Marketplace Partner Spotlight: Just
We sat down with Anton Velychko, the solo founder behind Just: AI Assistant for Jira — a Forge-native app that helps product and engineering teams turn rough Jira issues into structured, actionable work. Here's his story.
Question: What's your role at AI Apps?
Anton: I'm the founder — and also the developer, designer, marketer, and the guy writing this post. AI Apps is a one-person studio. I like to say my skills are less "T-shaped" and more "holding the weight of the world with spaghetti arms." But honestly, I wouldn't have it any other way. Solo means every decision is yours, and every win feels personal.
Question: How did you end up building Just?
Anton: It started with a job application, of all things. The task was to analyse a project and map out a transformation plan. I spent 30 hours on it — way out of scope, as tends to happen with me. The company passed. Too bold, too forward-thinking — their words. So I said: fine, I'll just do it myself. With blackjack and AI.
I've been in the Atlassian ecosystem for 15 years — since I was 20, studying development at university. Jira and Confluence have been part of almost every project I've worked on. So when I decided to build something, it felt natural to build it here, where the actual work happens.
See it in action — full 6-minute walkthrough:
From empty ticket to complete spec, including clarifying questions, image generation, apply back to Jira and founder walkthrough.
Question: The first version of Just wasn't quite right — what happened?
Anton: It was powerful but complex. Think of an IDE — lots of control, lots of options, lots of decisions required from the user before anything useful happened. People got lost in the first session. I watched it happen and knew something fundamental had to change.
So I rebuilt the core flow from scratch. The new version works like this: you open any Jira issue, click generate, pick an insight that feels relevant, optionally answer a few clarifying questions — and get a ready result immediately. Several design states, web search for current data, answers from the best AI models available. Two clicks, and a full spec for your ticket is ready.
That simplicity was hard to get to. But it was worth it.
Question: What makes Just different from what's already in Jira or from other AI tools?
Anton: Most AI tools for Jira just generate things. You give them a vague issue, they give you a vague output. Just works differently — it asks clarifying questions first, surfaces the decisions your team hasn't made yet, and only then builds a result.
There's a principle I keep coming back to: AI is like a genie — it does what you say, not what you mean. The gap between those two things is where most projects go wrong. Just helps you close that gap before a line of code is written.
On the technical side — Just connects to all five major AI providers simultaneously: Anthropic, Google, OpenAI, Mistral, and xAI. No lock-in. You take the best from each: reasoning from Claude, search from Gemini, images from Gemini Nano-Banana 2 model with consistent style across screens. Latest models, your own API keys, no middleman, no training on your data.
Question: Who is your typical user and what do they get in the first 10 minutes?
Anton: A product or project manager working fast — too fast to write a proper issue description, which creates blockers downstream. Engineering asks questions. Clarifications take days. Work starts on the wrong assumption.
In the first 10 minutes, I expect them to open Just on a real current task and see what it can do. But the magic really kicks in once they've added context about their project — stack, design system, audience. Set up once, reused across every ticket. With that in place, the outputs aren't just good — they're genuinely surprising.
The moment I knew it was working was when I added the clarifying questions feature. I'd seen it in Claude Code and OpenAI and was impressed. When I built it into Just and used it myself — on my own tasks, my own features — I realised: the questions don't just help the AI. They help me understand what I actually want. That's the real value.
Question: What's next for Just?
Anton: Two things. First — automatic context collection. Right now users fill in project context manually. I want Just to gather that automatically from their Jira history and structure. Less setup, better results from day one.
Second — image understanding. Currently attached screenshots and designs aren't read by the app. Adding OCR and vision support means a designer can attach a reference screen and Just will actually use it. That changes a lot.
The direction stays the same: shape your vision before you build. Not just for PMs — for anyone who touches a Jira issue and wants to start from clarity instead of guesswork.
Just is available on the Atlassian Marketplace. Built in Ukraine. 🇺🇦
2 comments
