DEV Community: Julie Yakunich

Product Thinking in a Project World: Delivering Software That Actually Moves the Needle

Julie Yakunich — Wed, 14 Jan 2026 15:12:46 +0000

At Leading EDJE, we’re obsessed with one thing: measurable value. Not shipped features. Not “percent complete.” Value - business outcomes your leaders can see, feel, and bank on.

Many organizations still run technology work as projects with fixed scope, budget, and timelines. Constraints are real. The difference with us is how we use those constraints: we prioritize by value, measure impact continuously, and make trade-offs transparent in business terms.

Our Approach: Value First, Always

1) Start with outcomes, not features

We clarify the business problem, define what success looks like, and agree on a small set of measurable outcomes (e.g., reduced cycle time, increased conversion, lower cost-to-serve). Features are a means; outcomes are the end.

2) Align goals from strategy to sprint

We align strategic goals, product goals, and sprint goals so day-to-day work directly supports what leadership values most. Teams understand why every item is in progress.

3) Order the work by value

Backlog decisions are grounded in expected value, ensuring evidence—not opinion—guides the roadmap.

4) Measure what matters

We use a handful of practical metrics tied to the target outcomes: adoption/usage, throughput/lead time, error rates, NPS - whatever best signals business impact. Then we close the loop by sharing results back to teams and stakeholders.

5) Make plans honest—and useful

Using agile forecasting and flow metrics, we give leaders credible timelines and options:

“Within the current budget/date, here’s the most value we can deliver.”
“To capture more value, here are the trade-offs.”

The Value Delivery Toolkit (How We Make This Real)

To make “value first” practical in project-driven environments, we bring a set of lightweight, repeatable practices we call the Value Delivery Toolkit:

Evidence-Based Management (EBM): Shared language and measures for value (e.g., Current Value, Time-to-Market, Ability to Innovate) so progress is judged by outcomes, not output volume.
Agile Forecasting: Probabilistic forecasts (including Monte Carlo and flow metrics) for realistic delivery windows that still leave room to optimize for the most valuable work.
Backlog & Goal Alignment: Clear product/sprint goals, value-based ordering, and explicit trade-offs that connect strategy to execution.
Incremental, Evidence-Driven Discovery: Small experiments to validate assumptions early, before big spend.

We don’t just “fill a seat.” We translate goals into outcomes, outcomes into measures, and measures into everyday decisions.

What This Looks Like in Your Organization

If scope and date are fixed: We maximize value within the constraints and show the cost/benefit of alternatives in business terms.
If success criteria are fuzzy: We co-define measurable outcomes and connect them to strategy.
If product roles are unclear: We act as translators—helping PMs, POs, BAs, and engineering align around outcomes.
If change feels hard: We start small (add sprint goals, instrument 1–2 outcome metrics, value-order the top of the backlog) and build momentum with proof.

Quick Wins You Can Apply This Month

Add sprint goals that are outcome-oriented and measurable and review them daily.
Create a product goal and tie it to a strategic objective.
Find one metric that best signals the outcome you want, and report it at sprint review.
Switch to probabilistic forecasts (ranges with confidence) instead of single-date promises (we love using Actionable Agile for this).

Why Clients Choose Leading EDJE

Value-Obsessed: positive business impact = success.
Pragmatic: We blend methods to fit your constraints and culture.
Transparent: Forecasts and metrics make trade-offs clear before money is spent.
Partner Mindset: We ask the right questions, care deeply about outcomes, and stay accountable.

Bottom line: We help you move from “was it delivered?” to “what value did it create?”

Ready to Turn Projects into Outcomes?

If you want your next initiative to prove its value—not just deliver scope—we’d love to help. Our Value Delivery Toolkit meets you where you are and raises the bar on what your technology delivers.

Navigating the Unexpected: How to Get Your Project Back on Track After a Setback

Julie Yakunich — Tue, 26 Aug 2025 18:27:53 +0000

We've all been there: your project is humming along nicely when suddenly, an unexpected interruption brings everything to a halt. Recently, our team faced a two-week break in the middle of a client project. When we reconvened, we encountered several challenges but also discovered valuable strategies for regaining momentum. Here's what we learned about getting back on track after an unexpected blip.

The Challenge of Resuming Work

Returning to a paused project is rarely as simple as picking up where you left off. Our team immediately faced several obstacles:

Access roadblocks: Regaining entry to necessary systems required navigating multiple layers of security and approval processes.
Timeline concerns: Stakeholders had legitimate questions about how the lost time would impact deliverables and deadlines.
Momentum loss: The team's rhythm and flow had been disrupted, requiring intentional effort to rebuild.

7 Effective Steps to Regain Project Momentum

Based on our experience, here are proven steps to help your team bounce back from an unexpected project interruption:

1. Rally Strong Leadership

Our project manager, scrum master, and product owner immediately aligned to advocate for the team's needs. This leadership triad created a protective buffer that allowed team members to focus on getting back to productivity while they handled administrative hurdles.

Part of this leadership alignment included rebaselining our project plan—a meticulous process of adjusting timelines, renegotiating commitments, and communicating changes transparently. While the team initially felt anxious about how the new baseline might affect delivery, seeing a clear, updated path gave them reassurance that we could move forward with confidence.

Action tip: Identify key leadership roles and ensure they're communicating frequently during the recovery period.

2. Cultivate Patience Deliberately

Frustration is natural when facing unexpected barriers. We made it a point to remind each other regularly that the process would take time and that patience would serve us better than impatience.

Action tip: Acknowledge frustrations openly but pair them with reminders about the temporary nature of the challenges.

3. Leverage Available Tools

We were fortunate to have access to an internal, secure AI assistant that helped us review code and write tests. This technological support accelerated our ability to get back up to speed.

Action tip: Audit what tools and resources might help your team recover more quickly, even if they weren't part of your original workflow.

4. Intensify Team Connection

Our scrum master made a conscious effort to check in with team members individually and frequently. We also increased team-building activities to rebuild the connection that had been temporarily lost.

Action tip: Schedule additional informal check-ins and create opportunities for the team to reconnect socially as well as professionally.

5. Create Psychological Safety

We established a safe space where team members could voice concerns without fear. This open dialogue led to creative solutions we might not have discovered otherwise. Even before the furlough, we had cultivated an environment of safety and trust where people could voice their concerns and opinions. This went a long way when we had an unexpected outage.

Action tip: Host a dedicated session specifically for airing concerns and brainstorming recovery strategies.

6. Reconnect with Purpose

Reminding ourselves why we valued this client and project rekindled our motivation. This connection to the work proved powerful in overcoming obstacles.

Action tip: Take time to explicitly discuss what team members find meaningful about the project to reignite intrinsic motivation.

7. Embrace Autonomy with Accountability

Having the freedom to solve problems creatively, backed by supportive stakeholders and mutual trust, allowed us to find the best path forward rather than the most obvious one.

Action tip: Give team members space to determine their own best recovery strategies while maintaining clear accountability for outcomes.

The Foundation for Resilience

Our experience highlighted that teams bounce back most effectively when they have:

Autonomy to solve problems creatively
Strong, supportive stakeholders who trust the team
Psychological safety to voice concerns and ideas
A foundation of trust among all parties

The unexpected pause in our project could have derailed our momentum permanently. Instead, by implementing these strategies, we not only recovered but ultimately delivered successfully.

When your team faces an unexpected interruption—whether it's two weeks or two months—remember that the path back to productivity is paved with intentional leadership, strengthened connections, and a renewed sense of purpose. The resilience you build through this process will serve your team well beyond the current project.

Have you experienced an unexpected project interruption? What strategies helped your team recover? I'd love to hear your stories in the comments below.

Evidence-Based Management: How Agile Forecasting and EBM Lead to Better Outcomes

Julie Yakunich — Thu, 27 Feb 2025 16:16:46 +0000

Evidence-Based Management (EBM) is a framework that helps organizations make better decisions by using empirical data to evaluate and improve performance. Created by the creators of Scrum, it pairs perfectly with Scrum and focuses on measuring value delivery through Key Value Areas such as Current Value, Time-to-Market, and Unrealized Value. By ensuring decisions are grounded in evidence rather than assumptions, EBM supports continuous inspection and adaptation, enabling teams to deliver measurable outcomes aligned with business goals.

EBM ties into Agile Forecasting by providing a framework for using empirical data to guide decision-making, measure progress, and forecast outcomes. EBM focuses on outcomes and the evidence of value delivered, making it a natural complement to Agile Forecasting, which uses data and probabilistic methods to predict future performance.

How Agile Forecasting Aligns with EBM

1. Focusing on Value Delivery

EBM emphasizes maximizing value delivery rather than focusing solely on outputs. Agile Forecasting aligns with this by helping teams predict when and how much value they can deliver based on historical data and flow metrics. For instance:

Throughput Forecasting: Estimates how many high-value features will be delivered by a target date.
Monte Carlo Simulations: Predicts the likelihood of delivering a prioritized set of features within specific timeframes, ensuring work aligns with business goals.

2. Data-Driven Decision-Making

EBM relies on empirical evidence (e.g., Key Value Areas like Current Value, Time-to-Market, and Unrealized Value) to make informed decisions. Similarly, Agile Forecasting uses historical data (e.g., throughput, cycle time) to make predictions. This shared reliance on data ensures:

Prioritization of High-Value Work: Teams focus on maximizing Current Value by delivering features faster to meet customer needs.
Estimating Time-to-Market: Forecasting helps stakeholders understand delivery timelines for specific features or product increments.

3. Continuous Improvement Through Metrics

Both EBM and Agile Forecasting depend on continuous measurement and feedback loops to improve processes and outcomes. For example:

Agile Forecasting: Uses metrics like cycle time, WIP, and throughput to refine predictions over time.
EBM: Incorporates these metrics into broader Key Value Areas to evaluate how process improvements impact business outcomes, such as reducing delays (Time-to-Market) or increasing stakeholder satisfaction (Current Value).

4. Risk Management

In EBM, understanding risks is crucial for managing Unrealized Value (the gap between potential and actual value delivered). Agile Forecasting supports risk management by:

Providing Probabilistic Forecasts: Shows the likelihood of meeting specific delivery targets, enabling teams to plan for variability.
Highlighting Bottlenecks: Flow metrics help identify areas where delays could jeopardize value delivery, allowing teams to mitigate risks proactively.

5. Transparency and Stakeholder Confidence

EBM encourages transparency by using metrics to communicate progress and value delivery. Agile Forecasting supports this by:

Offering Clear, Data-Driven Forecasts: Stakeholders can trust data-backed predictions.
Demonstrating Probabilities and Scenarios: For example, "There’s an 85% chance of delivering 15 features by the end of the quarter," which helps manage expectations effectively.

Benefits of Combining EBM and Agile Forecasting

By combining EBM’s focus on delivering measurable value with Agile Forecasting’s probabilistic methods, organizations can:

Improve Predictability: Maintain flexibility while providing reliable forecasts.
Align Team Efforts with Strategic Goals: Ensure that work prioritization supports broader business objectives.
Leverage Real-World Data: Use empirical evidence to guide decision-making and deliver incremental value.

Together, these practices create a virtuous cycle of data-driven improvement, ensuring teams deliver the right value at the right time.

Let me know if you'd like this tailored further for your context!

# Unlocking the Benefits of Agile Forecasting

Julie Yakunich — Tue, 18 Feb 2025 15:05:42 +0000

In our previous article, we explored the foundations of Agile Forecasting and how tools like Flow Metrics and Monte Carlo Simulations empower teams to predict outcomes with greater accuracy and confidence. While that overview highlighted the benefits of replacing deterministic forecasting with probabilistic models, there’s much more to unpack when it comes to the practical applications and impact of Agile Forecasting on software development projects.

In this article, we’ll take a closer look at why Agile Forecasting is a game-changer. We’ll examine how it helps teams set achievable goals, manage risks more effectively, and build trust with stakeholders by delivering transparent, data-driven insights. Whether you’re aiming to improve sprint planning, mitigate risks, or foster confidence in delivery timelines, Agile Forecasting offers a robust framework to navigate the complexities of modern software development. Let’s dive deeper into the specifics.

1. Set Achievable Goals

Agile Forecasting uses historical data and flow metrics (like throughput, cycle time, and WIP) to make realistic predictions about what a team can accomplish within a specific timeframe.

Steps:

Analyze Historical Data: Review past throughput (number of tasks completed per sprint or time period) to establish a baseline.
Incorporate Monte Carlo Simulations: Use simulations to estimate the likelihood of completing a certain number of items within a sprint or iteration. For example, predict that there is an 85% chance of completing 10 user stories.
Focus on Prioritization: Break down work into small, manageable tasks and focus on high-priority items, ensuring the goals align with capacity and priorities.
Monitor Progress: Use tools like Cumulative Flow Diagrams (CFDs) to visualize work in progress and adjust goals as necessary.

Outcome:

Teams avoid overcommitment, set realistic expectations, and align their goals with their capacity and historical performance.

2. Manage Risks Effectively

Agile Forecasting helps identify and address risks by introducing probabilistic thinking and visualizing potential bottlenecks or delays.

Steps:

Use Probabilistic Forecasting: Replace deterministic dates with ranges, such as, "There's an 85% chance we'll complete this by 20 days."
Visualize Bottlenecks: Use CFDs and flow metrics to identify areas where work is stalled or where WIP limits are exceeded.
Plan for Variability: Monte Carlo Simulations can help quantify the impact of variability, such as delays or fluctuating throughput, allowing teams to create contingency plans.
Limit Work in Progress (WIP): Apply WIP limits to prevent overloading the team, which reduces cycle times and mitigates the risk of work piling up.

Outcome:

Teams proactively address uncertainties, making them more resilient to unexpected changes while delivering value consistently.

3. Improve Stakeholder Confidence in Delivery Timelines

By providing data-backed forecasts and emphasizing transparency, Agile Forecasting builds trust with stakeholders.

Steps:

Present Data-Driven Forecasts: Use Monte Carlo Simulations to communicate delivery probabilities (e.g., "We have a 90% chance of completing this feature set by the end of the quarter").
Update Regularly: Continuously refine forecasts based on the latest data, showing stakeholders progress in real time.
Explain Variability: Help stakeholders understand the inherent uncertainty in software development and how probabilistic forecasts mitigate risks compared to fixed timelines.
Use Visual Tools: Share easy-to-understand visualizations like CFDs or cycle time scatterplots to demonstrate progress and areas of improvement.

Outcome:

Stakeholders gain confidence in delivery timelines as they see transparent, realistic predictions backed by data, along with plans to address risks proactively.

Agile Forecasting: Leveraging Monte Carlo Simulations and Flow Metrics for Predictability

Julie Yakunich — Tue, 11 Feb 2025 20:46:24 +0000

The Ongoing Battle of Agile vs Waterfall

Agile teams often face challenges in predicting project timelines and delivery dates due to the framework's inherent flexibility, which prioritizes adaptability over rigid schedules. While this approach aligns with Agile principles, most IT departments operate within yearly budgets that require some level of forecasting to understand when funded initiatives are to be completed. However, scoping out an entire year’s worth of work is inefficient and counterproductive to Agile’s iterative nature.

Traditional approaches, such as deterministic forecasting commonly used in waterfall methodologies, attempt to map out the entire scope, timeline, and resource allocation for a project upfront. This assumes that requirements, priorities, and team performance will remain static throughout the project lifecycle—a notion that rarely aligns with the realities of modern software development. When unexpected changes arise (e.g., shifting priorities, evolving customer needs, or unforeseen technical challenges), deterministic plans become obsolete, leading to missed deadlines, mismanagement, and stakeholder frustration.

What is a team to do when stuck in this conundrum? How can teams forecast work enough to satisfy the planning and budgetary needs of most IT departments while staying nimble enough to produce high-quality, useful software features? This is where Agile Forecasting using Flow Metrics and Monte Carlo Simulation comes in handy.

Bridging the Gap with Agile Forecasting

Predictive approaches like Agile Forecasting are better suited for today’s dynamic software development environment. By leveraging historical data and probabilistic models, Agile Forecasting embraces variability and uncertainty, allowing teams to provide a realistic range of outcomes rather than fixed, often unrealistic predictions. Techniques like Monte Carlo Simulations and flow metrics help teams forecast delivery timelines and throughput with greater accuracy, enabling more informed decision-making and proactive risk management.

Agile Forecasting bridges the gap between the adaptability of Agile and the need for predictability in IT planning. It empowers teams to achieve realistic goals without compromising flexibility, delivering value iteratively while meeting business expectations.

In this post, we’ll delve into how Agile Forecasting, powered by tools like Monte Carlo Simulations and flow metrics, can revolutionize the way teams plan, execute, and deliver their work.

The Case for Agile Forecasting

Agile forecasting predicts how much work a team can complete within a given timeframe by leveraging historical data and current performance metrics. Unlike deterministic approaches—such as the upfront, rigid project planning typical of waterfall methodology—it embraces variability and applies probabilistic thinking to deliver a more accurate range of potential outcomes. By adopting Agile Forecasting, teams can:

Set achievable goals: Use historical data and flow metrics to make realistic predictions about what the team can complete.
Manage risks effectively: Leverage probabilistic forecasting to account for variability and visualize bottlenecks with Cumulative Flow Diagrams (CFD) and flow metrics. Use Monte Carlo Simulations to create contingency plans and apply Work in Progress (WIP) limits to reduce cycle times.
Improve stakeholder confidence in delivery timelines: Provide data-driven forecasts using Monte Carlo Simulations and refine them continuously with updated metrics.

We’ll explore these benefits in more depth in a future article, but for now, let’s focus on Flow Metrics and Monte Carlo Simulations.

Flow Metrics: The Foundation of Agile Forecasting

Flow metrics are essential for understanding how efficiently work moves through a development process. The key metrics include:

Cycle Time: The time it takes for a work item to move from start to finish.
Work in Progress (WIP): The number of active items being worked on.
Throughput: The number of items completed in a given time period.
Work Item Age: The elapsed time since work on an item started.

Applying Flow Metrics in Practice:

Sprint Planning: Use throughput to determine how much work to pull into a sprint, moving away from deterministic velocity metrics.
Retrospectives: Analyze cycle time scatterplots to identify patterns or outliers and improve processes.
WIP Limits: Leverage Little’s Law (a principle linking WIP, throughput, and cycle time) to understand how limiting WIP improves cycle time and overall flow efficiency.

Monte Carlo Simulations: A Powerful Tool for Forecasting

Monte Carlo Simulations (MCS) is a computational algorithm that uses repeated random sampling to generate probabilities for a range of outcomes. In Agile Forecasting, MCS helps answer critical questions:

How many items can we close by a target date?
When will a specific number of items be completed?

How It Works:

Gather Historical Data: Collect throughput or cycle time data (e.g., completed tasks per sprint). Tools like JIRA simplify this process.
Run Simulations: Use this data to simulate thousands of possible outcomes, considering uncertainties and variations.
Generate Probabilities: Calculate the likelihood of meeting specific delivery targets.

For example, an MCS might predict there’s an 85% chance of completing 20 items in the next sprint. This approach is especially useful for fixed-scope or fixed-date projects, providing actionable insights into what’s achievable.

Tools like Actionable Agile or even Excel can be used to perform these simulations. If you prefer templates or tutorials, resources like @TheAgileLeanGardener on YouTube offer great guidance.

Continuous Improvement with Agile Forecasting

Agile Forecasting isn’t a one-time activity; it’s a continuous process of inspecting and adapting. By regularly analyzing flow metrics and leveraging probabilistic tools like Monte Carlo Simulations, teams can refine workflows, deliver value incrementally, and make data-driven decisions.

Tools and Resources to Get Started

Books:
- Actionable Agile Metrics for Predictability by Daniel Vacanti
- When Will It Be Done? by Dan Vacanti
Tools:
- Actionable Agile, LinearB, Sleuth
Articles:
- Monte Carlo Forecasting in Scrum
- Step-by-Step Guide to Monte Carlo Simulations

By adopting these techniques and tools, your team can shift from guesswork to reliable, actionable forecasts—delivering work with confidence and predictability.

Applying Evidence-Based Management (EBM) in Your Organization

Julie Yakunich — Tue, 16 Jul 2024 14:47:40 +0000

At Leading EDJE, we're on a relentless quest to enhance and measure the impact of our work. Harnessing the transformative power of Evidence-Based Management (EBM), a framework designed to complement your Agile practice, can revolutionize your approach to data-driven decision-making, empowering your teams to deliver value and achieve their goals.

The Foundation of EBM

EBM is deeply rooted in measurement — it's about translating data into actionable insights that align with your company's strategic goals. It's not just about doing work; it's about ensuring the work we do matters to the end customer. This approach can be particularly powerful in agile environments, like Scrum, but its principles are universally applicable.

Building an EBM Dashboard

The EBM Dashboard is a central tool in applying EBM. It visualizes your team's progress against key goals and measures. Here’s how you can construct an effective dashboard:

Start with Goals: Define Strategic, Intermediate, and Immediate Tactical Goals.

Strategic Goals: Long-term, big-picture objectives.
Intermediate Goals: Support the strategic goals, often aligning with or same as product goals.
Immediate Tactical Goals: The daily work driving towards your sprint goals.

Key Value Areas (KVAs): These are the lenses through which you view value:

Current Value: The value you're delivering now.
Unrealized Value: Potential value that could be delivered.
Ability to Innovate: How effectively you can introduce new capabilities.
Time to Market: Ho w quickly you can deliver new capabilities.

Key Value Measures (KVMs): These are the metrics you track under each KVA. They should be informative, actionable, and relevant to your goals.

Example Dashboard

*from Unlocking Business Agility with Evidence Based Management

Using the Dashboard

Once you've built your dashboard, it's pivotal to use it effectively:

Review Regularly: Update and review the dashboard in line with your sprint cadence.
Data-Driven Decisions: Use the dashboard to make informed decisions and adjust your approach when necessary.
Team Ownership: Ideally, the Agile team should collectively own and contribute to the dashboard.

The Role of Experimentation

Experimentation is integral to EBM and Scrum. Each sprint, release, or iteration, depending on your agile methodology, can be seen as an experiment to test hypotheses. By aligning your sprint goals with your hypothesis, you ensure each sprint is a step towards validating your approach and delivering value.

Applying EBM Across Various Scenarios

Whether you're on a Scrum team or not, EBM principles can be adapted to fit your work context. Here are some scenarios of how EBM could help:

No Clear Goals: If your team doesn’t have goals, begin by asking questions to identify goals at the company or project level.
Solo Contributors: Measure individual contributions and look for ways to connect them to larger goals.
Non-product Development Teams: Even if you’re not building a product, use EBM to focus on organizational capabilities (like Time to Market) and how your work indirectly add s value.

Engage with EBM

Implementing EBM is not about rigidly following a set formula; it's about thinking critically about what to measure and why. It involves continuous learning, adaptation, a willingness to ask the hard questions and have the hard conversations.

As we move forward, it's clear that EBM will become an increasingly important part of how effective teams work. By providing feedback and sharing our experiences, we can shape how EBM is implemented across our organizations.

EBM is a journey, not a destination. By adopting its principles and actively engaging with the process, we can all contribute to a culture of continuous improvement and evidence-based decision-making.

Harnessing the Power of Generative AI for Practical Business Solutions

Julie Yakunich — Tue, 11 Jun 2024 13:57:06 +0000

The rapid advancements in Generative AI have opened up a plethora of opportunities for businesses to innovate and solve complex problems. In a recent internal discussion, we explored various ways to leverage these technologies to build practical AI-driven solutions. Here's a rundown of the key takeaways and how they can be applied to your business.

Rethinking Data Retrieval with Retrieval-Augmented Generation (RAG)

One of the most intriguing applications of Generative AI is the concept of Retrieval-Augmented Generation. RAG combines the creative power of language models with the precision of information retrieval systems to generate responses that are both accurate and contextually rich. By integrating a RAG system, businesses can enhance customer service bots, improve search functionalities, and create more dynamic user interactions.

For instance, using a framework like Llama Index, companies can quickly bootstrap a RAG project that taps into their own data repositories. This means customer inquiries can be addressed by pulling relevant information from internal documents, providing responses that are both informed and tailored to the user's needs.

Structuring Unstructured Data with AI

Another powerful aspect of Generative AI is its ability to parse unstructured text and convert it into structured data. Imagine taking a block of text, such as a meeting transcript or a product description, and extracting key information in a structured format like JSON. This capability is invaluable for businesses looking to automate data entry, streamline content management, or enhance data analysis.

Utilizing libraries like Pydantic in Python, developers can create models that instruct AI on how to extract and structure data. This process can transform verbose product descriptions into concise, database-ready entries, saving countless hours of manual labor.

Localizing AI Models for Development Efficiency

The cost of running AI models on cloud platforms can quickly add up during the development phase. An effective strategy to mitigate this is to run local versions of AI models. Projects like EleutherAI's GPT-Neo and GPT-J provide open-source alternatives that can be used as stand-ins during development. Once the application is ready for deployment, it can then be switched to more powerful models such as GPT-4 for final testing and production use.

Beyond Pretty Pictures: Practical Uses for Generative Image Models

Generative image models like Stable Diffusion are not just for creating visually appealing images—they have practical business applications, too. For example, they can be used to visualize clothing on different body types without the need for a photoshoot. By identifying clothing items and human poses, these models can generate realistic images of how apparel would look on various individuals, offering a personalized shopping experience for customers.

Integrating AI with Existing Business Tools

AI doesn't work in isolation. It can be integrated with existing business tools to enhance their capabilities. For instance, wireframes or UI mockups can be analyzed by AI to generate code or to extract key performance indicators (KPIs). This integration can significantly speed up the development process and provide insights that might otherwise be missed.

Leveraging AI for Data-Driven Predictions

While Generative AI may not be the best tool for crunching numbers or making predictions based on statistical data, it can be a part of a larger analytical framework. By identifying patterns in text data, such as sports commentary or financial reports, AI can aid in the prediction process. However, for more precise numerical analysis, traditional machine learning techniques and tools like pandas in Python may be more appropriate.

In conclusion, Generative AI is reshaping how we approach problem-solving in the business world. From enhancing customer interactions to streamlining development processes, the potential applications are vast and varied. By staying informed and experimenting with these technologies, businesses can find innovative ways to leverage AI for practical and impactful solutions.

Insights from the Leading EDJE Round Table Discussion on AI: Bridging the Gap Between Virtual and Physical Spaces

Julie Yakunich — Tue, 16 Apr 2024 17:07:48 +0000

Artificial Intelligence (AI) is a driving force of technological innovation, shaping the future across diverse sectors. In a dynamic round table discussion, 30 professionals from Leading Edje congregated to dissect and debate the multifaceted nature of AI. They delved into current trends, best practices, and the ethical landscape that frames AI development and deployment. This gathering of keen minds was not only enlightening but also highlighted the importance of a nuanced and informed approach to creating and utilizing AI systems. The collaborative atmosphere fostered a rich exchange of ideas, emphasizing the collective effort required to navigate the complexities of AI responsibly.

Here are some key takeaways from the discussion:

Prompt Injection and AI Security:

The emerging threat of prompt injection in generative AI systems was a critical concern, with discussions on the need to design safeguards against such vulnerabilities.
Ensuring the security of AI systems against malicious inputs is vital for maintaining their integrity and trustworthiness.

Machine Learning Transparency:

The discourse emphasized the imperative for transparency in machine learning models, particularly about their limitations and inherent biases.
A transparent approach is crucial for fostering trust and enabling users to make more informed decisions when interacting with AI.

AI Deployment Considerations:

Technical considerations such as data privacy and the careful planning of AI infrastructure were recognized as fundamental to the successful deployment of AI technologies.
Protecting sensitive data and establishing robust infrastructure are essential steps in realizing AI's potential while mitigating risks.

Sanitization in AI Development:

The use of pipelines and chains in the development of AI models was discussed as a method for sanitizing both inputs and outputs, ensuring the cleanliness and safety of data.
This practice is crucial for preventing the propagation of errors or biases that could compromise the effectiveness of AI applications.

Human Expertise in AI:

The round table highlighted the irreplaceable value of human expertise in AI, particularly for complex tasks such as document analysis that require nuanced understanding.
Integrating human judgment with AI systems ensures higher quality outcomes and mitigates the limitations of automated processes.

Quality of Training Data:

Ensuring the high quality of data used for training and validating machine learning models is essential for their accuracy and fairness.
The conversation focused on the need for clean, unbiased, and representative datasets as the foundation for reliable AI models.

AI Explainability and Auditability:

Explainability in AI, or the ability to understand and trace the decision-making process of models, was recognized as a cornerstone of responsible AI development.
Having auditable explanations for AI decisions is crucial for accountability and building user confidence in AI systems.

Human Oversight in Healthcare AI:

Real-world examples of human oversight in critical AI applications, such as healthcare diagnosis, were discussed to illustrate the importance of human intervention in sensitive areas.
Such oversight is fundamental to ensuring the accuracy and reliability of AI-assisted decisions that can have profound implications on human lives.

Copyright and Data Source Considerations:

Issues surrounding copyright and the importance of transparency about data sources in machine learning were brought to the table, highlighting the ethical dimensions of data usage.
Ensuring that data is ethically sourced and used in compliance with copyright laws is vital for the legitimacy and ethical standing of AI initiatives.

AI Safety and Legal Risks:

Finally, safety considerations and the potential legal risks associated with AI models, particularly in content generation, were discussed.
Recognizing and addressing the legal implications and safety concerns is necessary to navigate the complex landscape of AI responsibly.

The round table discussion served as a microcosm of the larger conversation happening around the world about AI. It highlighted the importance of bringing diverse voices to the table to navigate the complexities of this transformative technology. As AI continues to break down the barriers between the digital and physical realms, events like this provide valuable forums for sharing knowledge, addressing concerns, and fostering collaboration.

In conclusion, the intersection of AI with various aspects of life and work presents both challenges and opportunities. As we forge ahead, the collective wisdom from discussions like these will be instrumental in shaping a future where AI is not only powerful but also responsible and inclusive.

Embracing Agility and Evidence: How EBM Enhances Scrum Practices

Julie Yakunich — Mon, 26 Feb 2024 14:58:09 +0000

In the fast-paced realm of software development, Scrum has revolutionized project management, offering a flexible and collaborative approach to tackling complex projects. It's known for its ability to improve product quality, enhance team collaboration, and speed up the delivery process. Yet, Scrum also encounters hurdles like the necessity for cultural shift, reliance on team dynamics, and the risk of too many meetings.

Evidence-Based Management (EBM) steps in as a strategic ally to Scrum, designed to bolster its strengths and address its challenges through a focus on empirical data and continuous improvement. Here's a closer look at how EBM enriches Scrum practices:

Enhanced Decision-Making
• EBM empowers teams with data-driven decision-making, enhancing Scrum's adaptability and flexibility. This approach ensures strategies are grounded in proven effectiveness, facilitating better responses to changing project requirements.

Improved Focus on Value Delivery
• While Scrum aims for quick value delivery, EBM emphasizes measuring the value delivered to customers. This helps teams prioritize high-impact work, streamlining the process and directly aligning with Scrum's efficiency goals.

Achieving Strategic Alignment
• EBM aligns team efforts with strategic objectives, focusing on key areas like Current Value, Unrealized Value, Time to Market, and Ability to Innovate. This ensures Scrum teams maintain a clear vision of organizational goals, enhancing long-term success.

Enhancing Risk Management
• Addressing the risks inherent in dynamic software projects, EBM's focus on continuous outcome measurement helps identify and mitigate issues early. This proactive stance complements Scrum's agile nature, ensuring high-quality product delivery.

Fostering Continuous Improvement and Organizational Health
• EBM extends Scrum's process improvement ethos to encompass overall organizational well-being. By promoting a culture of learning and experimentation, EBM helps overcome resistance to change, boosting innovation and market responsiveness.

Streamlining Delivery with Flow Metrics
• Utilizing flow metrics like Work In Progress (WIP), Cycle Time, and Throughput, EBM identifies and addresses bottlenecks. This optimization enhances workflow efficiency, reduces delivery times, and aligns development processes with the goal of delivering valuable, high-quality products to customers.

While Scrum lays a solid foundation for agile development, integrating EBM principles offers a comprehensive strategy to not only navigate Scrum's challenges but also amplify its advantages. Through the lens of empirical data, a focus on value, and a commitment to continuous improvement, EBM ensures that Scrum teams can more effectively achieve their objectives, delivering products that are not just developed faster but also offer higher quality and greater value to customers.

Insights and Innovations: Unpacking the HBR Future of Business Conference 2023

Julie Yakunich — Tue, 21 Nov 2023 15:21:05 +0000

Insights from the Harvard Business Review's Future of Business Conference

The Harvard Business Review's Future of Business Conference on November 9th offered a wealth of insights from industry leaders. With any conference, there are some awesome talks...and some duds. But I found almost all of HBR's speakers to be really good this year.

Here are my key takeaways from the event:

CEO Q&A: Jensen Huang of NVIDIA - A New Computing Era

Human in the Loop: Will AI Always Need a Human?

Jensen Huang emphasized the importance of keeping humans in the loop with AI. He believes that, for the foreseeable future, it's essential not to exclude human oversight and intervention in AI processes.

Theory of Management

Huang's approach to management at NVIDIA is characterized by a flat organizational structure. This model is designed to facilitate extensive communication across different levels of the organization.
A key principle in this structure is the rapid and barrier-free flow of information. Huang believes that information must move at "light speed," with no boundaries or barriers, to ensure efficiency and responsiveness in a dynamic tech environment.

Mini-Masterclass: Lynn Perry Wooten of Simmons University - Leading Through Uncertainty

Defining Resiliency

Wooten's definition of resiliency in leadership is about preserving and experiencing growth, making positive adjustments amidst challenges, changes, disruptions, and adversity. The goal is not just to survive but to thrive.

Mini-Masterclass: Frances Frei of Harvard Business School - How to Solve Any Problem Quickly

Key Strategies

Frei advocates for a shift from the "move fast, break things" approach to "move fast, fix things."
Trust is the foundation of effective problem-solving.
Deliberation in actions is crucial to avoid what Frei terms "exhausted mediocrity."

Mini-Masterclass: Prithwiraj (Raj) Choudhury of Harvard Business School - Success Strategies for Hybrid and Work-from-Anywhere Teams

Hybrid and Remote Work Insights

Choudhury emphasizes a trend towards 50% occupancy in offices, with employees showing a willingness to accept a 5-7% pay reduction for the flexibility to work remotely.
He highlights the unique model of Zapier: 270 employees, no physical office, presence in 18 countries, and a diverse workforce.
The ideal model, according to Choudhury, is "work from anywhere," not just from home. This involves flexible hybrid models where teams choose their working arrangements based on what suits them best.

Embracing Flexibility and Experimentation

Choudhury advocates for allowing teams to pick their model depending on what works for them and to experiment with different approaches.
One of his favorite examples of a flexible model in action is in Japan, where small conference rooms have been set up on train platforms. These rooms enable teams to assemble and collaborate effectively in unconventional but highly practical settings. This example illustrates the innovative ways in which workspaces can be adapted to suit the needs of a hybrid workforce.

Andrew McAfee of MIT Sloan School of Management - How Geeks Hacked Leadership

The Geek Way

McAfee discusses his book, "The Geek Way," noting how "geeks" effectively manage large, complicated efforts.
He contrasts the old "waterfall" method with the new "agile" approach.
The focus is on value creation and attractiveness, with businesses being built to be less brittle by embracing openness.

Brad Lightcap COO of OpenAI - The Generative AI Revolution

Evolution of Generative AI

Lightcap discusses the evolution towards personalized GPT models, emphasizing how the technology is becoming more adept at understanding individual work contexts and needs.
The focus is on making Generative AI highly assistive, capable of accomplishing tasks efficiently while regularly seeking human input for guidance.

Risk Mitigation Strategies

A key aspect of OpenAI's approach under Lightcap's leadership involves engineering models that are aware of their limitations and resilient to attacks. This includes developing a quantified sense of right and wrong within the AI models.
An example of this cautious approach is the delayed release of GPT-4. OpenAI chose to wait six months before releasing the model, dedicating this time to thoroughly evaluate risks and safety concerns.
Lightcap also addresses the challenge of runtime safety: ensuring models are not abused and developing strategies to prevent such misuse.

Social Responsibility and Adaptation

OpenAI is not only focused on technological advancements but also on helping society develop the necessary "antibodies" to absorb the impact of Generative AI. This involves fostering institutions, social norms, and regulations that can adapt to and integrate these advanced technologies.
A significant part of their mission is to assist people in adapting to the changes brought about by Generative AI, ensuring a balance between innovation and societal well-being.

The Future of Project Management: Leveraging AI to Streamline Operations and Enhance Communication

Julie Yakunich — Thu, 16 Nov 2023 21:33:40 +0000

In the realm of software development, developers have long been leveraging specialized tools designed to make their lives easier. From integrated development environments (IDEs) to version control systems, the focus has been on streamlining the development process. But what about the other key players in the project lifecycle, such as project managers, business analysts, and scrum masters? As a seasoned project manager, I've been researching tools that can make the often complex and multifaceted role of project management more manageable.

The Art and Science of Project Management

Project management is both an art and a science. It requires a delicate balance of technical skills, such as resource allocation and timeline management, and soft skills like communication and team leadership. In fact, communication alone constitutes about 90% of a project manager's job. This is where technology, and more specifically Artificial Intelligence (AI), can play a pivotal role.

Focusing on the Mundane Yet Critical Tasks

My research has primarily focused on tools that can alleviate the mundane, time-consuming, and downright annoying aspects of my day-to-day responsibilities. For instance, taking meeting notes is a task that, while necessary, can be incredibly tedious. This is an area where AI can offer significant advantages.

Cost and Data Privacy Concerns

As a consultant, two of the most critical factors I consider when evaluating new tools are cost and data privacy. It's imperative to ensure that any tool we implement is both cost-effective and compliant with data privacy regulations, especially when dealing with client data. We've been able to implement a private Generative Pre-trained Transformer (GPT) that doesn't send our data back to the model, known as a Local Language Model (LLM). While it doesn't have all the features of OpenAI's ChatGPT, such as useful plug-ins like the JIRA plug-in, it serves its purpose well.

The Current State of AI in Project Management

As of now, AI hasn't drastically altered the way we manage software development projects. However, it's not something that I see as a threat to my job security either. Rather, AI can be a valuable assistant that makes my days smoother, easier, and faster.

How I Use AI in My Role

Meeting Notes: AI assists in transcribing and summarizing discussions.
Status Reports: AI helps in automating the generation of comprehensive status updates, even if I have to copy the outputs into existing templates.
Project Charters: AI can help draft initial project charters based on predefined parameters.
Technical Language Interpretation: Complex jargon is simplified for better comprehension.
Document Summarization: AI condenses lengthy documents into digestible summaries, and with AI, I have the ability to tell it exactly what format I want.
Idea Validation: AI serves as a sounding board for brainstorming sessions.
Gap Analysis: AI reviews my work to identify any missing elements or areas needing more detail.
Content Creation: Even this article was facilitated by AI.

While AI may not have revolutionized project management yet, it's making steady inroads into making the lives of project managers more efficient. From automating mundane tasks to acting as a reliable assistant for complex responsibilities, AI is set to play an increasingly significant role in the future of project management.

By embracing these technological advancements, we can look forward to a future where project managers can focus more on strategic decision-making and creative problem-solving, thereby elevating the art and science of project management to new heights.