DigiFusion
DigiFusion

Agency — Case Studies

Automation that pays for itself

Three engagements. Three industries. One consistent result: small teams doing significantly more — without adding headcount or replacing the tools they already use.

Operations AutomationLogistics & FreightA mid-sized freight brokerage (12 employees)

From 4 Hours of Daily Admin to a 15-Minute Morning Review

The Challenge

Every morning the operations team spent 3–4 hours copy-pasting load confirmations from carrier emails into their TMS, updating a shared Google Sheet, and manually emailing status updates to shippers. Errors were common — the wrong load number on a confirmation caused a $4,200 invoice dispute that took two weeks to resolve. The team knew the process was broken but assumed fixing it required an expensive TMS upgrade.

What We Built

We built a three-stage AI pipeline that runs on their existing email inbox — no TMS change required. First, a document-parsing agent reads every inbound carrier confirmation PDF and email, extracts load number, pickup/delivery addresses, driver name and ETA with 98% accuracy. Second, a rules engine matches each confirmation to the correct load record in their TMS via API and pushes the update automatically. Third, a summarisation agent composes and sends a plain-English status email to the shipper the moment a milestone is hit — pickup, in-transit, delivered.

“I was convinced we needed to replace our whole TMS. Turns out the problem was never the software — it was the manual bridge between our inbox and the software. That bridge is gone now.”

Operations Manager

Results

  • Daily admin time reduced from ~3.5 hours to under 15 minutes (a 93% reduction)
  • Confirmation-to-TMS update lag dropped from 2–6 hours to under 3 minutes
  • Invoice disputes caused by data-entry errors fell to zero in the 6 months post-launch
  • The team redirected saved time to prospecting, adding 2 new carrier relationships in Q1

Stack Used

  • n8n workflow automation
  • OpenAI GPT-4o (document parsing)
  • Custom TMS REST API connector
  • Gmail + Resend for email delivery

Timeline: 6 weeks from discovery to go-live

Document & Client WorkflowProfessional ServicesA boutique employment law firm (3 solicitors, 2 paralegals)

Cutting Client Intake from 3 Days to 90 Minutes — Without Hiring

The Challenge

New client intake was a bottleneck. A potential client would email or call, the paralegal would schedule a conflict check, manually prepare an engagement letter from a template, chase the client for a signed copy, then re-enter the same information into the case management system. The firm was losing prospective clients who needed fast turnaround: two confirmed losses in six months totalling roughly £18,000 in fees. Senior solicitors were also getting pulled into admin to keep things moving.

What We Built

We redesigned intake as a fully automated sequence. A Typeform collects the prospective client's details and matter type. That submission triggers an AI agent that runs a conflict check against the firm's existing client database, generates a bespoke engagement letter using matter-specific clause blocks, sends it via DocuSign for e-signature, and — on receipt of the signed letter — creates and populates the case record in Clio automatically. The solicitor receives a single Slack notification with a summary. If the conflict check flags a potential issue, the case is routed to a human for review before the letter is sent.

“The bit that surprised me most was the conflict check. I assumed that had to be manual. Turns out it just needed to be structured correctly — and now it is.”

Managing Solicitor

Results

  • Average intake cycle shortened from 3 working days to under 90 minutes
  • Paralegal time per new matter reduced from ~4.5 hours to ~20 minutes (admin only)
  • Zero missed conflict checks in 8 months post-launch (previously 1–2 near-misses per quarter)
  • Firm onboarded 23% more new matters in the 6 months following launch with the same headcount

Stack Used

  • Typeform (client intake)
  • OpenAI GPT-4o (letter generation)
  • DocuSign API (e-signature)
  • Clio API (case management)
  • Slack notifications

Timeline: 8 weeks including compliance review

Customer Operations & ContentE-Commerce & RetailA DTC skincare brand (7-person team, ~$1.4M ARR)

Automating Customer Support and Product Descriptions — Saving 22 Hours a Week

The Challenge

The founder was personally handling ~60 customer service emails per day alongside supplier coordination, paid ads management and new product development. Response times averaged 14 hours, and three one-star reviews in a month all cited slow replies. Meanwhile, the brand had 140 SKUs but only 40% had SEO-optimised product descriptions — the rest were manufacturer copy-pastes. A freelance copywriter quoted $3,200 to fix the catalogue; the budget wasn't there.

What We Built

We deployed two parallel systems. For customer support: a GPT-4o-powered triage agent connected to their Gorgias helpdesk reads every inbound ticket, classifies it (order status / returns / product advice / complaints), drafts a response using brand-voice guidelines we co-wrote, and auto-sends for the three lowest-risk categories. Complex or negative-sentiment tickets are flagged and drafted but held for human review. For content: a batch pipeline ingests each product's ingredient list, claimed benefits and existing user reviews, then generates a structured, SEO-optimised product description with a meta title and meta description, reviewed and published in Shopify via API.

“I used to start every day drowning in emails. Now I open Gorgias once in the morning and once in the afternoon. The AI handles the rest — and it sounds exactly like me.”

Founder & CEO

Results

  • Average first-response time dropped from 14 hours to 38 minutes
  • Customer support time reduced from ~3.5 hours/day to 45 minutes (human review of flagged tickets only)
  • All 140 SKUs received unique, SEO-optimised descriptions within 11 days
  • Organic product page traffic increased 34% in the 90 days after description rollout
  • The founder reclaimed approximately 22 hours per week, redirected to partnerships and new product R&D

Stack Used

  • Gorgias API (helpdesk)
  • OpenAI GPT-4o (triage + drafting)
  • Shopify Admin API (product publishing)
  • Custom brand-voice prompt library
  • Slack (human escalation alerts)

Timeline: 5 weeks (support system: week 1–3; content pipeline: week 3–5)

Work with us

Ready to build yours?

Every engagement starts with a free 45-minute strategy session. We map your highest-value automation opportunities, estimate time savings and give you a clear picture of what a build looks like — before you commit to anything.

Book a strategy sessionSee all services