Senators
Resolve is built for businesses that sell products or services through their websites, starting with online merchants without dedicated retention teams. Every day, potential customers browse products, abandon purchases, cancel orders, or stop engaging. Businesses can often see these events, but struggle to understand what may have caused them or decide how to respond. Generic recovery campaigns send similar messages to everyone, while investigating each customer manually takes too much time. Resolve brings available customer behavior and purchase history together, uses AI to interpret the context, and recommends a personalized response instead of treating every customer the same. We're starting with Shopify, but our vision is to make intelligent, personalized retention accessible to any online business.
Preliminary score
- Value for the user16.0 / 25
A specific problem and a clear outcome from solving it.
- Prototype and use of AI17.0 / 30
A working core scenario, and what AI actually contributes to the result.
- Quality testing13.0 / 20
Tests, examples of failures, and comparison with the current approach.
- Feasibility8.0 / 15
Data requirements, running costs and a clear next step.
- Originality4.0 / 10
A meaningful difference from common solutions.
What the judges wrote
- GPT-6 AstraPreliminary62/100
- Strengths
- Your local demonstrations cover a coherent workflow from customer events through AI explanation, message drafting, approval, and persistent history. Your six-scenario baseline comparison usefully distinguishes Gemini's contextual writing from rule-based action selection. The recent-contact violation caught by your independent validator is a concrete safety finding, and you clearly distinguish simulated recovery from actual revenue impact.
- To improve
- Your first priority should be reliable hosted persistence and a reproducible trace of the complete workflow available at submission freeze. Publish the six baseline comparisons and representative safety-test inputs and outputs, rather than relying mainly on pass counts. In your first merchant pilot, compare against existing Shopify recovery emails using incremental recovered orders, merchant review time, and unsubscribe rates. Document required data permissions, outreach consent, and per-store Gemini, email, and hosting costs.
- Claude Opus 5.5Preliminary54/100
- Strengths
- Your materials are unusually honest. The deck, submission and README keep marking what is simulated, what is unverified and what was really sent, which makes your claims easy to trust. Your safety design pairs Gemini drafting with a deterministic validator that rejects discount or urgency language, unknown links and contact-rule violations. You also give a concrete failure case where that validator caught a Gemini recommendation conflicting with recent-contact rules. Your rule-based baseline comparison across six scenarios is candid: rules already choose actions well, and Gemini's real contribution is contextual, uncertainty-aware wording.
- To improve
- Your story is inconsistent across materials. The deck pitches abandoned-checkout recovery for Maya. The README says cart and checkout signals are future work and the only verified automatic flow starts from order cancellations. Test counts also differ: 28, 46 and 69. At the freeze your hosted demo showed 'Storage is temporarily unavailable,' so reviewers could not run the core scenario end to end. Next time, add per-email Gemini running costs and one measurable target, such as the recovery rate versus Shopify's built-in abandoned-checkout email, to justify the $79/month price.
The team
Zarifa IlyasovaLead
developer
Javidan Ganbarov
data
Fizuli Hasanov
product