# Katie Stone

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.

- Profile: https://b2binfluencer.marketing/creators/katie-stone/
- LinkedIn: https://www.linkedin.com/in/katiestonepa
- LinkedIn followers: 3,424
- Topics: artificial-intelligence
- Based on 20 dated posts, latest 2026-09-16

## Summary

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.
- Check first: Ask for recent results from webinars and CRM audit posts, plus audience analytics.

### 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.

## Main ideas

- **Generic AI tools produce generic work unless you control the prompt**: Stone argues that AI wrapper tools keep tutors dependent. 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.
- **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. 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.
- **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. 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.
- **Publishing AI failure stories is part of teaching people to use AI**: Stone treats failures as useful learning, not embarrassment. 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.
- **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. 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.
- **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. 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.
- **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. 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.
- **Build the connective layer between tools a business already owns**: Stone's systems work does not assume clients should replace everything. 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.
- **Teach from what tutors actually do, not what you think they should do**: Stone polls the audience before designing training. 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.
- **There is no single right way to run a small business**: Stone resists the idea that self-employment has one correct shape. 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.

## Work to read

- [Turning safeguarding survey data into school reports, with a human choosing the quotes](https://www.linkedin.com/posts/katiestonepa_educationbusiness-edtech-activity-7444640300695801856-jqLx)
- [Why they compare AI wrapper tools to stabilisers you never take off](https://www.linkedin.com/posts/katiestonepa_aifortutors-promptengineering-aiineducation-activity-7360922676586061824-DWn7)
- [The custom GPT that invented a meeting that never happened](https://www.linkedin.com/posts/katiestonepa_aifortutors-aiineducation-edtech-activity-7361273381574385664-GUYn)
- [The CRM build where a formula replaced copy-and-paste](https://www.linkedin.com/posts/katiestonepa_my-favourite-bits-of-a-crm-build-are-often-activity-7505869473434832896-zb5v)
- [The audit that starts with fourteen unlistened voice notes](https://www.linkedin.com/posts/katiestonepa_there-are-like-14-unlistened-to-voice-notes-activity-7490670456518250496-LZfV)
- [Hundreds of lessons all labelled “lesson”](https://www.linkedin.com/posts/katiestonepa_last-year-we-did-one-of-our-audits-on-a-tutoring-activity-7480889783649640448-otAQ)
- [Two days in Bath, with the laptop in the bag](https://www.linkedin.com/posts/katiestonepa_smallbusiness-tutoringbusiness-activity-7447538802178486272-M_Jk)
- [Out of office, but the business is not](https://www.linkedin.com/posts/katiestonepa_im-not-really-here-no-really-by-the-activity-7483449118619627520-iPnb)

## Brand relationships

- **Elevated Support Ltd** (disclosed_partnership)
- **TutorCruncher** (mention)
- **Airtable** (mention)
- **Claude** (mention)

## Partnerships

The LinkedIn posts do not say whether Katie Stone has a sponsorship programme, rate card or brand-collaboration format. 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.

## Alternatives

- Lucy Oldershaw: Choose Lucy Oldershaw when the collaboration leans on compliance and the operational side of a CRM rather than on teaching AI. 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.
- J. Abraham: Choose J. 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.
- Jacob Galajda: Choose Jacob Galajda when the collaboration is about data quality, plumbing or product engineering rather than day-to-day AI and CRM use. 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.

---
AI-assisted summary of public posts. Audience demographics are not available. Engagement appears only where we checked recent posts. Method: https://b2binfluencer.marketing/methodology/