Key takeaways
- Staff are already using AI, often more than their employers: plan for that rather than pretending it is not happening[1].
- Be specific and honest about jobs; most UK businesses using AI report no change in headcount[1].
- Let reception co-write the scripts and escalation rules; they know what callers really ask.
- Train before go-live, and keep training short, practical and repeated.
- Measure adoption by what staff do, not what they say in the launch meeting.
Why staff decide whether the tool works
A new phone assistant or drafting tool can be bought by the partners, but it is used, or quietly worked around, by reception, secretaries, paralegals and junior fee earners. If they do not trust it, they will transfer every call straight back to themselves, retype every note and tell clients the system is unreliable. The tool then fails for reasons that have little to do with its quality.
Staff are also not starting from zero. ONS figures show 55% of UK employees report using AI for work or education, compared with around 35% of businesses reporting use of at least one AI technology[1]. In many firms, people are already experimenting with free tools on their own. The Legal Ombudsman warns that not all free AI tools are secure or keep what users type confidential[2], so bringing staff into an approved, supervised approach also reduces risk.
What support staff worry about
Concerns are predictable, and naming them early takes the heat out of them. In conversations with reception and support teams, the same themes come up:
- Jobs: whether the tool is the first step towards replacing them.
- Blame: whether they will be held responsible when the tool gets something wrong.
- Extra work: whether they will spend their day correcting notes and calling back confused clients.
- Losing the good parts: many receptionists like talking to people and solving problems, and fear being left with only the admin.
- Being monitored: whether call reviews will be used to judge them rather than the tool.
Each worry deserves a direct answer from a partner, not just from the practice manager, and the answers should be written down so they can be checked later.
Be honest and specific about jobs
Vague reassurance does more harm than a clear statement. The evidence from UK businesses is calmer than the headlines: of businesses using AI to improve operations, 63% reported no change in headcount, 1% an increase and 6% a decrease[1]. That does not guarantee anything for your firm, so say what you actually intend.
Source: Office for National Statistics, 2026[1]
Show the data
| Item | Value |
|---|---|
| No change | 63% |
| Decrease | 6% |
| Increase | 1% |
A useful statement covers three points: what the tool is for, such as answering overflow and evening calls that currently go to voicemail; what will happen to the time it frees, such as more time for client follow-up or onboarding checks; and what the firm is not planning, if that is true. Receptionists are a large workforce, with about 195,000 jobs in the ONS earnings data[3], and many are weighing up what AI means for them. If roles may change, say how, when and what support will be offered.
Eight steps for introducing AI with the team
Show the problem with your own data
Share the firm's missed-call figures, voicemail backlog or after-hours enquiries. People accept a change more readily when they can see the problem it addresses.
Map tasks together
Ask reception to list what they do in a typical day and mark which tasks they would happily give away and which they want to keep.
Co-write the scripts
Let receptionists draft the greeting, the intake questions and the lines for awkward callers, then have fee earners check them for accuracy.
Choose champions
Pick one or two respected team members to test the tool first, report problems and explain changes to colleagues.
Train before go-live
Run short, hands-on sessions: how the tool works, what it will not do, how to read its notes and how to report a problem.
Build a feedback loop
Give staff one simple way to flag a bad call or note, and show them what changed as a result each week.
Recognise the fixes
When a receptionist's suggestion improves the rules, say so in the team meeting. It shows their expertise is shaping the tool.
Review roles after three months
Sit down with each affected team member to discuss how their work has changed and what they want next.
A script for the first team meeting
Example: 'We miss calls at lunchtime and every evening call goes to voicemail. We are going to trial an assistant that answers those calls using questions you help us write. It will not give advice, and it will pass anything urgent or unusual to a person. Nobody's job is being cut because of this trial. We want you to tell us every time it gets something wrong, because you know our callers better than anyone. In a month we will look at the results together and decide whether to keep it.'
Then stop talking and take questions. Write down every concern raised, answer what you can on the spot and come back on the rest within a week.
Training that people remember
Few businesses have trained most of their people. Only 11% of UK businesses with ten or more employees say more than half of their workforce has received AI-related training[1]. In a small firm, a little goes a long way.
| Who | What they need | Format | When |
|---|---|---|---|
| Reception | How calls are handled, reading notes, taking over a call, reporting problems | Hands-on session with test calls | Before go-live, then a refresher in week two |
| Fee earners | What the notes contain, what the tool will not do, how escalation reaches them | Short briefing in team meetings | Before go-live |
| Secretaries and paralegals | How enquiries reach them and how to correct records | Walkthrough with real examples | First week |
| Partners and COLP | Supervision duties, change approval, review of samples | Short briefing plus the review pack | Before go-live and monthly |
Set rules for staff use of AI tools
Introducing one approved tool is a good moment to set rules for everything else. The Law Society's generative AI guidance, aimed particularly at small and medium-sized firms, highlights the risk of misuse or disclosure of confidential information[4]. A short policy should cover:
- Which AI tools are approved for work, and that client information goes into nothing else.
- That outputs are checked by a person before they reach a client or a file.
- Who to ask before trying a new tool, and how quickly they will answer.
- How to report a mistake or a possible disclosure without fear of blame.
Human checking is the norm, not a sign of distrust: 84% of UK businesses using AI apply at least some human input or checking to AI outputs[5].
When someone stays opposed
Some staff will remain unconvinced after the meeting and the training, and that is not always a problem to be managed away. Persistent objections often contain useful information: a receptionist who insists the tool cannot handle a particular group of callers may be right. Ask them to keep a short log of the calls they think went wrong, review it together each week and act on what it shows.
If the objection is about change itself rather than the tool, give the person a defined role in the trial, such as checking overnight notes, so they see the results first-hand. Avoid making adoption a test of loyalty. The aim is a team that uses the tool because it helps, and a tool that improves because the team is willing to say when it does not.
Signs it is working, and signs it is not
- Working: staff report problems quickly, suggest wording changes and stop transferring routine calls back to themselves.
- Working: fee earners act on enquiry notes without ringing reception to ask what the caller wanted.
- Not working: reception keeps a private list of calls 'just in case', or retypes every note.
- Not working: complaints about the tool arrive only through partners, never through the feedback route.
Example: in a fictional four-partner firm, the senior receptionist rewrote the probate intake questions after the first week because callers kept describing a death before saying what they needed. Misrouted calls dropped and the team began suggesting changes for other practice areas. For the questions to ask suppliers before you start, see choosing an AI phone provider; for how roles are shifting more widely, see how AI is changing legal support roles.
Frequently asked questions
Will an AI phone assistant replace our receptionists?
It depends on what the firm decides, which is why partners should say clearly what the tool is for. Most UK businesses using AI to improve operations report no change in headcount, and in many firms the tool covers calls that currently go unanswered.
How should we involve reception staff?
Ask them to map their tasks, co-write the greeting and intake questions, test the tool first as champions and report problems through one simple route, then show them what changed.
How much training do staff need?
Short, hands-on sessions before go-live and a refresher after a week or two. Reception needs the most practical training; fee earners mainly need to know what the notes contain and how escalation works.
Do we need a policy on staff using AI tools?
Yes. A short policy should list approved tools, ban putting client information into anything else, require human checking of outputs and make it easy to report mistakes.
References
Office for National Statistics (2026). Artificial intelligence in UK businesses: 2023 to 2026 (opens in a new tab) ↑
Legal Ombudsman (2026). AI and Complaints: A Consumer Guide (opens in a new tab) ↑
Office for National Statistics (2026). Earnings and hours worked, occupation by four-digit SOC: ASHE Table 14 (2026 provisional) (opens in a new tab) ↑
The Law Society (2026). Generative AI - the essentials (opens in a new tab) ↑
Department for Science, Innovation and Technology (GOV.UK) (2026). Artificial Intelligence Adoption Research (opens in a new tab) ↑