Akhil
Raveendran
Nissan
GEN-AI Governance & Budget Monitoring System
Designed an enterprise solution to mitigate Operational Risk and Ensure Compliance with financial governance protocols by controlling LLM token usage.
My Role
Product Designer
Duration
Q3 FY24
Business Problem.
-
Bridging the governance gap : As Generative AI adoption accelerated uncontrolled token usage presented a major financial and compliance risk. we addressed the challenge where 65% of projects faced budget overruns, leading to complex, and delayed finance approval workflows.
Based on the session with BA and PM, an initial draft biz process was created, which was used as a base for the session with the stakeholders in the first discovery session.
Impact
100% reduction
in budget overshoots by establishing essential financial controls, effectively mitigating operational and financial risks.
Live tracking
Track token usage or budget usage across multiple variables like projects, departments, and teams.
Approach
-
Understanding the current Business process.
-
This was done by conducting multiple discovery workshops with all the stakeholders of the project.
-
-
Define the users, their roles, and responsibilities.
-
As an outcome of the workshops derived and defined the exectations needs and wants.
-
-
Define the workflows of the product.
-
Based on the current Business process and user definition, revised workflows and prototypes were created for validation with stakeholders in feedback sessions.
-
Discovery
The first draft of the user hierarchy and process flow was prepared as an outcome of sessions with BA and PM. These were used to collect feedback from the stakeholders in the following discovery session.
Based on the session with BA and PM, a user hierarchy was created.
Based on the session with BA and PM, the process flow.
Task flow created post-first session with business stakeholders.
Outcome of Discovery
As an outcome of multiple discovery workshops, created and grouped actions that can be done by the super user and where they can be performed in the process flow by each user type.
The complete set of tasks that can be performed by the super user. Multiple user types were discussed, and their access to the features was controlled with this as the master list of tasks.
With the tasks and roles defined, the workflow was created for the analytics to work for the viewer user type.
Outcome
The Figma file can be accessed.
Learning
Biz Alignment
In the early stages, set expectations and prevent last-minute pivots, emphasising stakeholder involvement.
Cross-team alignment
Helped set the technical and development scope and limitations which made the design buildable in time frame