Case Study: How a Denver Client Saved 20 Hours a Week with AI Workflows

Client details have been generalized for privacy.

The client

A 22-person professional services firm headquartered in Denver, with a mix of partners, associates, and admin staff. Strong revenue, growing client roster — but the team was drowning in repetitive document work and inbound triage.

The problem

Three high-volume workflows were eating most of the team's "non-billable" time:

• New client intake (forms, document collection, conflict checks)
• Initial document review and clause extraction
• Inbound email triage and routing

A baseline audit estimated the team was spending roughly 30 hours per week across the firm on the manual portions of these three flows.

What we did

Over a 7-week engagement we delivered:

• An AI-powered intake portal that captured client data, requested missing documents automatically, and pre-populated the firm's matter system.
• An AI document reviewer that read incoming documents, extracted key clauses and dates, and flagged anything outside the firm's standard playbook.
• An email triage agent that routed inbound mail into the right matter folder and drafted first responses for repeat questions.

Each piece used a combination of off-the-shelf AI (Claude for the document reviewer, ChatGPT for the email drafting) plus custom integration into the firm's practice management system.

The result

• ~20 hours per week saved across the firm within 60 days of launch.
• Average intake time reduced from 4 days to ~36 hours.
• Document review accuracy remained at or above the human baseline once the AI playbook was tuned.
• The partner team reclaimed enough time to take on a meaningful number of additional matters per quarter.

What made it work

• Workflow first. We didn't start with "let's add AI." We started with the three highest-cost manual flows.
• Pilot first. Each of the three workflows went live as a small pilot before being rolled out firm-wide.
• Training in context. Associates learned the new tools using their own active matters, not a generic demo.
• Measurement built in. Hours saved, accuracy, and turnaround time were tracked from day one — not bolted on later.

Week-by-week: how the engagement actually ran

The 7-week timeline is worth walking through, because it's representative of how most of our small business engagements unfold:

• Week 1 — discovery and audit. We interviewed the partners, two associates, and the firm's office manager. We shadowed two full days of intake and email triage. We pulled a baseline measurement of hours spent on the three target workflows.
• Week 2 — design and SOW finalization. We mapped each of the three workflows from current state to target state on a single page. The firm signed off on what was in scope and, just as importantly, what was out of scope (no document drafting from scratch, no CRM replacement).
• Weeks 3–4 — build of the intake portal. The intake portal was the largest piece, so it went first. We integrated with the firm's existing matter system rather than replacing it, and stood up an AI agent that requested missing documents from clients automatically.
• Week 5 — build of the document reviewer. Claude was the right model for nuanced clause extraction. We tuned a custom playbook against 30 historical documents the firm provided.
• Week 6 — build of the email triage agent. We added an AI agent inside the firm's existing email system to categorize, route, and draft responses. Drafts were never auto-sent — every reply was reviewed by a human before going out.
• Week 7 — pilot and training. We ran each workflow with a single partner-and-associate pair for the full week, recorded the issues, fixed them, and held a firm-wide rollout session at the end of the week.

What surprised the team

Three things consistently surprised the team in the first 30 days. First, the document reviewer was more consistent than the human baseline on standard clauses, but weaker on edge cases — exactly as we'd predicted, but it took seeing it firsthand for the partners to trust the split-of-work. Second, the email triage agent was a much bigger morale win than anyone expected; the team's "Monday morning inbox dread" effectively disappeared. Third, the intake portal compressed the new-client experience in a way clients themselves commented on within the first two weeks.

What we'd do differently

Honest reflection from our side: we under-budgeted training time. We planned a single 2-hour workshop and ended up running three additional 60-minute office hours sessions in the first month. For future engagements of this size, we now bake those office hours into the SOW rather than treating them as overflow. The firm got the same end result, but we'd rather build it into the plan than scramble for it.

The numbers, 6 months later

Six months after launch, the firm sent us its own retrospective. Hours saved had ticked up from 20 to closer to 24 per week as the team got more comfortable handing edge-case documents to the AI reviewer. The intake portal had reduced pre-engagement churn (clients dropping off before signing) by an estimated 18%. The email triage agent had drafted more than 4,000 responses, of which the team sent ~70% with light edits. None of those numbers existed before the engagement, because nobody was measuring this work as a system. Putting AI on top of it forced the firm to start.

Could we do this for your business?

If your team is doing repetitive document, intake, or triage work — almost certainly yes. The same playbook applies across professional services, construction, healthcare, logistics, and SaaS. Read our workflow automation services page for how we scope these engagements, or book a free discovery call and we'll tell you honestly whether we'd recommend an engagement.