A three-partner accountancy firm in West London was losing 18 hours of collective billable time per week to non-billable administrative tasks: scheduling client meetings, managing inbound calls, processing routine client queries, sending deadline reminders, and coordinating document requests. After deploying Softomate's AI personal assistant across all three partner workflows, non-billable admin fell to 6 hours per week. At an average partner billing rate of £175 per hour, the 12 hours recovered translates to £2,100 per week or £9,100 per month in additional billable capacity. In practice, the firm recovered approximately £4,200 per month of that capacity in new billable work during the first 90 days - with the remainder used for business development activity that had previously been deferred.
The practice serves 340 active business clients across a range of sectors: property, professional services, retail, and hospitality. Its work mix is split between annual accounts and tax returns (60%), ongoing management accounts and bookkeeping (25%), and advisory and planning work (15%). Three equity partners handle client relationships; a team of six qualified and semi-qualified accountants handles the technical work. Two administrative staff members support the full team.
The admin burden the partners described was not primarily data entry or bookkeeping - those tasks were handled by the accountancy team. The burden was in relationship management overhead: the calls, emails, and scheduling coordination that partners handled personally because client relationships were their responsibility. The firm had not added administrative headcount in three years despite growing its client base by 22%. The admin-to-partner ratio had stretched to the point where partners were beginning client calls 15 minutes late, missing follow-up windows after meetings, and losing new enquiry leads to slower-than-expected responses.
Softomate configured an AI PA for each of the three partners, with a shared inbound call handling layer for the main office number. Each partner's AI PA connected to their individual Microsoft 365 calendar and email, the firm's practice management system (IRIS), and the shared VOIP phone system.
The call handling configuration: the main office number routes to the AI during busy periods and after hours. The AI identifies the caller's name and client reference, looks up the relevant partner, and routes or messages accordingly. New client enquiries are captured with company name, query type, and preferred contact time, then routed to the business development partner with a brief. Existing client calls are either transferred (if the relevant partner is available and not in a client meeting) or logged with a message and a committed callback time.
The email configuration for each partner: the AI triages incoming email into four categories - immediate action required (contracts, HMRC correspondence, urgent client issues), client queries requiring a response within 24 hours, administrative (scheduling, confirmations, non-urgent queries), and noise (newsletters, promotions, generic supplier communications). Partners review only the first two categories in real time; the AI drafts responses for the third category for approval. The fourth is archived without manual processing.
Scheduling: the AI handles all meeting scheduling using a booking link with each partner's availability rules configured. Clients receive the link after initial contact; the AI confirms, sends reminders 24 hours and 1 hour before meetings, and attaches relevant client account information as a pre-meeting briefing.
| Metric | Before | After (90 days) | Change |
|---|---|---|---|
| Partner non-billable admin per week | 18 hrs total | 6 hrs total | -12 hrs/week |
| Average time to respond to new enquiry | 4.2 hours | 22 minutes | -92% |
| Missed calls (unanswered, no voicemail) | 34% of inbound | 4% of inbound | -88% |
| New enquiries converted to clients (90 days) | 38% conversion | 51% conversion | +34% |
| Additional billable hours per month | - | +24 hrs (collective) | - |
| Monthly revenue impact | - | +£4,200 | - |
The new enquiry conversion improvement from 38% to 51% was the outcome the managing partner found most significant. The improvement was attributable to speed of first response: enquiries that reached a partner within 30 minutes of initial contact converted at 68%, while enquiries where the first partner contact was more than 2 hours after initial enquiry converted at only 31%. The AI ensured that every enquiry reached a partner (via message or live transfer) within the first response target window. The 13-percentage-point improvement in conversion added approximately 8 to 10 new clients over the 90-day period - representing £60,000 to £80,000 in additional annual fee income beyond the immediate admin recovery benefit.
The deployment began in November, meaning the firm's peak period - January through April (self-assessment deadline, then corporation tax deadlines) - fell entirely within the first six months of AI PA operation. Tax season is the period where the admin burden in accountancy practices traditionally spikes: client queries about deadlines, document requests, chasing outstanding information, explaining tax positions under time pressure. The AI handled this volume increase without additional cost or configuration change.
Inbound call volume in January increased 340% above the November baseline. The AI handled 91% of calls without escalation to a partner. Partner interruptions during client-facing work dropped by 67% compared with the previous January, despite the higher call volume. Clients received consistent, accurate deadline information from the AI call handler, reducing the number of "quick deadline check" calls that previously consumed partner time throughout January.
The managing partner noted that this was the first January in four years where no partner worked past 7pm on a weekday. The admin overhead that had previously extended working hours during peak season was absorbed by the AI, allowing the team to work more efficiently during standard hours.
The AI PA handles HMRC correspondence routing but does not respond to it autonomously. Correspondence arriving by post is scanned and uploaded to the client file (the administrative team handles scanning). Correspondence arriving by email or via the HMRC online agent account is flagged immediately to the relevant partner with a priority tag and a 24-hour response reminder. The AI does not draft responses to HMRC correspondence - this is explicitly outside its scope, as such responses require qualified professional judgment. The AI's role is to ensure no HMRC correspondence is missed or delayed, not to respond to it.
For routine client requests relating to HMRC - payment reference queries, UTR number requests, PAYE code queries - the AI provides accurate information from the client file. This handles approximately 40% of the "quick HMRC question" calls that previously interrupted partners.
The managing partner cited two factors that differentiated the Softomate deployment from a previous attempt to use an off-the-shelf virtual assistant tool. First: the IRIS integration. Previous tools could not connect to the practice management system, which meant the AI had no access to client records and could not provide account-specific information or route calls correctly. Softomate's IRIS integration resolved this. Second: the accountancy-specific knowledge base. The AI was loaded with correct answers to the 50 most common client questions in an accountancy context - deadlines, payment methods, document requirements, HMRC processes. This removed the need to configure the knowledge base from scratch.
AI PA services for UK accountancy firms start at £299 per month from Softomate. The three-partner firm in this case study paid £449 per month (Professional tier, covering three user integrations and IRIS PMS integration). Against the £4,200 per month in billable hour recovery and the longer-term new client acquisition benefit, the ROI on the subscription is clearly positive from month one. For further information on AI PA services for accountants, see AI PA for UK accountants and AI PA ROI for UK businesses.
The most concrete way to understand the AI PA's impact is to compare a specific partner's day before and after deployment. The managing partner - responsible for the firm's largest clients and most of its business development - was tracking his time for the two weeks before deployment as part of the onboarding process. He was spending an average of 4.2 hours per day on email, 1.1 hours on scheduling-related tasks, and 1.8 hours on inbound calls and call-backs. That is 7.1 hours per day on administrative tasks that an AI PA could handle in full or in part - against a target billing day of 7.5 hours.
After six weeks, the same tracking exercise showed: 35 minutes per day on email (reviewing AI drafts, approving or editing), 12 minutes on scheduling (approving AI-proposed slots), and 28 minutes on calls and call-backs (returning the 3 to 4 calls per day that required personal partner involvement, rather than 18 to 22). Total daily admin time: approximately 75 minutes. The remaining time was reclaimed for billable client work, business development calls, and one afternoon per week dedicated to strategic planning - a luxury the partner had not had in the previous three years.
The impact on the other two partners followed a similar pattern, with variation based on their specific client mix. The partner handling the most SME bookkeeping clients (who generate high call volume relative to fee value) saw a 68% reduction in call interruptions. The partner handling advisory and planning work (whose calls are less frequent but more complex and longer) saw the largest email time reduction - 71% - because client update communications dominated her inbox and the AI handled the bulk of standard update requests automatically from the IRIS client file.
The partners were cautious about how clients would respond to an AI handling their initial contact. The firm's positioning is built on a personal, relationship-led service model - the partners were concerned that AI-handled calls and emails might undermine that positioning. The reality, based on client feedback collected at the 90-day mark, was the opposite: clients responded positively to the improved responsiveness and consistency.
Of 87 clients surveyed at the 90-day mark, 76% rated the firm's responsiveness as "better than before" or "significantly better than before". 12% reported no change. 12% were unaware that any system change had occurred and had no comment on responsiveness. Zero clients reported a negative experience with the AI-handled initial contact. Several mentioned specifically that they appreciated receiving a call back from the AI within minutes of leaving a voicemail, rather than waiting hours.
One client - the owner of a 40-person professional services firm - noted in his feedback: "I don't particularly care whether it's an AI or a person who takes my initial message. What I care about is that my message is received and acted on promptly. [Partner name] now calls me back faster than any other professional I work with." This feedback captures the client sentiment accurately: for the routine touch-points that AI handles, clients value outcome (fast, accurate response) over mechanism (human vs AI).
The firm's professional indemnity insurer reviewed the AI PA deployment before the firm went live. The key questions from the insurer related to three areas: whether the AI could provide advice to clients (no - the AI provides information from the client file and routes advisory queries to the partner), whether client data was stored on third-party servers (yes, on UK-based encrypted servers under UK GDPR compliance, covered by a data processing agreement), and whether the AI could impersonate a qualified accountant (no - the system identifies as an automated assistant and routes any query requiring professional judgment immediately to a partner).
The insurer confirmed that the deployment did not affect the firm's professional indemnity coverage. The firm updated its client engagement letters and privacy notices to disclose AI-assisted communication handling. This was a one-page addition to existing documentation, reviewed by a solicitor, and took one week to finalise and update across all active client files.
For ICAEW-registered firms, the AI PA deployment falls outside the scope of technical guidance on AI and professional standards published by the Institute in 2025, which focuses on AI use in audit and assurance work. Administrative AI that handles scheduling, routing, and communication does not affect technical professional standards and requires no specific ICAEW notification or approval. The firm consulted the ICAEW helpline to confirm this position before deployment.
The 12-month data available for this case study shows that the improvements established in the first 90 days compounded over the following nine months. Admin hours per partner stabilised at 6 hours per week collectively - the 12-hour weekly saving was maintained without regression. Billing per partner increased by an average of £3,100 per month compared with the same period in the prior year, controlling for growth in the client base. New client conversions attributable to faster first response generated an estimated 22 additional clients over 12 months, representing approximately £210,000 in new annualised fee income.
The firm has since extended the AI PA service to cover its six qualified accountant staff members, who handle client communications independently for their allocated client portfolios. The Extended tier subscription covers the full team at £599 per month - a 33% increase in subscription cost that the partners estimate generates a further £8,000 to £12,000 per month in additional billable capacity across the wider team. Total annual ROI on the AI PA subscription at 12 months: approximately 45x the subscription cost in measurable financial benefit.
After the initial 90-day partner-level deployment, the firm extended the AI PA to its six qualified accountant staff members in month four. Each accountant manages a portfolio of 40 to 60 clients and handles significant client communication independently. The configuration was a lighter version of the partner setup: email triage, scheduling, and call routing, without the CRM-based briefing note function reserved for partners handling business development.
The impact at team level differed from the partner impact. Partners had been losing billable hours to admin. Accountants had been managing client communications alongside technical work in a way that created context-switching costs: stopping a reconciliation mid-task to answer a client call, then spending 10 minutes re-establishing context. The AI reduced interruption frequency. Accountants reported completing technical tasks 20 to 30% faster after the reduction in mid-task interruptions.
The combined productivity gain at 12 months, across the 9-person professional team, represents approximately 85 recovered billable hours per month. At an average blended billing rate of £120 per hour across the team, that is £10,200 per month in recovered billing capacity against a £599 per month subscription for the full team tier. Return at 12 months: 17x the monthly subscription cost. The firm now includes the AI PA subscription in its core annual budget, alongside software licences and professional subscriptions, as a permanent operational tool rather than a technology experiment.
For firms assessing whether an AI PA is appropriate for a wider team beyond principals, the accountancy firm experience provides a clear benchmark. The configuration for staff members requires careful definition of authority levels but delivers the same core benefit: fewer interruptions, faster client response, and more consistent follow-up. See AI PA for UK accountants and AI PA ROI for UK businesses for full details.
For accountancy practices evaluating an AI PA deployment, Softomate recommends beginning with a 30-minute scoping call to map current workflow and quantify admin overhead. The call uses a standard time-audit questionnaire covering call volume, email volume, scheduling time, and follow-up frequency. The output is a specific hour-recovery estimate and an ROI projection using the practice billing rate. This analysis is provided at no cost and takes one week to complete. Most practices find the resulting projection both more specific and more conservative than their initial estimate - and still compelling enough to proceed to the 14-day trial. See AI PA for UK accountants for detailed service information.
UK accountancy firms that are ICAEW, ACCA, or CIOT members and are considering an AI PA deployment can request a copy of the data processing agreement and compliance summary that the firm in this case study used to satisfy its professional indemnity insurer and update its engagement letters. This documentation package is provided by Softomate at no cost as part of the pre-deployment due diligence process. It covers UK GDPR compliance, data residency, professional confidentiality, and the scope boundaries of the AI PA role. Having this documentation ready significantly accelerates the governance sign-off that most firms require before proceeding to a live deployment, reducing the typical governance review from three to four weeks to five to seven working days.
Partners recovered 12 hours of collective weekly admin time, reducing non-billable admin from 18 hours to 6 hours per week. At an average billing rate of £175 per hour, that is £2,100 per week in recovered billable capacity. In the first 90 days, the firm converted approximately £4,200 per month of that capacity into actual additional billing.
The AI flags and prioritises HMRC correspondence for immediate partner attention but does not draft or send responses. Regulatory correspondence always requires qualified professional judgment and partner sign-off. The AI ensures no correspondence is missed or delayed, which previously happened during peak periods when the team was under pressure.
Softomate integrates with IRIS, CCH, Xero Practice Manager, QuickBooks Accountant, and most major UK accountancy practice management platforms. Integrations are built and maintained by the Softomate technical team. The firm does not need to change its existing software to use the AI PA service.
Response time was the primary driver. Enquiries answered within 30 minutes converted at 68%; enquiries answered after 2 hours converted at only 31%. The AI ensured every new enquiry reached a partner within the firm's target response window, lifting conversion from 38% to 51% over 90 days and generating an estimated 8 to 10 new clients during that period.
Yes. The Standard tier at £299 per month covers a single-user integration and handles the inbound call, email triage, and scheduling functions that are most valuable for small practices. A sole practitioner recovering even 5 hours of weekly admin time at a £120 billing rate generates a monthly return of £2,400 from a £299 investment. Sole practitioners and two-partner firms are among Softomate's most active accountancy clients.
We protect the real names of all clients featured in examples and case studies. Every testimonial is from a real client.
More from the blog
Work with us
Every project we take on has a measurable outcome. Talk to our London team and we will show you exactly how we would approach your challenge.
Softomate Solutions Limited
Deen Dayal Yadav
Online
