← Andre DeNardo Proof of Concept
Applied AI · Independent Venture

Clarity AI Agent

A natural-language front end for enterprise portfolio management: built so a Portfolio Lead can ask a question in plain English and get a governed answer back, instead of opening a report builder to find it themselves.

Discuss this with Andre ← Back to portfolio
The Opportunity

The answer already exists. Getting to it is the hard part.

Enterprise PPM platforms like Broadcom Clarity hold the answer to nearly every resourcing and investment question an organization has. But reaching that answer usually means learning the tool: building a report, running a query, or waiting on someone who already knows how. That gap between having a question and having an answer is where decisions lose their speed.

Today

  • Getting an answer means building or requesting a report
  • Self-service tools go unused because they still require training
  • Simple questions pass through several people before reaching a decision

With the Agent

  • Ask in plain language, get a governed answer immediately
  • No training curve: if you can describe what you need, you can get it
  • Portfolio Leads self-serve the same data governance already trusts
How It Would Feel

A conversation, not a report

Reading the portfolio and acting on it, in the same conversation — asking a question and making the change it leads to, without switching tools.

Mockup of a Clarity Agent workspace showing connected tools in a sidebar, a chat conversation about project resourcing, and a portfolio snapshot panel.
Concept illustration. A mockup of what the workspace could look like, built to show the intended experience — not a screenshot of the running proof of concept.
Why It Matters

Value an executive can act on

Faster Decisions

Questions that used to wait on a report get answered in the time it takes to ask them.

Real Self-Service

No training curve. If you can describe what you need in a sentence, you can get it.

Time Back for PMs

Less time spent building status reports, more time spent managing the work itself.

Consistent Governance

Every answer comes from the same governed data Clarity already protects, so speed never costs accuracy.

High-Level Architecture

Kept intentionally simple

The agent sits between the people asking questions and the system that already holds the answers. It never becomes a second system of record.

Where People Ask Web chat Slack / Teams / embedded The Agent Claude-powered understanding & orchestration The Adapter Translates intent into Clarity operations Clarity PPM Projects · resources financials · governance asks tool call REST / XOG answer, back in plain language
Requests flow left to right through the agent and its adapter into Clarity PPM; the answer returns the same way, translated back into plain language.

Clarity PPM stays the single system of record. The agent doesn't store or duplicate portfolio data; it interprets what someone is asking, calls the right tool, and hands back what Clarity already knows.

Where This Stands Today

Proof of concept, with a path to production

This began as a proof of concept: a working front end running a mock agentic loop, built to test whether natural language could realistically stand in for Clarity's native reporting tools. It integrates six emulated tools against representative Clarity data to demonstrate the core interaction pattern shown above.

The architecture above reflects the path from that proof of concept to a production system: Claude-powered tool-calling for understanding and orchestration, a middleware adapter bridging REST and XOG for real Clarity SaaS integration, and delivery through the channels people already use, whether that's web chat, Slack, Teams, or a widget embedded directly in existing tools.

Curious what this could look like inside your organization?

Happy to walk through the architecture, the proof of concept, or what a pilot would take.

andre.denardo@gmail.com LinkedIn ↗