The workflow before#
This fast-growing accounting and advisory firm set a goal of doubling recurring revenue in two years without growing its back office at the same pace. Its tools were solid, Microsoft 365, HubSpot, DocuSign, Sage Intacct and BQE Core, but nothing connected them. A proposal took about an hour to assemble by hand. Contracts took two days to get through signature. Onboarding a new client meant 30 days of tasks tracked in spreadsheets, client credentials included. Billing errors sometimes reached clients before anyone caught them. Advisors spent hours on email and admin instead of client work, and the only way to grow was to hire.
What we built#
We built an automation layer across the tools the firm already owned instead of replacing them. A discovery assessment picked the six processes with the biggest payoff, and we automated them in three phases over 12 months: proposals and contracts first, then email and onboarding, then billing checks and financial analytics. Together they return more than 1,800 hours a year.
Proposals in 15 minutes
The system pulls opportunity details from HubSpot, extracts what the client needs and drafts the proposal in the firm's format, so a partner reviews and sends it in under 15 minutes instead of an hour. Accepted proposals flow into DocuSign, and contracts now clear in under 2 hours instead of 2 days. We started here because faster proposals bring revenue in sooner, which built the case for the later phases.
An email assistant for routine replies
An assistant built on retrieval-augmented generation reads incoming email, finds the relevant client and engagement context, and drafts replies to routine requests. Staff review and send, and email handling time dropped 40%. We kept a person on every send because client communication at an accounting firm carries professional responsibility.
Onboarding with a secure credential vault
Onboarding steps that lived in spreadsheets now run as an orchestrated workflow in n8n, and a secure credential vault replaced the spreadsheet of client logins. Onboarding takes 15 days instead of 30. The vault also moved the firm closer to SOC 2 readiness, which matters to the larger clients it wants to win.
Billing checks before invoices go out
Machine learning models on Azure ML compare each invoice against billing patterns and flag anomalies before the client sees them, and the firm expects billing errors to fall by about 80%. Real-time pipelines feed Power BI so partners can watch financial trends as they happen. The models flag and a person decides, so no invoice changes without someone approving it.
How we delivered it
The program ran 12 months in three four-month phases. Phase one covered proposals, contracts and the DocuSign integration. Phase two delivered the email assistant, the onboarding workflow and the credential vault. Phase three added invoice quality checks and financial trend analysis. Each phase put working automations into daily use, and we ran training workshops and wrote documentation so the firm's own staff could run and adjust them.
Before and after#
| Measure | Before | After |
|---|---|---|
| Proposal creation | About 60 minutes | Under 15 minutes |
| Contract processing | About 48 hours | Under 2 hours |
| Client onboarding | 30 days | 15 days |
| Email management time | Baseline | 40% less |
| Billing errors | Caught after invoices went out | Flagged before sending, about 80% fewer expected |
What's next#
With six processes automated and its systems connected, the firm can take on more clients without matching headcount. The next step is pointing the same approach at advisory work: using the real-time financial trend data to spot client issues early, so advisors call with an answer before the client asks.
