to show range.
Ask for sponsorship terms, rates and partner-programme details.
Abraham when the buyer wants the bottleneck argument rather than a tutoring-specific one.
The posts focus on operations and delegation: clients arrive asking for an executive assistant or somebody to follow up on leads, when the real problem is that nobody owns the follow-up.
Leads sit in a CRM untouched because the owner is selling, managing the team and handling clients at once.
The companion post puts a number on it: people worth $200 an hour spending half the day on $20-an-hour work, with the test, ‘what am I still doing because nobody else owns it?’ That fits a vendor positioned as taking repetitive follow-up and admin off an owner’s plate.
The framing is deliberately audience-agnostic: the writer says the pattern shows up in small businesses, companies doing millions, and leadership teams alike.
It’s a good route to reach founder and operations audiences rather than build depth in one sector.
Nothing here is about tutoring, schools, safeguarding or the mechanics of a specific platform: no Airtable, TutorCruncher or prompt-level detail.
A tutoring vendor would reach a wider but less qualified audience.
The two posts offer a point of view, not a demonstrated body of instruction.
Choose Jacob Galajda when the collaboration is about data quality, plumbing or product engineering rather than day-to-day AI and CRM use.
Read more
One post walks through a data cleanup for an M&A deal: records scattered across a CRM, spreadsheets and exports from tools the company had retired; fields renamed or abandoned as priorities shifted; and entire categories—customer retention patterns, support volume by cohort, pipeline velocity by segment—incomplete or missing.
The transferable point for a systems vendor is that data has to be collected consistently before it feels urgent, and renaming a field breaks the record.
The other post covers building his own shopping product and the infrastructure decisions behind it: moving from in-house computing to cloud resources, cutting cost while improving speed and recommendation accuracy, and a public beta waitlist.
That suits a partner selling infrastructure, data tooling or an engineering-facing story.
The build-in-public approach may resonate with some technical audiences.
The subject is a transaction-ready data estate and a consumer shopping tool, not tutoring businesses or AI instruction, so a tutoring-focused vendor would reach a different audience with a different problem.
The natural format is a technical or founder-story segment rather than a teaching session.
The same creator is the subject of both posts, offering little independent evidence of how they handle a partner’s brief.
Ask for examples of their teaching style and audience fit before considering a recurring collaboration.
This page describes 13 LinkedIn posts from Katie Stone and the LinkedIn profile as of 9 October 2026.
The comparison and similar-creator sections draw on six posts from three other creators.
Katie Stone's posts do not include publication dates, so ideas are described as recurring arguments rather than a dated timeline.
Katie Stone's posts do not identify a signature book, show or earlier body of work, or state a sponsorship policy, rate card, pricing, per-post performance data, or audience breakdown by job role, business size or country.
Similarity to other creators rests on the arguments and methods in the posts, not on client lists or commercial availability.
Commercial terms and audience detail would need to be established directly.
