A Sydney advisory firm was spending 14+ hours every week on admin tasks that didn't need a human: answering repetitive inbound queries, triaging contact form submissions, and manually scheduling initial consultations. Three weeks later, an AI chatbot and automated workflow were handling it.
Meridian Advisory is a specialist firm. Their advisors are senior, their work is complex, and their time is expensive. But when we did a time audit in the discovery session, the picture was sobering: between the team, they were collectively spending 14 hours every week on tasks that had nothing to do with advising clients.
The bulk of that time was falling into three buckets. First, answering inbound enquiries via email and the website contact form — many of which were asking the same five questions about services, fees, and process. Second, manually triaging those enquiries to figure out whether a lead was qualified and which advisor should handle them. Third, going back and forth to schedule initial consultations, a process that often took three to five emails per booking.
None of this required a senior advisor. All of it was getting done by senior advisors anyway, because the firm didn't have dedicated admin support and hadn't found an off-the-shelf tool that fit their workflows without a significant configuration burden.
The brief was tight: find and automate the 80% of inbound that doesn't need a human, get it live quickly, and do it in a way that doesn't feel like the firm is hiding behind a bot. Clients still need to feel like they're dealing with a serious, responsive organisation — the automation just needs to handle the parts that don't require judgement.
Workflow audit and automation mappingSpent two days with the team logging exactly where time was going. Categorised every type of inbound communication by volume, required response complexity, and whether a human decision was actually needed. Result: 68% of inbound queries were repeatable questions answerable from existing documentation. Those were the automation targets.
AI chatbot build and trainingBuilt a RAG-based chatbot (retrieval-augmented generation) trained on Meridian Advisory's service documentation, FAQ content, fee structures, and process guides. The chatbot was designed to answer the firm's most common inbound questions with authority, escalate to a human where it detected complexity or urgency, and never fabricate information outside its knowledge base. Deployed as an embedded widget on the website's contact and services pages.
Intake triage automationBuilt an n8n workflow that connects the website contact form to a qualification logic layer. New enquiries are automatically scored against criteria the firm defined — service type, client type, enquiry complexity — and routed to the right advisor's inbox with a pre-populated summary. No more manual triage decisions. Advisors open their inbox and see qualified, sorted leads, not a raw pile of unprocessed form submissions.
Automated scheduling integrationConnected Calendly to the post-triage workflow so that once an enquiry clears the qualification filter, a personalised scheduling link is automatically sent to the prospective client. The link only shows availability for the relevant advisor. Average time from enquiry to booked consultation dropped from 3.2 days to same-day for eligible leads.
Human escalation and override protocolsBuilt explicit escalation paths into every automation layer. The chatbot flags and escalates when it detects distress, legal complexity, or queries outside its training scope. The triage workflow sends high-urgency leads directly to a principal's inbox with an alert rather than through the standard queue. Every automation has a visible handoff point so the team always knows what the system is doing and can override it in one click.
Testing, tuning, and staff onboardingRan two weeks of parallel testing before switching off manual handling — the automations ran alongside the old process so every output could be spot-checked. Staff training covered how to monitor the dashboard, how to add new Q&A content to the chatbot knowledge base, and how to adjust triage routing rules when service offerings change.
In the first four weeks post-deployment, the chatbot handled 68% of all inbound queries without human intervention. Of those, fewer than 4% required a follow-up correction or escalation — the training coverage was strong enough that the bot was reliably accurate on the queries it was fielding.
The team's weekly time-on-admin dropped from 14 hours to roughly 3.5 hours — the remaining manual time covering genuinely complex or sensitive conversations that warranted a human. That 10+ hours recovered per week has been redirected into billable client work.
The triage automation has also improved lead quality. Before, advisors were spending time on enquiries that weren't a genuine fit before realising that fact. Now, poorly-matched enquiries are identified early and routed to a polite, automated response explaining the firm's focus — freeing advisor time for leads with real potential.
The scheduling integration cut the average time to booked consultation from 3.2 days to same-day for qualified inbound leads, which has had a secondary effect on conversion: fewer leads are going cold between enquiry and first conversation because the friction has been removed.
"We were sceptical about AI handling client-facing communication — reputation matters in our sector and we can't afford to sound like a bot. But the chatbot was trained well enough that we've had clients come to their first consultation having already formed a positive impression from the website interaction. The time saving was real from week one."— Managing Director, Meridian Advisory | Professional Services, Sydney NSW
30-minute discovery call. We'll do a quick time audit with you, identify your highest-value automation targets, and tell you what's realistic to build and how long it takes.
