AI & Workflow Automation
← Back to Rocinante Systems

Stop paying peopleto do whatsoftware can.

We design and deploy n8n workflows and AI agents that take over the slow, repetitive work eating your team's time — so they can focus on the work that actually needs a human.

Hours → Minutes
typical task cycle time after automation
60–90%
reduction in manual operations cost
24 / 7
your workflows run while your team sleeps
Real scenarios

What we actually build

Every automation starts with a real process. Here are six common ones we've designed — with the exact workflow steps behind each.

Sales / Marketing

Inbound lead enrichment & CRM routing

A sales team receives 80–150 leads per day via web forms, LinkedIn, and email. Each one needs to be qualified, researched, scored, and entered into the CRM — a task that takes SDRs 4–6 minutes per lead.

Before

2 SDRs spend ~5 hrs/day doing nothing but copy-pasting lead data.

After automation

Every inbound lead is automatically enriched via LinkedIn/Clearbit, scored by an AI model against your ICP, routed to the right rep in HubSpot or Salesforce, and a personalised first-touch email drafted — ready to send with one click.

~9 hrs/daysaved across the team
Workflow steps
1
Lead arrives via form / email / LinkedInTrigger
2
Enrich with Clearbit + LinkedIn dataHTTP / API
3
AI scores against ideal customer profileClaude / GPT-4
4
Route to correct rep based on territoryn8n Logic
5
Create CRM contact + draft outreach emailHubSpot / Salesforce
6
Notify rep via Slack with full summarySlack
Finance / Operations

Invoice & document processing pipeline

An operations team receives 200+ invoices per week across PDF, email, and supplier portals. Finance staff manually open each one, extract line items, check against POs, and enter into the accounting system.

Before

3 hours of manual data entry per day, high error rate, late payment penalties.

After automation

Documents are automatically parsed by AI, key fields extracted (vendor, amount, due date, line items), cross-referenced against open POs, and pushed into Xero or QuickBooks. Exceptions flagged for human review only.

~15 hrs/weekof finance team time
Workflow steps
1
Invoice arrives via email or uploadGmail / Webhook
2
Extract PDF text + structure with AIClaude / GPT-4 Vision
3
Parse vendor, amounts, line itemsStructured output
4
Match against open purchase ordersDatabase / ERP
5
Push approved invoices to accountingXero / QuickBooks
6
Flag mismatches for human reviewSlack / Email alert
E-commerce / SaaS

Customer support first-response agent

A SaaS company receives 300+ support tickets per day. 70% are variations of the same 20 questions. Support staff spend most of their time writing responses to questions the knowledge base already answers.

Before

Average first response time: 6 hours. Support team burned out on repetitive queries.

After automation

An AI agent reads every incoming ticket, classifies intent, searches the knowledge base, and drafts a contextual response. Simple questions are resolved automatically. Complex ones get a drafted reply and summary for the human agent.

70% of ticketsresolved without human touch
Workflow steps
1
Ticket arrives (email, Zendesk, Intercom)Trigger
2
Classify intent + sentiment + urgencyClaude / GPT-4
3
Search knowledge base + past ticketsVector DB / RAG
4
Draft personalised responseLLM generation
5
Auto-send if confidence high; else queuen8n Logic
6
Log resolution + update KB if new patternDatabase
Strategy / Consulting

Daily competitive intelligence briefing

A strategy team needs to stay on top of competitor moves, industry news, job listings, and regulatory changes. Currently a junior analyst spends 2 hours each morning aggregating and summarising sources.

Before

2 hrs/day of analyst time, delayed by timezone, inconsistent coverage.

After automation

An agent runs every morning: scrapes configured sources, filters for relevance, runs AI summarisation, and delivers a structured briefing to Slack or email before the team starts their day.

~10 hrs/weekanalyst time reclaimed
Workflow steps
1
Scheduled trigger (06:00 daily)n8n Cron
2
Scrape competitor sites, news, job boardsHTTP / RSS
3
Filter for relevance to watchlistAI classification
4
Summarise and extract key signalsClaude / GPT-4
5
Format into structured briefingTemplate engine
6
Post to Slack channel + email digestSlack / Email
Operations / BI

Cross-system data sync & weekly reports

Operations managers pull data from Shopify, Google Sheets, PostgreSQL, and a third-party logistics API every Monday to produce a performance report. It takes 3–4 hours and is always slightly out of date.

Before

Manual Monday morning ritual. Errors from copy-paste. Always late.

After automation

A scheduled workflow pulls fresh data from all sources, transforms and joins it, generates a formatted report with AI-written commentary on trends and anomalies, and delivers it to stakeholders before 8am.

~3.5 hrsevery Monday morning
Workflow steps
1
Trigger on schedule (Sunday 23:00)n8n Cron
2
Pull data from Shopify, Sheets, Postgres, APIHTTP / DB nodes
3
Transform, join, and aggregaten8n Code node
4
AI writes trend summary + anomaly flagsClaude / GPT-4
5
Render report (PDF or Google Doc)Template / Drive
6
Email to stakeholders + post to SlackGmail / Slack
HR / Recruiting

CV screening & candidate shortlisting

A recruiting team receives 200–400 CVs per open role. Initial screening to create a shortlist takes a recruiter 6–8 hours. Quality varies and unconscious bias creeps in.

Before

Full day of recruiter time per role just to get to a shortlist.

After automation

CVs are parsed by AI, mapped against the job requirements, scored on structured criteria, and ranked. A shortlist with per-candidate rationale is delivered to the recruiter in minutes. They only read the top candidates.

~7 hrsper open role
Workflow steps
1
CVs received via email or ATS uploadGmail / Webhook
2
Extract and parse CV contentGPT-4 Vision / Parser
3
Score against role requirements rubricClaude structured output
4
Rank candidates + write rationale per CVLLM generation
5
Build shortlist report with scoresn8n + Template
6
Deliver to recruiter via email + ATS noteEmail / Greenhouse

Don't see your use case? Most operational processes can be automated.

Tell us what you're doing manually today
How it works

From messy manual process to running workflow — in weeks.

We've done this enough times to have a process that works. Here's what working with me looks like.

00

Free intro call

Free · 30 min

A no-commitment call to understand your situation, what you're doing manually today, and where automation might help. If there's a good fit, we'll talk next steps.

Deliverable:Clarity on whether we're a good fit
01

Automation audit

€500 · 1–2 days

A paid, structured deep-dive into your workflows. We document the highest-ROI automation targets, estimate time and cost savings, and deliver a prioritised roadmap you own — regardless of whether you hire me to build it.

Deliverable:Priority automation roadmap + ROI estimates
02

Workflow design

1–5 days

We map the exact process end-to-end: triggers, data sources, decision logic, AI prompts, error handling, and human-in-the-loop checkpoints.

Deliverable:Annotated workflow diagram + spec
03

Build & integrate

1–4 weeks

We build the workflow in n8n, wire up all integrations (your CRM, email, database, APIs), write and tune the AI prompts, and connect everything to your existing tools.

Deliverable:Working automation in your environment
04

Test & tune

1 week

We run the workflow against real data, measure accuracy, adjust AI prompts and logic, and stress-test edge cases before handing over.

Deliverable:Validated automation with documented edge cases
05

Handover & support

Ongoing optional

We document everything, train your team on monitoring and adjustments, and offer optional ongoing support packages for maintenance and expansion.

Deliverable:Runbook + training + optional retainer
Tools & models

We work with what you already use.

Model-agnostic and stack-agnostic. We choose the right AI model for each task and connect to the tools your team already relies on.

Orchestration

n8nPrimary
Primary orchestration layer
Self-hosted or cloud. Connects everything.
Webhooks & REST APIs
Trigger & integration layer
Any service with an API can be connected.
Cron / Scheduling
Time-based triggers
Run workflows hourly, daily, weekly, or custom.

AI models

Claude 4.6 SonnetPreferred
Anthropic
Reasoning, long-context, tool use
GPT-4o
OpenAI
Vision, function calling, speed
Gemini 1.5 Pro
Google
Very long context, multimodal
Llama 3.1
Meta / local
Self-hosted, no data leaves your premises
Mistral Large
Mistral AI
EU-hosted, strong reasoning
Ollama models
Various
Local deployment, full control

Integrations (sample)

HubSpotSalesforcePipedriveGmailOutlookSlackNotionAirtableGoogle SheetsXeroQuickBooksStripeZendeskIntercomFreshdeskPostgreSQLMySQLSupabaseShopifyWooCommerceWebflowGreenhouseWorkableLinear+ 400 more via n8n

If it has an API or webhook, we can connect it. Custom integrations built where native connectors don't exist.

FAQ

Questions we get asked.

Honest answers. If yours isn't here, email me — I respond quickly.

Ask a question →
Get started

Let's talk about your workflows.

Start with a free 30-minute intro call — no commitment, just a conversation about where you're losing time. If there's a clear fit, the next step is a structured automation audit at €500.

Usually responds within 24 hours

Automation Audit
€500 · 1–2 days · Yours to keep

A paid, structured analysis of your operations. Delivered as a document you own — whether or not you hire me to build anything.

  • Deep-dive into your current workflows
  • Identify highest-ROI automation targets
  • Estimate realistic time and cost savings
  • Deliver a prioritised roadmap you own

The audit stands on its own. Most clients use it to get internal buy-in before committing to a build — and some just take it and run with it themselves.