Post Project Audit 3 Update
The third and penultimate project audit has just been completed by the Concept AI team. There were many updates and additions all parts of the project to prepare for this audit, which was presented on Wednesday the 19th of August by Dylan and Fraser. A recording of the presentation and subsequent question time is available here, with the key points of this audit's progress summarised beneath.
Repository Reorganisation
The Concept AI Google Drive repository has been reorganised to be easier to navigate. The folder and artefact locations are more intuitive and consistent, and the directory guide has been updated to reflect the new locations. This should make it easier for both people trying to read our work, as well as for ourselves in locating our previous work.
Functional Breakdown and System Architecture
The system is comprised of eight core functions, which are listed in the table below, with their interconnections shown in the diagram following:
Function | Purpose |
User Interface | Used to accept inputs, display outputs, and collect feedback |
API Communication | Used to transfer request information between the User interface, the backend of the system, and the AI model being used. |
Prompt Processing | Used to interpret and generate the requirements from the user input |
Image Generation | Sends through the deduced requirements and returns the resulting generated images |
Output Display | Showing the resulting generated requirements and images, allowing for users to save, select, or to continue refinement. |
Iterative Refinement | Combines the prior user prompt, as well as the updated user feedback. |
Session Storage | Stores the prompts, generated images, and user feedback history. |
Error Handling | Detects invalid system inputs and failed internal requests. |
Stakeholder engagement
The team's engagement with stakeholders includes both our host as well as external Subject Matter Experts (SMEs). Our project host signed off on the newest version of our Concept of Operations, as well as our final design criteria and requirements. He also provided us with funding for API keys for the system through OpenRouter, which will cover all the testing needed for the remainder of the project. Contact with external SMEs has given the team a variety of feedback about the system's current state, and potential additions or modifications that could be made.
What next?
From here the team will conduct the testing of the system against our design criteria, collect more external and internal feedback, and construct the final MVP iteration of our system. Following this, the necessary evaluation documentation will be completed ahead of Project Audit 4, and the handover materials including our SWOT analysis will be given to our host. We will present the product and our future recommendations and analysis to Frazer Nash, and complete our pitch and poster for the ANU showcase.
Comments