This week’s AI Team Tuesday post is a full case study rather than a single-workflow deep dive. We’re walking through how Meridian Consulting Group (name changed at client’s request), a 70-person professional services firm specializing in financial and compliance advisory, went from a fully manual operations backbone to an AI-first operating model across three functions in 60 days.
The Starting Point: A Firm Growing Faster Than Its Operations Could Support
Meridian had grown from 40 to 70 employees in 18 months, driven by strong client demand, but their operations infrastructure hadn’t scaled with headcount. The COO described the situation candidly during our initial discovery call: “We’re basically running a 70-person firm on the operational muscle of a 25-person firm, and everyone knows it — we just haven’t had the bandwidth to fix it.”
Three functions were identified as the highest-friction, highest-volume candidates:
H3: Consultant Scheduling and Project Staffing
A single operations manager manually matched consultants to client engagements based on availability, skill set, and utilization targets — tracked in a spreadsheet that was perpetually a day or two out of date.
H3: Client Invoicing
Meridian bills by the hour across dozens of active engagements. Time entries came from six different consultants’ varying habits (some daily, some weekly, some monthly “catch-up” sessions), and invoices were manually compiled by one staff accountant, taking roughly two full days per billing cycle.
H3: Client Support and Status Inquiries
Clients frequently emailed asking for project status updates, which required the account lead to manually pull together information from multiple sources before responding — often taking 24-48 hours.
The Before: By the Numbers
| Function | Before |
|---|---|
| Time to staff a new engagement | 2-3 days |
| Utilization tracking accuracy | Updated weekly, often stale |
| Time to compile monthly invoices | 2 full days per cycle |
| Billing errors requiring client correction | 3-4 per month |
| Client status inquiry response time | 24-48 hours |
The 60-Day Rollout
Rather than deploying all three functions simultaneously, Xact AI staged the rollout — a decision the COO specifically credited with making adoption smoother.
H3: Days 1-14 — Scheduling and Staffing Agents
Following the 14-day deployment process detailed in Week 5, Xact AI deployed a staffing agent team that tracked real-time consultant availability and utilization (pulling directly from calendar and time-tracking data), and proposed optimal staffing matches for new engagements based on skill tags and current utilization.
H3: Days 15-35 — Invoicing Agents
With scheduling live and trusted, the invoicing agent team was deployed next: a Time Aggregation Agent that pulled and normalized time entries regardless of how or when each consultant logged them, a Draft Invoice Agent that compiled client-ready invoices matching each client’s specific billing format requirements, and a Review Agent that flagged anomalies (unusually high hours, missing entries) before invoices went to the staff accountant for final approval.
H3: Days 36-60 — Client Status Agents
Finally, a Status Response Agent was deployed that could pull current project status from the firm’s project management tool and time-tracking data to draft accurate, specific status updates for routine client inquiries, escalating anything requiring judgment (scope changes, delays, difficult conversations) to the account lead.
The After: By the Numbers
| Function | Before | After |
|---|---|---|
| Time to staff a new engagement | 2-3 days | Under 4 hours |
| Utilization tracking accuracy | Stale, weekly | Real-time |
| Time to compile monthly invoices | 2 full days | 3 hours (review only) |
| Billing errors requiring correction | 3-4/month | 0 in the two months since launch |
| Client status inquiry response time | 24-48 hours | Under 2 hours |
What Changed for the People, Not Just the Metrics
The operations manager who had been manually tracking staffing in a spreadsheet moved into a business development support role — work the firm had wanted to resource for over a year. The staff accountant, freed from two days of manual invoice compilation each month, now handles financial reporting analysis that previously got pushed to quarter-end crunches. No one was let go; the firm redirected existing talent toward growth-oriented work instead of hiring reactively to keep up with operational debt.
The COO’s Own Assessment
At the 60-day mark, the COO’s summary was direct: “The honest version is that we were about six months from needing to hire two more operations people just to keep pace with growth. Instead we have a system that scales with us without adding headcount, and it cost less than one of those hires would have in year one alone.”
Why Staged Deployment Worked Better Than “Automate Everything at Once”
Meridian’s team could see, function by function, exactly what changed and why — building trust before the next function was introduced. By the time the Status Response Agent went live in the final phase, the team had two successful rollouts of evidence behind them and effectively no resistance to the third.
What This Case Study Tells Other 50-500 Person Firms
Meridian isn’t unique in facing this gap between growth and operational capacity — it’s the default state for services firms scaling past 50-60 employees. The pattern that worked here — start with the highest-friction function, prove it in shadow mode, then expand — is repeatable regardless of industry, which is why we recommend the same staged approach in the 14-day deployment guide from Week 5.
Want a Rollout Plan Like Meridian’s for Your Firm?
Book a free demo and tell us about your growth trajectory — we’ll help you identify which function to deploy first and map a realistic staged rollout timeline.
[Book Your Demo →]
—