01
Time tracking and billing
Consolidate time entries, check billing rules, prepare draft invoices and associated emails before approval.
Custom AI agents
Developed withinBOULEUMA j.d.o.o.↗Automate complex workflows without losing control.
I design artificial intelligence agents for law firms, compliance teams, finance departments and regulated organizations. Each solution is built around your rules, data and existing tools.
Human-in-the-loopTraceabilityExplicit permissions
Agentic system
From signal to action, with human control.
controlled
Email
Documents
CRM / tools
AI
AI Agent
Context + rules
✓
Human approvalHuman-in-the-loop
↗
Controlled actionAPI · task · draft · update
Approach
Not another SaaS. A tool built around real work.
The starting point is not the technology, but a repetitive, expensive or difficult-to-control task.
I analyze the workflow, the data involved, the required approvals and the relevant risks. The agent is then designed to support the team, automate what can be automated and leave sensitive decisions to qualified professionals.
The result may be a focused automation, an internal interface or an agent connected to several tools. Replacing your information system is not a prerequisite.
Use cases
Agents designed for specific workflows
A useful agent does not start with a model. It starts with a trigger, defined sources, an allowed action and an approval point.
02
Matter and client intake
Collect required information, identify missing documents, organize files and route the matter to the appropriate professional.
03
Regulatory monitoring
Monitor defined sources, detect relevant changes, prepare summaries and create follow-up actions.
04
Document review
Extract information, flag inconsistencies, apply a control framework and prepare a structured human review.
05
Financial reporting
Consolidate data, highlight variances, prepare commentary and generate a verifiable first reporting layer.
06
Internal knowledge support
Answer from approved procedures and documents, cite internal sources and escalate sensitive situations to a responsible person.
Architecture
AI agent or conventional automation?
I do not add AI where a deterministic workflow is simpler. The right architecture depends on how much context needs to be interpreted.
Recommended architecture
Deterministic automation
Known triggers, explicit rules and predictable steps: automate the workflow without adding an AI model unnecessarily.
Sectors
Designed first for regulated environments
01
Law firms
Matter management, time tracking, document preparation, research and internal organization.
02
Compliance and risk
Monitoring, controls, internal procedures, decision records and follow-up actions.
03
Finance and controlling
Data collection, closing, reporting, variance analysis and administrative automation.
04
HR and sensitive organizations
Documented workflows, access rights, records, handovers and operational coordination.
Architecture
What an agent can integrate
The solution is adapted to your technical environment and required confidentiality level.
✓Approved documents, knowledge bases and templates
✓Email, calendar, document storage, CRM or specialist tools
✓Permission rules, approval steps and activity logs
✓One or more AI models depending on task, cost and risk
✓Monitoring dashboard, alerts and human escalation
What I do not offer
An opaque automation that makes professional decisions on your behalf.
In regulated fields, the objective is to increase the team’s capacity while preserving a clear chain of accountability.
AI agents are assistance tools. They do not replace professional judgment, validation or applicable obligations.
Method
From workflow to operational pilot
01Identify the workflow
02Map data and risks
03Build a focused pilot
04Test with users
05Deploy and improve
Start with a specific workflow, not an abstract promise.
Describe a repetitive task, a bottleneck or a control that is difficult to maintain. I will help determine what can realistically be automated and under which conditions.