How a Mumbai D2C brand shipped 3 AI systems in 12 weeks instead of building a team that never came.
Loftwell Furniture is a Mumbai-based D2C home furnishings brand. Series B funded (₹110 Cr raised), 165 employees, ₹180 Cr ARR across modular furniture, lighting, and home decor. They ship to 84 cities and run 6 experience stores.
Karan Mehta, CTO, joined 14 months ago with a board mandate to make Loftwell "AI-first by FY26" — and a hiring loop that wouldn't close.
Karan had three AI engineer roles open since January. He'd made 8 offers across 9 months. Five candidates ghosted post-offer for AI startups paying 1.8× his band. Two joined; one quit inside 4 months. Meanwhile, two direct competitors launched AI room visualizers and an AI styling assistant. Loftwell's customer service still ran on a Zendesk queue with 11.2-hour first-response times. Demand forecasting lived inside a 412-row Excel managed by one analyst — last quarter's misforecasts caused ₹3.2 Cr in inventory writedowns.
The board meeting opened with a slide of a competitor's AI demo and ended with the CEO publicly committing to ship three AI features in 90 days. Karan had no team. Then a ₹14 Cr B2B interior design contract walked out the door because Loftwell couldn't demo AI-assisted moodboarding.
Founders pinging on Slack, procurement asking when the forecast fix would land. The hiring strategy was over.
Vynara embedded a 4-person pod with Loftwell for 12 weeks: 1 PM, 2 AI engineers, 1 designer. Instead of a roadmap deck, the pod shipped to production: a demand forecasting agent that replaced the Excel model; a room visualizer that turns a customer's photo into furnished mockups using Loftwell's actual catalogue; and a customer service triage agent that auto-resolves 38% of tickets.
Stack stayed boring: Python, Postgres, OpenAI and Anthropic APIs via OpenRouter, on Loftwell's existing AWS. Pricing, returns approval, and VIP escalations stayed human. Forecast runs, first-touch CS, and visual recommendations became automated.
I spent 9 months in hiring cycles and shipped zero AI. Vynara shipped 3 systems in 12 weeks for less than one senior hire's annual CTC. We're hiring people to maintain what Vynara built, not invent from scratch.
The forecasting agent paid for the entire engagement inside its first quarter. Procurement actually trusts it more than the Excel — which tells you everything.
We embed, ship to production, train your team, and leave the runbooks. If your last board meeting opened with a competitor's AI demo, you don't have a hiring problem — you have a shipping one.