19 Hours of Manual Reviews.
Down to 2. Every Week.
A services business was spending nearly half their operations assistant's working week manually reviewing employee time recordings in Harvest. Quixas built an AI-powered review pipeline using n8n and Claude that now handles 100% of recordings automatically, routing only the 10% that genuinely need human judgment to the assistant.
Industry
Services / Operations
Solution
AI-Powered Review Automation
Stack
n8n + Claude AI + Harvest API
Time Saved Per Week
17 hrs
From 19 hours manual review to 2 hours
From 19 hours manual review to 2 hours, every week.
89%
Time reduction
~10%
Need human review
19→2
Hours per week
Before and after automation
0%
Time reduction
Measured post-deployment
0%
Recordings reviewed
By AI on every cycle
~10%
Need human review
Escalations only
Half a Working Week.
On a Single Repetitive Task.
19 hours every week spent on manual recording reviews
The operations assistant was exporting Harvest recordings, reviewing each entry for completeness, accuracy, correct project tagging, and policy compliance, then drafting individual messages to employees whose entries needed clarification. Nearly half their working week, every week.
Inconsistent reviews at high volume
Manual review at this scale introduced inconsistency. The same type of entry might be flagged one week and missed the next, depending on fatigue or workload. There was no standardised evaluation criteria applied consistently across every recording.
Slow feedback loops for employees
When an entry needed correction, employees waited for the assistant to work through the review cycle before being notified. By that point, the context for the entry was often days old, making corrections slower and less accurate.
An AI That Reviews
Every Entry. Routes the Rest.
Quixas built a five-stage automated review pipeline using n8n as the orchestration layer and Claude as the intelligent review agent. The system ingests every Harvest recording, evaluates it against the client's criteria, and routes only genuinely ambiguous cases to the human assistant.
AI-Powered Recording Review
Claude evaluates every Harvest entry across multiple dimensions: description completeness, time reasonableness, project accuracy, and policy compliance. Each entry receives a structured rating and a confidence score. No entry is skipped, no entry is reviewed inconsistently.
Automated Classification and Routing
Based on Claude's output, entries are classified into three categories: Approved (auto-cleared), Flagged (employee notified automatically), and Escalated (routed to the assistant with full context). Only the roughly 10% of genuinely ambiguous entries reach a human.
Multi-Channel Notifications and Reporting
Employees receive immediate Slack or email alerts when their entries are flagged, with specific guidance on what needs correcting. The assistant gets a daily digest of escalations. Management receives automated weekly reports covering approval rates, common issues, and compliance trends.
Five stages. Fully automated.
Automated Data Ingestion
n8n triggers on schedule, pulls all new Harvest recordings via API. Entries normalised with employee metadata, project details, and historical patterns.
AI Review by Claude
Each recording sent to Claude with the client's review criteria and rating rubric. Evaluated across description completeness, time reasonableness, project accuracy, and policy compliance.
Classification and Routing
Entries automatically classified: Approved, Flagged, or Escalated. Only escalated entries land on the assistant's desk.
Notifications and Alerts
Employees notified immediately on flags. Assistant receives daily escalation digest. Management gets weekly compliance reports, all automated.
Human-in-the-Loop Resolution
Assistant reviews only genuinely ambiguous cases, with Claude's analysis and confidence score presented alongside the original entry for fast, informed decisions.
Built on n8n (self-hosted) · Claude AI (Anthropic) · Harvest API · Slack / Email
89% Less Time.
Same Quality. Better Consistency.
89% reduction in weekly review time
From 19 hours of manual review every week to 2 hours of human escalation review. Approximately 8 hours per month returned to the operations assistant for higher-value work.
Consistent, standardised evaluation on every entry
Every recording evaluated against the same criteria every time, no variation based on fatigue, workload, or inconsistent interpretation. The same type of entry is treated the same way on every review cycle.
Employees notified immediately on flagged entries
Automated Slack and email alerts fire the moment an entry is flagged, with specific guidance on what needs correcting. No waiting for the assistant to work through the review queue.
Management visibility without manual reporting
Weekly automated reports delivered to management covering total recordings reviewed, approval rates, common issues, and compliance trends, compiled and sent without anyone manually pulling the data.
We were spending 19 hours every week on something that should not have required a person. Now it takes 2. The quality of the reviews is actually better, because the criteria never changes.
Operations Director
Services Business · Harvest Automation Client
Client name withheld pending approval.
Which Repetitive Review Process
Is Consuming Your Team This Week?
In 30 minutes we map the process, calculate exactly how many hours it is costing per week, and show you what an automated version looks like, before you commit to anything.