Understanding the Agiloop Loop: From INVENT to IMPLEMENT to INSPECT to ITERATE

Agiloop is designed around a simple idea: software should not stop evolving after the first build. The Agiloop Loop helps you move from an initial idea, to an execution-ready plan, to working software, to real-world product insights, and then back into the next round of improvement.

The loop has four stages:

1. INVENT — define and plan what should be built.
2. IMPLEMENT — build, verify, and deploy the approved work.
3. INSPECT — measure how the product and its features perform in real-world use.
4. ITERATE — turn insights into recommended improvements and feed them back into the backlog.

This article walks through the full process step by step.

————————

Before You Begin

Before moving through the full loop, it helps to understand a few key Agiloop concepts.

A Project represents the application, system, or product you are planning. Within a project, Agiloop organizes work into Feature Bundles, Features, Stories, and implementation-level details.

In general, the planning structure is:

Project → Feature Bundle → Feature → Story

A feature is a meaningful unit of functionality that delivers user or business value. A story breaks that feature into smaller pieces of work that can be implemented, tested, and reviewed.

For more background, see:

• Core Concepts in Agiloop INVENT
• Welcome to Agiloop INVENT
• Take a Tour

————————

Step 1: Start in INVENT

INVENT is where the loop begins.

Instead of starting with a blank requirements document, INVENT guides you through a structured discovery process. The goal is to turn your idea into a clear, execution-ready plan that can be reviewed, refined, exported, or sent forward into IMPLEMENT.

What INVENT Does

INVENT helps you:

• Capture the intent behind your project or enhancement.
• Identify the users, workflows, goals, rules, and constraints.
• Generate structured functional and technical specifications.
• Break the project into feature bundles, features, stories, and tasks.
• Estimate scope, cost, duration, team makeup, and story point complexity.
• Prepare work for execution by Agiloop IMPLEMENT or by external tools such as Jira, Azure DevOps, Trello, or Excel.

The result is not just a list of tasks. INVENT creates a structured software plan that explains what should be built, why it matters, and how it can be implemented.

————————

Step 2: Complete the Agiloop Interview

The Agiloop Interview is the guided discovery experience inside INVENT.

During the interview, specialized AI personas ask focused questions about your idea. The Business Analyst persona focuses on goals, users, value, rules, and business needs. The Software Architect persona focuses on architecture, integrations, data, security, and technical constraints.

Questions adapt based on your answers, helping surface details you may not have considered at the start.

What You Can Provide

For each question, you can usually provide input in several ways:

• Select from suggested answers.
• Add free-form comments.
• Provide extra context, exceptions, or constraints.
• Clarify how your business or users actually work.

Your responses directly influence the features, stories, specifications, and estimates that INVENT generates.

What the Interview Produces

When the interview is complete, INVENT generates:

• A feature bundle.
• A feature list.
• Stories and implementation-level work breakdowns.
• Initial functional specifications.
• Initial technical specifications.
• Initial estimates and story point complexity.

You remain in control. The generated output is a starting point that can be reviewed, edited, refined, archived, or pushed forward.

Related article:

• What Is the Agiloop Interview?

————————

Step 3: Review the Generated Features

After the interview, review the generated feature list.

The feature list is the primary backlog view for your project. It helps you understand the scope of the system at a glance and decide which pieces of functionality should move forward.

What to Review

For each feature, review:

• Whether the feature represents a meaningful business or user outcome.
• Whether the description accurately reflects what you intended.
• Whether the feature belongs in the current bundle or should be moved, split, combined, or archived.
• Whether any important dependencies exist between features.
• Whether the feature is ready to move toward implementation.

Agiloop features are intentionally higher-level than tasks or tickets. Each feature should describe a coherent piece of functionality rather than a small implementation chore.

Active and Archived Features

Active features are included in estimates and planning rollups. Archived features remain available for reference but are excluded from active planning totals. This allows you to explore ideas without committing every generated feature to the current scope.

Related article:

• Understanding the Features List

————————

Step 4: Review Specifications, Estimates, and Work Breakdown

Once the feature list looks right, review the details behind each feature.

INVENT provides multiple views to help you move from high-level intent to execution-level detail.

Functional Specification

The functional specification explains what the system should do from a user and business perspective. Review this section to confirm that the workflows, rules, user behaviors, and expected outcomes are correct.

Technical Specification

The technical specification explains how the system may be built. Review this section to confirm assumptions about architecture, integrations, data, security, infrastructure, and technical approach.

Estimates

INVENT generates estimates to help with planning and decision-making. Estimates may include cost, duration, team makeup, and story point complexity.

Use estimates to compare scope, understand tradeoffs, and plan budgets. They are intended as planning guidance, not fixed commitments.

Work Breakdown

The work breakdown view shows how features are decomposed into stories and execution-level detail. This is especially useful when preparing for engineering review, stakeholder walkthroughs, or delivery.

Related articles:

• Understanding Estimates in Agiloop INVENT
• Understanding the Work Breakdown View

————————

Step 5: Decide How the Work Should Move Forward

After reviewing the plan, decide where execution should happen.

You have two main paths:

1. Use Agiloop IMPLEMENT to generate, verify, and deploy the feature.
2. Push the planned work into an external execution tool such as Jira, Azure DevOps, Trello, or Excel.

The right path depends on how your team wants to build, review, and manage the work.

————————

Step 6: Configure Project Settings

Before pushing work forward, review your project settings.

Project settings control important project-level configuration, including naming, export destinations, integrations, and execution settings.

Push To Settings

If your team uses external project management tools, configure the Push To section. This allows INVENT to send features, stories, and tasks into supported tools while preserving hierarchy, descriptions, and estimates.

Supported export destinations include:

• Jira
• Azure DevOps
• Trello
• Microsoft Excel

Each destination requires different configuration details, such as credentials, target project or board identifiers, and mappings between Agiloop concepts and destination work item types.

Related articles:

• How to Configure Project "Push To" Settings
• Pushing Features to Execution Tools

————————

Step 7: Move Approved Work into IMPLEMENT

IMPLEMENT is where approved plans become working software.

IMPLEMENT uses the work defined in INVENT to build the feature as a complete unit. The goal is to move from approved specification to deployable code without forcing a developer or product owner to manually translate requirements into disconnected tasks.

What IMPLEMENT Does

IMPLEMENT can:

• Generate code from the approved feature, stories, and specifications.
• Work from the project’s defined backlog and feature structure.
• Show the story point cost before generation begins.
• Use your connected Git repository so you own the generated code.
• Support review, verification, and acceptance criteria validation.
• Prepare the generated work for deployment.

Because IMPLEMENT works from the structured output of INVENT, the generated work is grounded in the requirements, technical context, and acceptance criteria already captured during planning.

————————

Step 8: Review the IMPLEMENT Checklist

Before code generation can begin, Agiloop may ask you to review or complete required setup items.

The checklist helps ensure that the project is ready for safe and successful implementation.

Items You May Need to Review

Depending on your project configuration, you may be asked to review:

• Project naming in generated specifications.
• Feature dependencies.
• Code generation skills or technology assumptions.
• Whether a feature requires human review.

Required Items

You may also need to complete required configuration, such as:

• Git repository configuration.
• Story point balance in the project wallet.
• Railway deployment configuration, if you want Agiloop to deploy completed services to Railway.

These checks help prevent avoidable generation, build, or deployment issues.

————————

Step 9: Generate, Verify, and Review the Feature

When a feature is ready, IMPLEMENT moves it through the build process.

A typical flow may include:

1. The feature is queued.
2. Story points are held from the project wallet.
3. Code generation begins when the feature moves into active work.
4. The feature is generated as a complete unit.
5. Automated verification and review steps run against the generated work.
6. Human review may occur when required by the project or feature settings.
7. The feature moves toward completion once it passes the required checks.

The purpose of this process is not just to create code quickly. It is to create code that can be inspected, reviewed, tested, and owned by your team.

————————

Step 10: Deploy the Generated Work

When a feature reaches the appropriate status, IMPLEMENT can prepare the generated service or services for deployment.

Depending on your project setup, deployment may happen through your own CI/CD process or through an integrated deployment destination such as Railway.

If Railway deployment is configured, Agiloop can help deploy generated services and make deployment and runtime logs available from within the Agiloop experience.

The exact deployment path depends on your project settings, repository configuration, and selected deployment tools.

————————

Step 11: Move into INSPECT

INSPECT begins after working software is available for users.

The purpose of INSPECT is to help you understand how the product is actually being used. Instead of guessing whether a feature is successful, you can measure behavior, usage, flows, and goals.

Because IMPLEMENT instruments features as they are built, INSPECT can begin collecting useful telemetry without requiring you to manually wire up a separate analytics system.

What INSPECT Helps You Understand

INSPECT can help you answer questions such as:

• Are users discovering the feature?
• Are users interacting with it as expected?
• Where do users drop off?
• Which flows are working well?
• Which features are underused?
• Which goals are on track, at risk, or behind?
• Are users reaching the outcomes the product was designed to support?

INSPECT turns the deployed product into a source of feedback for the next planning cycle.

————————

Step 12: Define and Track Goals

INSPECT becomes more powerful when you define goals.

A goal describes what success should look like for a feature, flow, or product behavior. Goals may include adoption targets, usage targets, completion targets, engagement thresholds, or other measurable outcomes.

Examples include:

• Reach 50% adoption for a new feature within 60 days.
• Increase completion of an onboarding flow.
• Reduce drop-off between two key steps.
• Confirm that users are engaging with a new workflow after launch.

Once goals are defined, INSPECT can track progress over time and help identify where the product is performing well or falling short.

————————

Step 13: Use ITERATE to Identify Improvements

ITERATE closes the loop.

Where INSPECT measures what is happening, ITERATE helps decide what to do next.

ITERATE reviews underperforming, at-risk, or missed goals and provides AI-powered analysis of what may be driving the gap. It then recommends targeted improvements that can be added back into the backlog.

What ITERATE Produces

ITERATE may produce:

• A diagnosis of why a goal is underperforming.
• A recommended product change.
• A suggested enhancement to an existing feature.
• A new feature recommendation.
• A refinement that can be sent back into INVENT for planning.

This allows your team to move from “we have a problem” to “here is a concrete improvement we can evaluate and build.”

————————

Step 14: Send Recommendations Back to the Backlog

The final step is to feed ITERATE recommendations back into the planning and build cycle.

A recommendation can become a new backlog item, a refinement to an existing feature, or the start of a new feature bundle. Once it is back in INVENT, the same process begins again:

1. Review the recommendation.
2. Refine the feature or requirement.
3. Review the specifications, estimates, and work breakdown.
4. Approve the work for IMPLEMENT.
5. Build and deploy.
6. Measure with INSPECT.
7. Improve again through ITERATE.

This is the Agiloop Loop in action.

————————

The Full Process at a Glance

————————

Best Practices

To get the most value from the Agiloop Loop:

• Treat INVENT output as a strong first draft, not a locked plan.
• Review generated features before sending them into IMPLEMENT.
• Use estimates to compare options and understand tradeoffs.
• Confirm project settings before generating code.
• Keep feature dependencies accurate.
• Use human review where business risk, security, compliance, or product judgment requires it.
• Define measurable goals in INSPECT as soon as features are deployed.
• Review ITERATE recommendations regularly.
• Feed useful recommendations back into INVENT so the product continues to improve.

————————

Summary

Agiloop connects planning, building, measurement, and improvement into one continuous process.

INVENT helps you turn an idea into a structured, execution-ready plan. IMPLEMENT turns that approved plan into working software. INSPECT shows how the product performs in the real world. ITERATE uses those insights to recommend what should improve next.

Together, these stages help teams move faster while staying grounded in clear requirements, measurable outcomes, and continuous learning.