How an accounting firm reduced admin work by 63% with an AI receptionist
Partners and managers were losing hours every week to scheduling, intake and routine client questions. After Boafo Agent took over first-touch on web, phone and email, admin work dropped 63%, every enquiry was captured instantly, and partners got back to chargeable work.
Results at a glance
The problem
The firm has six partners and twelve managers serving roughly 480 active clients across tax, audit and advisory. First-touch with prospects and routine client questions both landed on whoever was nearest the phone, which was usually a partner.
The partners had become an expensive switchboard. Time-and-billing records showed that 11% of partner time was spent on intake, scheduling and answering questions that could have been handled by a competent receptionist with access to the diary and the client file.
Out-of-hours was a different problem. Prospects who searched on a Sunday evening for an accountant found the firm website, filled in a contact form, and received an automated thank-you. The first human response typically arrived three working days later, by which point the prospect had usually retained someone else.
Hiring a receptionist had been discussed and rejected. The firm did not want a single point of failure on the phone, did not want to manage holiday cover for a front-of-house role, and wanted something that could answer routine technical questions about deadlines, document requirements and fee structures with partner-level accuracy.
The solution
Boafo Agent was deployed on the firm website chat, the main inbound line and a monitored intake mailbox. The AI was trained on the firm service catalogue, fee schedule, partner specialisations and the standard intake checklist for each service line.
Prospect intake follows one flow. The AI asks about the prospect entity type, current accountant status, services needed and ideal start date, matches the prospect to the right partner by specialisation and capacity, and offers two confirmed slots in that partner calendar.
Existing-client questions follow a separate, more capable flow. The AI authenticates the client, looks up their file, and can answer the most common questions directly: upcoming deadlines, document requirements, fee status and meeting history. Anything outside the playbook is escalated to the responsible manager with a full transcript.
Out-of-hours behaviour is explicit. Prospects who arrive after 18:00 or at the weekend get the full intake experience, with a confirmed slot in the relevant partner diary for the next available window. The partner walks in on Monday to a fully-briefed meeting, not a backlog of contact forms.
The results
Admin work across partners and managers fell 63% in the first quarter, measured against the firm own time-and-billing data.
Intake-to-meeting time fell from a three working day average to 11 minutes for the great majority of prospects, including those who arrived out of hours.
Each partner recovered an average of 9 hours per week, which translated into a 12% lift in chargeable hours with the same headcount.
New-client conversion rose 34% in the first quarter, driven primarily by the firm now being the first to respond instead of the third or fourth.
“Our partners are not employed to chase calendars or take name and address details. The AI handles all of that, sends us a clean brief and books the meeting. We have recovered nearly a day per partner per week of genuinely chargeable time without hiring a single person.”
Before vs. after
| Metric | Before | After Boafo Agent |
|---|---|---|
| Intake-to-meeting time | 3 working days | 11 minutes |
| Out-of-hours intake | Contact form, 3-day delay | Live AI intake, instant booking |
| Partner time on admin | 11% of billable hours | 4% of billable hours |
| Client enquiries captured | around 85% | 100% |
| New-client conversion | Baseline | +34% |
| Chargeable hours / partner | Baseline | +12% |
| Headcount | 6 partners, 12 managers | Unchanged |
Implementation playbook
The deployment was sequenced carefully because the firm operates in a regulated environment and the partners wanted explicit assurance that client data would never be exposed inappropriately. Week one was a closed pilot on the prospect intake flow only, with no client data in scope. The pilot ran for ten business days and the conversion lift was measurable inside the first week.
Week three opened the existing-client question flow, gated behind explicit authentication. The AI is only allowed to read from a tightly scoped set of data sources per service line, and every read is logged with timestamp, message ID and the responsible partner. The compliance partner reviewed the first 200 conversations personally and signed off on the controls.
Adoption among partners was driven by one specific behaviour change: the AI sends a one-paragraph brief to the relevant partner ten minutes before any booked meeting, including the prospect entity type, services needed, prior research and the AI confidence score on fit. Partners stopped walking into meetings cold for the first time in years.
The single biggest behavioural shift was at the front desk. The firm had been quietly debating a receptionist hire for 18 months. The AI removed the need entirely and freed the partners from being the firm switchboard. The managing partner has stated publicly that the firm will not hire a front-of-house receptionist for the foreseeable future.
Operational discipline keeps the system honest. Every Monday the practice manager reviews the previous week dashboard with the firm operations partner, with particular attention to escalation accuracy, fee-quote accuracy and any client question the AI declined to answer. Any pattern that appears twice is fixed the same week. That cadence is the reason the 63% admin reduction has held for two quarters.
What is next
The firm now uses Boafo Agent as the standard first touch for every channel. The managing partner has set an explicit policy that partners should not be the first human a prospect speaks to, and the AI makes that policy enforceable in practice.
The roadmap for 2026 is to extend the AI into post-engagement work: deadline reminders, document chasing and fee-status questions, all of which currently consume manager time at scale.