This is our first time participating in Codegeist, and it has been quite a challenging yet rewarding journey.
Our decision to participate was mainly driven by one important question: How can we leverage AI and make it an active ideation and collaborative partner within Confluence?
Additionally, at the start of October, we had a new CTO which kind of led everyone, especially developers, to try and give a good first impression.
Our fast growing team already consists of nearly 20 members from different backgrounds, experience levels, and especially working remotely from different locations. This hackathon was a great opportunity to gather everyone at the office at the same time and see how we can collaborate together on such an important project.
In this story, we’d rather avoid technical jargon and instead highlight the team dynamics during this journey.
How the journey started
The fact that we’ve already had an Ideation app within Confluence helped us gather valuable insights in terms of usage patterns and client preferences. This is when we realized two important findings:
First, the process of turning a thought into an idea and properly categorizing it can be challenging for some. The process often gets hindered by the complexities of turning a vague notion into a structured concept.
Second, the initial evaluation process could quickly become tiring and time consuming. This was a crucial pain point for teams, and it was clear that there was an opportunity to make this process more efficient and data-driven.
The idea started to take shape after just a couple of brainstorming sessions.
We thoroughly discussed the two important findings and decided to include AI as just an assistant (not to take over the ideation process completely). This way, the final say, be it during creation or evaluation rests solely on the user’s hands.
The product design team drafted a couple of mock-ups that more or less mirror what we had in mind. Then, all the involved parties came together to discuss the designs and share their feedback. Once the decision was made for the entire user experience and feel of the app, development started shortly after.
Meet the team
| Name | Role | Role in Project | What They Brought to the Table |
| Sofien | COO | Oversee Project | Big brother vibes and feeding the team |
| Alex | CTO | Oversee Project | Well…He brought the table |
| Melek | CPO | Oversee Project | Code reviews and preventing people from quitting |
| Fares | Marketing | Tell this story & prepare the demo | Boundless enthusiasm and marketing magic |
| Siwar | Marketing | Communication | That entry-level energy and fresh ideas |
| Salma | Design | UX & UI Design | Creative flair and eye-catching designs |
| Sami | Dev | Back and Front end development | Sleepless nights and Batman energy |
| Yasmine | Dev | Back and Front end development | Everything great about the youth of today |
| Firas | Dev | Back and Front end development | Pulling like there is no tomorrow |
| Tarek | Dev | Back and Front end development | Pulling like there is tomorrow |
| Skander | Support | Prompts | Testing the limits of Chat-gpt |
| Manel | Dev | Back and Front end development | Crafting code with a touch of brilliance |
| Nadine | Dev | Back and Front end development | Development expertise and wisdom |
| Chayma | QA | Testing | Thorough testing and expert bug extermination |
| Amal | QA | Testing | Expert guidance and quality control |
| Ali | Dev | Back and Front end development | Notable work during his notice period |
| Haythem | Dev | Back and Front end development | Started the whole thing and went back to school |
- What’s VectorIA and why you should give it a try
VectorIA is designed to facilitate the process of creating, categorizing and evaluating ideas inside Confluence.
During the idea creation process, VectorIA acts as an ideation assistant helping users generate idea titles, correct grammar & spelling mistakes, automatically associate ideas with predefined categories and more.
When it comes to idea evaluation, we’ve decided to equip evaluators with AI generated scores based on criteria such as feasibility, sentiment, competition, etc. These scores will help evaluators grasp an initial understanding of a given idea, but the final evaluation score depends on their own assessment.
How we built VectorIA
The development process lasted for nearly a week. The team worked simultaneously on the back and front end development.
All the while, everyone contributed with some prompts on Chat-gpt especially for the evaluation process. Then through trial and error, we started to combine prompts, properly guide OpenAI responses and came back with ones that kind of reflect what we wanted to see from the start.
With Forge and Confluence Cloud RestAPI, We’ve built a global interface accessible via the Apps Menu. This is where users can create and submit their ideas, which are then transformed and converted into Confluence pages within selected spaces. Idea pages act as blank canvases (or templates) that can be enriched over time with comments and insights by idea contributors. Ideas can also be voted upon, commented, and shared.
During idea creation, VectorIA becomes the companion of creators, carefully reviewing syntax and offering suggestions to enhance the idea’s structure. For this, we’ve leveraged OpenAI APIs to convert specific fields into prompts that AI can go through and check for potential enhancements. The results will then be visible by selecting Enhance AI. The AI also plays a role in validating the selection of the appropriate category, ensuring that each idea finds its perfect place.
When it comes to idea evaluation, we’ve also relied on OpenAI APIs to evaluate ideas based on various criteria including feasibility, innovation, impact, and more.
What we’ve learned:
- The Power of friendship and teamwork:
One of the most valuable lessons we took away from this incredible journey is the age-old cliché, “teamwork makes the dream work.” It might sound like a well-worn phrase, but its truth became clear throughout our journey. With team members from diverse backgrounds,, coming together to develop VectorIA, we grew closer as a team. The hackathon acted as a crucible, bringing out the best in each of us, and it was this collaborative spirit that propelled us to successfully create VectorIA.
- Embracing Forge:
While some of us had a degree of familiarity with Forge, none of us had previously embarked on a project of this magnitude using this framework. The hackathon served as an invaluable opportunity for us to dive into Forge headfirst. This rich framework provided us with the means to turn our ideas into tangible, functional reality. The experience was enlightening and expanded our horizons as developers.
- The Art of Prompt Engineering:
Delving into the realm of AI for the very first time, we embarked on a journey of prompt engineering. We immersed ourselves in understanding how the AI model responds to different instructions, conducting extensive tests to discover the most effective way to harness AI’s potential for VectorIA. It was a path filled with trial and error, but the knowledge we gained was immensely rewarding, offering insights that enriched the entire team’s skill set.
We hope you give VectorIA a try. With your feedback and bug detection (you won’t find many hopefully :p), we can take this idea to Amsterdam and beyond.
