AI sprint
Challenge
A lot of repetition in assessment
Startup assessments at Wayra consist of research, structured assessment and checks against internal criteria, steps with a lot of repetition and pattern recognition.
Relieve the team without losing depth
Find out where AI workflows can take over parts of the assessment process without giving up the depth of the analysis.
Agent or rule?
-
suits an agent
-
LLM stage
-
-
stays rule-based
-
n8n workflow
-
Deliberately not full automation: traceability and expert control are preserved.
Solution
Process mapping
The assessment workflow mapped in Excalidraw together with the Wayra team, so that automation starts in the right place.
Technical requirements
Defined which steps suit LLM-based agents and which are better kept rule-based.
Deliberately hybrid
Not full automation, but n8n workflow steps combined with LLM stages, so that traceability and expert control are preserved.
Handover at the end of the sprint
The defined solution went to Wayra; alm's engagement ended with the sprint.
Clear about what can be automated.
In one month, we produced a hybrid architecture for AI-assisted startup assessment that was ready to implement and could go straight into development. The value lies in the clarity about which parts of the process suit agentic automation and which deliberately do not.
Stack
- Excalidraw
-
n8n
- Agentic workflow patterns
- LLM integrations
Project facts
- AI consulting
- Workflow automation
- Agentic workflows
- Process analysis
- Patrick Müller
- Jonathan Häberle