Services / Connect
Automation & AI
Automation on your data. Systems built on the CRM, the documents and the tools you already run, not a chatbot bolted onto the homepage.
every lead arrives with a source
Who it is for. Teams losing hours to manual handoffs between systems, companies whose lead data arrives incomplete or late, and organisations sitting on years of documentation nobody can search.
What we do
CRM and lead operations
- CRM implementation and cleanup (Zoho, HubSpot, Pipedrive)
- Form to CRM pipelines carrying full source attribution, not just a name and an email
- Lead routing, assignment rules and deduplication
- Custom fields, layouts and pipeline design
- Scheduled sync jobs that repair records the real time push missed
- Migration between CRMs without losing history
Workflow automation
- n8n development, self hosted or cloud
- Zapier and Make where they are genuinely sufficient
- Multi step workflows across marketing, sales and operations tools
- Scheduled reporting, alerting and escalation
- Approval flows and internal notifications
- Data enrichment and validation between systems
Operational automation
The unglamorous work that removes recurring incidents.
- Deployment, cache and backup routines
- Automated site health and integrity checks
- Recurring campaign and content workflows
- Monitoring with alerts that reach a person, not a log file
Integrations
- API integrations with Google, Meta, LinkedIn and platform specific endpoints
- Custom middleware and webhook handling
- Data pipelines between marketing, sales and finance systems
- Building against APIs that have no ready made connector
AI on your data
A generic chatbot answers from the public internet. These systems answer from your documents, your CRM and your history.
Connecting your knowledge
Most of the value is already written down somewhere. The work is making it reachable.
- Notion, Obsidian, Confluence, SharePoint, Google Drive
- Slack and email archives
- ClickUp, Asana, Jira
- CRM history and support tickets
- Contracts, policies and internal documentation
Document processing
- OCR and parsing of PDFs, scans and forms
- Extraction from contracts, invoices and structured documents
- Classification and routing of incoming documents
Assistants and agents
- Internal assistants over your own documentation, using retrieval and vector search
- Knowledge search across repositories, past work and institutional memory
- Lead qualification and support assistants connected to your CRM, with real customer context
- Custom agents for defined repetitive tasks
- MCP integrations, connecting your internal systems to the AI tools your team already uses
Where it lives
An assistant nobody opens is not a deliverable.
- Slack apps and bots
- Web interfaces and internal dashboards
- Telegram and WhatsApp Business
- Email automation and inbox triage
Cloud or self hosted
Where your data is allowed to go decides the architecture. We build for both.
Hosted models
When speed and quality matter most and the data permits it. Claude and OpenAI.
Open weight models
When data cannot leave your infrastructure, or when cost at volume makes it the better choice. Llama, Mistral, Qwen, Hermes, running on your own servers.
Hybrid
Where sensitive content is processed locally and everything else uses hosted models.
We will tell you which one your case actually needs, including when the answer is the cheaper one.
How we approach AI
We start with the data, not the model.
If lead sources are wrong and the CRM is half empty, an AI layer will produce confident nonsense. We usually fix the plumbing first.
We build what has a measurable job.
Hours saved, errors removed, response time reduced. If we cannot name the number, we will say the project is not worth doing.
We measure whether the AI is actually right.
Answer quality is evaluated before anything reaches production, and traced afterwards, with full visibility on cost per workflow. The same standard we apply to analytics.
Your data stays yours.
Self hosted where it matters, clear boundaries on what is sent where, and no training on your content.
Nothing is a black box.
Documented, and buildable by another team if you ever move on.
Where this usually starts
Most automation work begins as a smaller problem:
- Leads arrive in the CRM without a source
- Two systems hold different versions of the same customer
- Someone rebuilds the same report every month
- A team keeps answering questions that are already written down somewhere
- Documents pile up faster than anyone can read them
Any of those is a good first conversation.
How we work
| Engagement | Scope | Timeline |
|---|---|---|
| Automation Audit | Map current workflows, identify what to automate first | 5 days |
| CRM Integration | Form to CRM with attribution, routing, sync | 2 to 3 weeks |
| Workflow Build | Scoped automation across your systems | 2 to 4 weeks |
| AI Assistant | Discovery, data preparation, build, evaluation | 4 to 8 weeks |
| Automation Retainer | Monitoring, changes, new workflows | monthly |
- Automation Auditfrom 900 EUR
- CRM Integrationfrom 2,500 EUR
- Workflow Buildfrom 3,000 EUR
- AI Assistantfrom 6,000 EUR
- Automation Retainerfrom 500 EUR per month
Stack
n8nZapierMakeZohoHubSpotClaudeOpenAIopen weight models (LlamaMistralQwenHermes)OllamaMCPQdrantpgvectorNeo4jVoyage AILangfuseFastAPINodePostgresSupabaseDockerCloudflare WorkersDigitalOcean