AloBizness
Mekan AI targets four groups in Azerbaijan: homeowners and renters planning interior improvements, furniture retailers seeking digital sales opportunities, real estate agents presenting unfurnished properties, and professional interior designers looking for clients. Customers struggle to visualize furniture in their own rooms, compare suitable products, and stay within budget. Furniture stores face challenges reaching relevant buyers, while real estate agents struggle to demonstrate the potential of empty properties. Existing workflows often require multiple disconnected tools, manual coordination, and significant time. Mekan AI aims to connect AI-assisted interior visualization, real furniture discovery, virtual staging, and professional design services in one platform.
Preliminary score
- Value for the user14.0 / 25
A specific problem and a clear outcome from solving it.
- Prototype and use of AI12.0 / 30
A working core scenario, and what AI actually contributes to the result.
- Quality testing10.0 / 20
Tests, examples of failures, and comparison with the current approach.
- Feasibility7.5 / 15
Data requirements, running costs and a clear next step.
- Originality4.5 / 10
A meaningful difference from common solutions.
What the judges wrote
- Claude Opus 5.5Preliminary46/100
- Strengths
- Your web application covers a broad flow beyond image generation: hover-to-product cards on a design image, cart, order requests, designer handoff and role-based dashboards, and it is backed by 23 unit tests and Playwright browser tests. You are open about limitations. You state clearly that the public demo uses prepared images and that live Qwen and Grounding DINO generation is still unvalidated, and that honesty is valuable. Your deck is clear and carefully hedged on market size and pricing, and it ties the product to local Baku retailers with a concrete set of revenue hypotheses.
- To improve
- The AI core is not yet demonstrated. Run the real pipeline (Qwen Image Edit + Grounding DINO) on 15–20 real Baku room photos and show the successes, failures, generation time and cost per image. Your tests check UI and matching logic, but not AI quality. Measure detection accuracy and how often the matched products truly resemble the generated furniture. Narrow your four audiences to one buyer, for example a furniture retailer, and name the metric that would justify adoption, such as inquiry conversion rate versus today's Instagram or WhatsApp sales. Also align the submission text, which says the model is 'pending verification', with the README so it states clearly which model is in use.
- GPT-6 AstraPreliminary50/100
- Strengths
- Your Baku-focused concept addresses a concrete gap between room-design inspiration and finding locally available furniture within budget. Your reported tests cover product-matching constraints, accessibility and authorization, and identify specific hover-card and test-cleanup failures that were corrected. You also clearly distinguish prepared demo images from live AI functionality that still needs validation.
- To improve
- Your next priority should be to demonstrate one live photo-to-design-to-product-match-to-order-request workflow, identifying the actual generation and detection models used. Provide reproducible test results, including failures involving room-layout preservation, furniture detection and unsuitable catalog matches. For the enterprise track, choose one primary business user and measure an adoption metric, such as retailer time to prepare a suitable furniture shortlist, against the current manual process. Strengthen feasibility with a concrete retailer-catalog acquisition and update plan, measured generation costs and latency, and a narrowly scoped pilot.
The team
Yusif Mammadov
developer
Akif KarimovLead
product
Ahmad Afandiyev
data