Independent research · Part of the Harvard Business School (HBS) Foundry Bootcamp, Fall 2026

Every project is a game of getting things live. How does yours play?

Every automation, AI agent and agentic workflow should move cleanly from idea to operations and end with value you can prove. In practice, someone builds every record, task, review and status update by hand. In regulated industries, each square must also leave a traceable record of who approved what, when, and why. This exploratory study asks how respondents describe their delivery practices. Findings describe participants and do not establish representative industry benchmarks or validated peer rankings.

Take your turn: the research survey The study aims to share descriptive findings; publication timing is not confirmed.

PROJECT: the game of getting things live. 76 seconds, no sound.

The board, and how it changes for AI. Shaded squares are where practices vary most between organizations, and where this study is looking.

AutomationsRule-based, deterministic
  1. Idea
  2. Require­ments
  3. Assess­ment
  4. Feasibility
  5. Platform fit
  6. Develop­ment
  7. Testing
  8. Readiness
  9. Deploy­ment
  10. Hypercare
  11. Operations
AI agentsOne model, tools, goals
  1. Define role
  2. Pin model
  3. Tools & access
  4. Ground in data
  5. Evaluate
  6. Approve
  7. Deploy
  8. Monitor drift
  9. Re-version
  10. Retire
Agentic workflowsMany agents, handoffs
  1. Use case
  2. Autonomy level
  3. Risk class
  4. Orchestrate
  5. Build
  6. Scenario tests
  7. Shadow rollout
  8. Trace handoffs
  9. Re-evaluate
  10. Retire

Why this study

Requests arrive through email, chat and forms. Approvals wait in inboxes. The same work is tracked in several tools, hypercare ends informally, feedback gets lost, and few teams compare the value delivered with the value promised.

Underneath it all, delivery leads spend hours turning requirements into epics, stories and tasks, keeping backlogs current and answering the same status questions for every stakeholder. AI agents and agentic workflows add new stages and new risks on top. This study measures where that time goes and which views teams and leaders actually need.

What you get

  • Descriptive findingsPatterns reported by participants, with sampling and evidence limitations.
  • Questions to considerReflect on your own practices. No validated individual peer comparison is available.
  • Practitioner perspectivesReported practices to investigate, without claims that they are proven effective.

Your turn

The survey asks about delivery practices and includes a bonus round that can be skipped. Within the bonus round, some questions are required if you choose to enter it. Any player label or scorecard is a self-reported reflection, not a validated maturity rating or peer ranking. It's for anyone who runs, approves or delivers automation, AI or digital initiatives: project managers, project coordinators, scrum masters and product owners, CoE and transformation leaders, PMO and portfolio leaders, AI platform leads, and quality and validation leaders.

Loading the survey. If it does not appear, use the hosted survey link below.

Trouble with the form? Open the survey in a new tab.

The survey is provided by Tally, an external service. Loading this page connects to Google Fonts; initializing the survey connects to Tally, which receives browser request information and may use cookies or other services. Details entered in the survey are processed by Tally. Review Tally’s privacy policy and Google’s privacy policy before participating. Do not submit confidential business information or sensitive personal data.

How your answers are used

  • Tally processes the information entered in this survey. Its contact-details page currently requires your name, email address and company. Do not submit confidential business information or sensitive personal data.
  • Owner-approved policy permits aggregate research publication. Identifiable quotations or acknowledgments require separate permission. Participation is not permission for commercial outreach: a separate explicit opt-in is required. Removing names alone does not guarantee anonymization.
  • The retention target for identifiable research responses is six months after documented study completion, subject to justified exceptions. Privacy and deletion requests: [email protected].
  • This notice is effective October 10, 2026. Policy requirements do not establish that the survey or downstream systems enforce permission and deletion controls. This website does not enforce those account workflows.
  • Tally processes survey responses. Google Fonts supplies the page typography, and Cloudflare hosts this website; these providers receive technical requests needed to deliver their services.

Who is running it?

This is independent research conducted as part of the Harvard Business School (HBS) Foundry Bootcamp, Fall 2026. The researcher has spent more than 20 years in software engineering and enterprise AI, including a decade leading RPA and automation programs and building an AI and Automation Center of Excellence in regulated industries.

The researcher is also exploring Serviceiou, an early-stage AI value consultancy concept. Research participation does not authorize commercial outreach. A requested walkthrough concerns your research results. Any commercial outreach or identifiable publication requires separate permission, as described in the policy above; these policy requirements do not establish that provider or downstream controls enforce them.

This study is not sponsored or endorsed by Harvard Business School.

Questions: [email protected]