Shipped work: AI systems in production
In short
This is the work Mirai Minds has shipped: AI systems built for clients, and platforms we run ourselves. Each case study covers the problem, how the system works, the models and stack behind it, and what changed, with the source of every number. Startup clients are named; enterprise and public-sector clients are anonymised.
Shipped to productionUpdated
| Project | What we built | Result | Year | Status |
|---|---|---|---|---|
| Resin manufacturer (India) | Catalog chat assistant with text and photo product search | Pilot on a 15-product catalog sample | 2026–present | Pilot |
| Photo Experience | Generative photo editing, face search and the image pipeline | 18.7 s per generative photo on one A100 (was 60–90 s) | 2023–present | Pilot |
| iKoMatch | WhatsApp and voice fundraising assistant with explainable matching | 6,717 investor records; each day's matches graded by an LLM judge | 2026 | In production |
| CallAuditAIIn-house | Call audit service: recordings scored against a weighted rubric | Scorecards for an admissions counseling team's recorded calls | 2026 | In production |
| Ethnic-wear e-commerce brand (India) | Visual product search and abandoned-cart recovery calls | Visual search live for store staff since July 2026 | 2026 | In production |
| WhatsApp AgentsIn-house | WhatsApp agent service with memory, summaries and document search | Runs iKoMatch's founder conversations on WhatsApp | 2026 | In production |
| AI outbound-calling startup | Pipecat voice bot, campaign dialer and per-call recording | Live for outbound campaigns | 2025–2026 | In production |
| Voice AgentsIn-house | Multi-tenant platform for AI phone agents and calling campaigns | Live at voice-agents.miraiminds.co | 2025–2026 | In production |
| Indian public-sector refinery | Document AI: engineering drawings to SAP-ready bills of material | Delivered and handed over; every row reviewed before it reaches SAP | 2025–2026 | Delivered |
| US logistics software company | Chat agent over the client's platform APIs through MCP | Delivered; every tool call's arguments are checked before it runs | 2025 | Delivered |
| Gurbani learning app | Transliteration, explanation and speech API for Gurbani lines | Delivered to the app team as an authenticated API | 2025 | Delivered |
| Myva.ai and CallPaaS | Document-grounded chatbot platform and call-center AI with GPT | Both products built and extended across 2023–2025 | 2023–2025 | Delivered |
| Rouh | Desktop computer-use agent with MCP tools, scheduling and memory | Working desktop prototype, reviewed by the founder | 2024–2025 | Prototype |
AI agents and automation
Agents that read, decide and act inside your tools, with every tool call logged and a person approving the risky ones.
Resin manufacturer (India)
Catalog assistant with photo search for a resin manufacturer
Pilot on a 15-product catalog sample
AI agents and automation · Computer vision
iKoMatch
iKoMatch: an AI fundraising assistant on WhatsApp and voice
6,717Unified investor records
AI agents and automation · WhatsApp AI agents · Voice AI agents
US logistics software company
Operations chat agent for a US logistics software company
Delivered; every tool call's arguments are checked before it runs
AI agents and automation
Rouh
Rouh: a computer-use agent for the desktop
Working desktop prototype, reviewed by the founder
AI agents and automation
Voice AI agents
Phone agents that answer, qualify and follow up in English or Hindi, and bring in a person when the caller needs one.
Custom LLM and document AI
Language models grounded in your documents and data, measured on your own test set, with people reviewing what matters.
Indian public-sector refinery
Engineering drawings to SAP-ready bills of material
16SAP material-master fields per extracted row
Custom LLM and document AI
Gurbani learning app
Explanation and pronunciation tools for a Gurbani learning app
Delivered to the app team as an authenticated API
Custom LLM and document AI
Myva.ai and CallPaaS
Myva.ai and CallPaaS: document chatbots and call-center AI
Both products built and extended across 2023–2025
Custom LLM and document AI · Voice AI agents
Computer vision
Vision models that find faces and products, cut out backgrounds and edit photos, fast enough for a queue of real customers.
Photo Experience
AI photo pipeline for theme-park souvenir photos
18.7 sPer generative photo on one NVIDIA A100 80GB
Computer vision
Ethnic-wear e-commerce brand (India)
Visual product search and cart-recovery calls for an ethnic-wear brand
Visual search live for store staff since July 2026
Computer vision · Voice AI agents
Built in-house
Products we build and run ourselves
CallAuditAI
CallAuditAI: rubric scoring for counseling and sales calls
Scorecards for an admissions counseling team's recorded calls
Custom LLM and document AI
WhatsApp Agents
WhatsApp Agents: our service for AI WhatsApp conversations
Runs iKoMatch's founder conversations on WhatsApp
WhatsApp AI agents
Voice Agents
Voice Agents: our platform for AI phone agents
Live at voice-agents.miraiminds.co
Voice AI agents · WhatsApp AI agents
Method
How we measure and what we leave out
Every entry in the table above is a system that exists: live in production, delivered and handed over, in pilot, or a working prototype. The status column says which.
Case studies follow the same order: the problem, what we built, how it works in production, and what we would change. That last part is deliberate. Every system has trade-offs, and knowing them is more useful to you than a clean story.
FAQ
Asked on the first call
How are the numbers in these case studies measured?
Each metric names its source next to it: a benchmark run, a load test, a production log or the client's own records, with the month it was measured. If we could not measure something, we describe the outcome instead of estimating a number.
Why are some clients anonymised?
Enterprise and public-sector clients often can't be named publicly. We describe them by industry and country instead, and leave out anything that would identify them, such as locations, partner names or document samples.
Can we speak to a client you worked with?
Where the client agrees, yes. Ask on the first call and we will check who is happy to talk about the project.
Have a system in mind? Let's scope it.
A 30-minute call with an engineer who has shipped this before. You leave with a plan, a rough timeline and what it would take — whether or not we build it.