How One Automation Saved 11 Hours Per Week

Here’s the exact workflow, step by step.

Most articles about AI automation are written to impress. They use vague language, aspirational claims, and carefully curated results. This one is written to inform. We’re going to walk through a real engagement — what the firm looked like before, what we built, exactly how it works, and the measured results.

The firm details have been anonymised for confidentiality, but the workflow, the numbers, and the process are real.

The Firm

A small litigation practice in New South Wales. Three lawyers (one principal, one senior associate, one junior), two paralegals, and a part-time practice manager. Total staff: six. Annual revenue in the mid-seven figures. They handle personal injury matters with a focus on personal injury and complex litigation — birth injuries, surgical errors, delayed diagnosis, and aged care neglect.

They came to us with a specific complaint: their intake process was drowning them.

The Problem: Intake Was Consuming 17.5 Hours Per Week

The firm receives approximately 30 new enquiries per week. These arrive through multiple channels: email, phone calls, their website contact form, and occasionally a direct message on social media. Each enquiry needs to be captured, assessed, responded to, and either converted into a new matter or declined.

When we mapped the process, the time per enquiry averaged 35 minutes. That number surprised them — they’d estimated 15 to 20 minutes. The discrepancy was because nobody had ever timed the full process end-to-end, including all the interruptions, context switches, and follow-ups.

Seventeen and a half hours per week. That’s more than two full working days — every week — consumed entirely by the intake process. Split between one paralegal (who handled roughly 60% of the volume) and the senior associate (who reviewed every enquiry for viability and handled the more complex ones directly).

The Before Workflow: Six Steps, Thirty-Five Minutes

Here’s exactly what the intake process looked like before automation, broken down by step.

Step 1: Enquiry Arrives (0 minutes of processing, but variable delay)

A potential client contacts the firm. Email, phone, or web form. Phone messages get transcribed by whoever answers, usually with incomplete details. Emails sit in a shared inbox. Web form submissions arrive as email notifications. There’s no unified view of incoming enquiries — just multiple inboxes being checked by multiple people at variable intervals.

The first problem: response time. Because enquiries weren’t centralised, the average time from enquiry to first human acknowledgement was 4.2 hours during business hours and up to 18 hours for anything received after 5pm. In personal injury, where potential clients are often contacting multiple firms simultaneously, that delay was costing them cases.

Step 2: Manual Data Entry (8–12 minutes)

The paralegal opens the enquiry and manually enters the details into the practice management system. Name, contact details, date of incident, treating facility, nature of injury, current treatment status. If the enquiry came via phone, she’s working from handwritten notes with missing details. If it came via email, she’s copying and pasting between windows.

Error rate on manual data entry: the firm estimated around 5%. We measured it at 11% — missing fields, transposed numbers, misspelled names. Each error creates downstream problems: letters returned, follow-ups sent to wrong contacts, conflicts checks returning false negatives.

Step 3: Senior Lawyer Review (5–8 minutes)

Every enquiry is flagged for the senior associate to review. She opens the file, reads the details, and makes a preliminary viability assessment. Is this a matter the firm handles? Is there an identifiable breach? Is the injury significant enough to justify the cost of pursuing? Are there limitation concerns?

This step cannot be skipped or delegated — it requires legal judgement. But most of it is pattern recognition that follows consistent rules: the firm has clear criteria for what it will and won’t take on. The senior was effectively applying a checklist, matter after matter, thirty times a week.

Step 4: Initial Response (5–7 minutes)

The paralegal drafts and sends an initial response. If viable, it’s a letter inviting the client to provide further information. If not viable, it’s a decline letter with a suggestion to contact the Law Society for a referral. Both are based on templates, but each requires manual customisation — inserting names, dates, specific details about the enquiry.

Step 5: Document Requests and Follow-Up (8–12 minutes, spread over days)

For viable matters, the paralegal sends document request letters to treating providers, requests authority forms from the client, and adds follow-up reminders to her diary. She manually tracks what’s been requested and what’s outstanding in a spreadsheet. When a follow-up falls due, she checks the spreadsheet, checks the file, and sends a chaser — or forgets, because she’s managing the same process across 30+ active enquiries.

Step 6: File Opening (4–6 minutes)

For matters that proceed, the paralegal creates a new file in the practice management system, generates a costs agreement from a template, assigns a file number, sets up the folder structure in the DMS, and notifies the responsible lawyer. Each step is manual. Each step is necessary. None of it requires legal reasoning.

What We Built: The After Workflow

The automated system we deployed replaced or augmented every step of the intake process. Here’s exactly how each step works now.

Step 1: Unified Enquiry Capture (Automatic)

All enquiry channels — email, web form, phone — now feed into a single AI-powered intake system. Email enquiries are parsed automatically. Web form submissions are ingested directly. Phone calls are transcribed by the firm’s receptionist into a simple structured form (name, phone number, one sentence about their matter) and submitted to the same system.

The AI agent reads each enquiry, regardless of format, and extracts the key facts: client name, contact details, date of incident, treating facility, nature of injury, and any other relevant information mentioned. It handles messy, unstructured emails just as well as clean web form submissions.

Response time from enquiry to automated acknowledgement: under 3 minutes. The client receives a professional, personalised email confirming receipt and explaining next steps — generated from an approved template, populated with their specific details, sent automatically.

Step 2: Automated Data Entry and Conflicts Check (Automatic)

The extracted data is entered directly into the practice management system via API integration. No manual copying. No transposition errors. No missing fields — the system flags any enquiry where key information couldn’t be extracted and routes it for manual completion.

Simultaneously, the system runs an automatic conflicts check against the firm’s existing client and party database. Any potential conflict is flagged immediately with the matching records displayed for human review.

Step 3: AI Viability Assessment (Automatic, with Lawyer Override)

This was the most sophisticated part of the build — and the part the firm was most sceptical about. The AI assesses each enquiry against the firm’s viability criteria: matter type (does it fall within the firm's practice areas?), apparent severity of injury, potential limitation issues, identifiable treating provider, and general merit indicators.

The system doesn’t make decisions. It produces a viability score and summary for each enquiry, which appears on the senior associate’s dashboard. She can review a pre-assessed enquiry in 60 to 90 seconds instead of 5 to 8 minutes — because the work of reading, extracting, and categorising has already been done. She just confirms or overrides the assessment.

In the first three months, the system’s viability assessment agreed with the senior associate’s judgement 91% of the time. The remaining 9% were borderline matters where the AI correctly flagged uncertainty rather than making a definitive call.

Step 4: Automated Response (Automatic, Template-Approved)

Based on the viability assessment (confirmed by the senior associate with a single click), the system generates and sends the appropriate response. Viable matters receive an engagement letter with next steps. Non-viable matters receive a professionally written decline with referral suggestions.

Every response is generated from firm-approved templates, populated with matter-specific details, and logged to the file automatically. The senior associate can review any outgoing correspondence before it’s sent — but in practice, after the first month, she stopped reviewing the standard responses because they were consistently accurate.

Step 5: Automated Document Requests and Tracking (Automatic)

For viable matters, the system automatically generates document request letters to the treating providers identified in the enquiry, sends authority forms to the client for signature, and creates a tracking record for each outstanding request.

Follow-ups are automated on a configurable schedule: if a requested document hasn’t been received within 14 days, a chaser is sent automatically. If it’s still outstanding at 28 days, the system escalates to the paralegal for manual follow-up. No spreadsheet. No diary entries. No forgotten chasers.

Step 6: Dashboard Review (2–3 minutes of human time)

The senior associate’s morning now begins with a dashboard showing all new enquiries from the previous 24 hours, each with a viability score, extracted key facts, conflicts check result, and recommended action. She reviews the list, confirms or adjusts the assessments, and approves the responses.

Thirty enquiries that used to consume 17.5 hours of combined staff time per week are now processed with approximately 1.5 hours of total human involvement per week — almost entirely the senior associate’s review time.

The Results: 11 Hours Per Week Recovered

We measured the results over a 12-week period following deployment.

But intake processing was only part of the picture. The automated follow-up system, document tracking, and file opening automation saved an additional 5 hours per week of paralegal time that had previously been consumed by chasing outstanding documents, updating tracking spreadsheets, and performing repetitive file setup tasks.

However, the new system introduced approximately 1.5 hours per week of dashboard review time for the senior associate (time she hadn’t spent before in this specific format) and roughly 30 minutes per week of system oversight by the practice manager. Net additional time: 2 hours.

Subtracting that from the combined savings:

We report the headline figure as 11 hours per week because that’s the conservative number the firm’s principal agreed to after excluding time savings that were harder to measure precisely (fewer interruptions, reduced context-switching, less time spent on error correction). The true measured figure was higher, but we prefer to understate.

The Annual Value

Two hundred thousand dollars in productive capacity recovered annually. From automating a single workflow at a six-person firm.

The Unexpected Benefits

The time savings were the primary objective. But three months into operation, the firm reported several outcomes they hadn’t anticipated.

No More Lost Enquiries

Before automation, the firm estimated they were “probably” losing some enquiries — messages that came in on a busy Friday afternoon and weren’t actioned until Monday, by which time the potential client had engaged another firm. They couldn’t quantify it because they didn’t know what they were missing.

After automation, every enquiry receives an acknowledgement within minutes, regardless of when it arrives. The principal reported that two new clients specifically mentioned the fast response as a reason they chose the firm. At an average case value of $80,000 to $120,000 in fees, even one additional converted client per quarter represents significant revenue.

Consistent Data Quality

The 11% error rate on manual data entry dropped to effectively zero on AI-extracted fields. When the system can’t extract a field with confidence, it flags it for human input rather than guessing. The downstream benefits — accurate conflicts checks, correct correspondence addresses, reliable reporting — eliminated a category of small, persistent problems that had been consuming time across the firm.

Better Viability Decisions

With structured viability assessments on every enquiry, the firm noticed that the senior associate was making more consistent accept/decline decisions. Previously, her assessment depended partly on how much time she had when she reviewed the enquiry. On busy days, she’d spend less time on each one. The AI pre-assessment removed that variability — every enquiry received the same level of analysis regardless of workload.

What It Cost

Full transparency.

In year two and beyond, with the build cost amortised, the ongoing cost is $6,000 per year against $200,000+ in recovered value. An annual return of approximately 33x.

What This Means for Your Firm

Your intake process may look different from this firm’s. Your volume may be higher or lower. Your channels may be more or fewer. But the pattern is almost always the same: multiple manual steps, spread across multiple people, consuming hours that could be spent on legal work, with errors and delays that cost you cases you never know you’ve lost.

We showed you this case study in full detail for a reason. Not to impress you with numbers, but to show you that the process is straightforward, the results are measurable, and the investment is modest relative to the return.

Every firm has an equivalent. A workflow that consumes disproportionate time relative to its value. A process that everyone knows is inefficient but nobody has the bandwidth to fix. A bottleneck that constrains growth without anyone quite realising it.

The first step is identifying it. That’s what the free workflow audit is for.