Project Manager
Assistant
AI-assisted project reporting and performance monitoring
01. THE NEED
Our client manages a number of projects on behalf of their own customers, and their Project Managers were spending a large share of their time on recurring manual work: tracking how each project was progressing, keeping customers informed, and reporting on project health internally.
#AI (Artificial Intelligence)
#Local AI
#NLP (Natural Language Processing)
02. THE SOFTWARE
We built the assistant on Hermes, an agent framework using Claude Sonnet as its underlying engine. We chose this combination because it gave the client room to grow the assistant over time as their needs developed.
Security and controlled access were a priority throughout. We installed and configured Hermes on a dedicated machine with limited access within the client’s network.
We created a read-only account for the agent in their source control system (TFS), scoped to a pre-selected set of projects, and set up a dedicated email inbox so the agent could receive messages from both the client’s team and their project customers. In practice, this means the assistant has the access it needs to do its work and nothing beyond that.
03. SOLVED PROBLEMS
The assistant addressed two problems the client’s Project Managers had been dealing with for some time.
The first was the effort involved in keeping customers up to date. Producing regular progress updates had been a manual task that pulled Project Managers away from other work; the assistant now handles this with little day-to-day involvement from the team.
The second was the lack of a consistent view of project performance. Management didn’t have a steady, objective picture of how projects were actually tracking. The assistant provides that on an ongoing basis, for both the responsible Project Manager and company management.
04. KEY FEATURES
- Weekly progress reports for customers. The assistant monitors customer communication and project advancement through Pull Requests, and compiles weekly progress reports. This keeps customers informed without adding to the Project Managers’ workload.
- Ongoing project performance evaluation. It continuously reviews each project and reports to both the responsible Project Manager and company management. These reports cover progress, how closely delivery matches what was promised, responsiveness in communication, overall customer sentiment, and similar measures.
- Secure, isolated deployment. The agent runs on a dedicated, access-restricted machine within the client’s network.
- Scoped, read-only data access. Access is limited to pre-selected projects through a read-only TFS account.
- Dedicated communication channel. A dedicated inbox handles messages from both the team and project customers.
05. TECHNOLOGIES
Claude Sonnet
TFS
Have a project in mind?
Get in touch with us for your software development needs!