PILOT22Kevin M. Clement

Built by Kevin M. Clement

AI agents built around real delivery problems.

I designed PORTIA, STRATA and Project Nexus to automate recurring work, improve leadership visibility and make project knowledge easier to use.

The operating need

Less preparation. Better information. Clearer ownership.

Project teams spend time rebuilding artifacts, assembling updates and searching scattered records. Leaders need current information they can act on. These agents address different parts of that work.

How the three agents complement each other
AgentPrimary jobPrimary audience
PORTIAAutomate project, portfolio and governance functionsPMO, project managers and governance participants
STRATAPrepare the monthly strategic leadership reviewExecutives, technology leaders and portfolio reviewers
Project NexusAnswer questions from an individual project’s recordProject teams, sponsors and new stakeholders

PORTIA · Portfolio Reporting & Intelligence Assistant

Project, portfolio and governance automation

An agent that supports the recurring work around delivery, from project artifacts and task reminders to request routing and portfolio reporting.

The problem

Project artifacts and executive reports take repeated preparation. Updates, notifications and due tasks require follow-through. Intake requests need a consistent route to the right decision forum.

What I built

  • Automation for charters, status reports and other project artifacts.
  • Updates, notifications and due-task tracking.
  • Automatic request routing and prioritization by rock size: sand, pebble or rock.
  • Portfolio updates, demand pipeline reports and executive exports.

The value

Reduces recurring preparation and follow-up work. Applies a consistent reporting pattern and makes delivery risks, governance findings and data gaps visible for review.

Gives the PMO a repeatable way to support both individual projects and portfolio activities.

Who it serves

Project managers: artifacts, updates and due-task visibility.

PMO and portfolio leaders: demand, reporting and governance support.

Executives and Council members: consistent decision materials.

Reported evidence and controls

The PORTIA overview reports 8 artifacts across 7 active commands, with 3 builds rejected before delivery. Issues included shifted columns, dropped data and stale counts.

Registered artifact workflows verify approved toolkit files and run validation gates. Failed verification or validation stops delivery. Those artifact checks do not establish source-data truth or demonstrate that every notification or task workflow uses the same gates.

STRATA

Monthly strategic project updates

A portable leadership review assembled from live Smartsheet data, with portfolio context and detailed project narratives in one document.

The problem

Monthly reviews require someone to reconcile the project population, collect narratives and assemble a readable report. Inconsistent formats and incomplete records can obscure the decisions leaders need to make.

What I built

  • A self-contained HTML review that opens offline.
  • Portfolio measures, an agenda, project summaries and detailed project cards.
  • Executive Highlight projects grouped by Tech Lead.
  • Fresh source data each cycle, recorded narratives and findings with source provenance.

The value

Brings leadership review materials into one consistent, portable document. Preserves project narratives and makes missing or malformed data visible.

Creates a repeatable reporting cycle that supports discussion of progress, risks and decisions.

Who it serves

Executives and technology leaders: strategic portfolio visibility.

Portfolio reviewers and PMO: a consistent monthly review format.

Project owners: visibility into source gaps that need correction.

Scope and release controls

STRATA reviews all Executive Highlight projects by Tech Lead. PORTIA’s executive reporting commands can also apply a technology-domain filter, so the two report populations may differ.

A 27-check release gate covers completeness, health logic, narratives and document structure. Current proof covers 26 of 27 checks; baseline approval remains pending. Blank health is identified as missing or derived rather than automatically treated as Green.

Project Nexus

Project intelligence and knowledge

One conversational agent per project, using the systems that hold its history, plan, delivery work and specifications.

The problem

Project knowledge sits across documents, meeting records, plans, tickets and specifications. Published status can become stale. Decisions may never reach the log, and new stakeholders need repeated catch-up briefings.

What I built

  • A standard project agent connected to SharePoint, Smartsheet, Jira and Confluence.
  • Plain-language questions with source-cited answers.
  • Reconciliation that reports conflicts and deltas across records.
  • Executive status decks, meeting recaps, ownership checks and data-quality audits.

The value

Makes project knowledge easier to find and verify. Surfaces stale status, missing decisions and overdue work that may be buried in the record.

A standard configuration pattern lets each project use the same approach with its own sources, scope and team.

Who it serves

Sponsors and executives: current status and decisions needing attention.

PMs and workstream leads: overdue work, owners and decision history.

New or returning stakeholders: a project briefing from recorded information.

Findings from the pilot

The live plan held 23 overdue tasks while the published report showed 12. The pilot also surfaced decisions and actions missing from the log, a status headline conflicting with an open issue, and an overdue decision blocking workstreams.

Nexus is read-only across connected systems. It identifies contradictions for named owners to resolve. Its answers depend on records being posted and kept current; it is not the system of record.

How I build

Start with the workflow. Keep accountability clear.

Business need first

Define the user, the recurring work and the decision the output should support. Use standard inputs and outputs where consistency matters.

Controls in the workflow

Use fresh source records, provenance, reconciliation and artifact checks. Report gaps instead of concealing them. Keep review and decision ownership with people.

Measure the right value

Track preparation time, artifact acceptance, source gaps, adoption and decision usefulness. Distinguish potential value from measured results, and freed capacity from cash savings.

Know the boundaries

Artifact validation does not prove original data is correct. Project owners validate facts and recommendations; Finance, Compliance and governance partners retain their responsibilities.

Examples and pilot results are drawn from my project and agent overview materials. They are not an independently audited ROI calculation.

Make AI useful in the work people actually do.

Connect with me about practical AI, project delivery and portfolio governance.

Discuss AI delivery