Last Updated on 10. September 2026
With the Cosmo Agentic Underwriting Workbench (AUWWB), mgm technology partners has developed the first agentic underwriting solution in Germany. It lets underwriters and autonomous AI agents work as one team within the same underwriting process – system-agnostic, as an orchestrating layer across existing systems, and, as one of the first solutions on the market, consistently designed for the EU AI Act.
Business Value Across the Entire Underwriting Value Chain
Qualified underwriters are scarce, and submission volumes are rising. The Cosmo Agentic Underwriting Workbenchsupports not just individual steps but the entire underwriting value chain – from submission intake through risk assessment and decision-making to documentation, as well as renewals and remediation. That means: more capacity without additional underwriters, shorter quote turnaround times, and end-to-end portfolio steering across the entire underwriting value chain.Agentic AI is thus translated end to end into measurable results:
- Freeing up capacity: Straight-through processing of standard cases takes the load off underwriters – the capacity freed up goes into complex, high-margin risks.
- Shorter turnaround times: Automated intake and score-based prioritization shorten time-to-quote and improve the hit ratio.
- Active portfolio steering: Real-time visibility across the entire underwriting pipeline – capacity can be deliberately allocated between new business, renewals, and remediation.
- Better risk selection: Structured, quality-assured data prioritizes attractive risks and screens out unsuitable ones early.
A True Team Model Instead of Mere Assistants
Unlike AI assistants that merely support the underwriter, the Cosmo Agentic Underwriting Workbenchintegrates specialized AI agents as full team members. As distinct “characters” they take on defined roles – for example as a junior underwriter who prepares and pre-assesses submissions, or as a portfolio analyst who keeps an eye on the pipeline. They hand off tasks, recommend next steps, and escalate wherever human judgment counts. At every binding decision point, however, the human remains the final authority.
The team is deliberately designed to be extensible: existing agents can be built out and new agents added flexibly, matching the respective line of business and the organization’s level of maturity. In doing so, the agents take their cue from the decisions of experienced underwriters and keep getting better over time: they take on progressively more of the recurring routine work, while underwriters can focus more on complex, high-margin risks.
Further Innovations Make the Difference
Processing is steered granularly for each process step between automated (“dark”) and manual (“light”). Expertise is deployed only where it is needed. Being system-agnostic, the AUWWB can be connected as a layer to all existing systems, orchestrating and automating the processes on top of them instead of costly replacement of policy administration, pricing, or claims systems. It is also one of the first solutions on the market designed to be EU AI Act compliant from the ground up: with traceability, human oversight, and local hosting. The provisions for high-risk AI in the financial sector take effect in August 2026. “The real progress does not lie in an AI reading documents faster, but in agents and underwriters acting as one team – with the human as the final authority on every binding decision,” explainsMarc Philipp Gösswein, Member of the Executive Management at mgm technology partners.
Built on the A12 AI Low Code Platform, scoring criteria, dashboards, and workflow rules can be configured by the business itself. The solution is aimed at insurers, brokers, and MGAs in commercial and specialty lines.
Further information:
- Website AUWWB: https://www.mgm-tp.com/uwwb.html
- Website mgm technology partners: www.mgm-tp.com/insurance.html
