AI development and automation
Agents and workflows that do the busywork inside your tools.
LLM features inside your product, agents that move data between the systems you already run, and computer-vision pipelines on live video. Built as software, with logs, tests, and a person in the loop where it matters.
What we build
- 01
LLM features in your product
Summaries, extraction, drafting, classification, chat. Wired into the app your people already open, with cost and latency budgets.
- 02
Agents on your systems
Workflows that read the inbox, update the CRM, reconcile the sheet, and ask a human when they are unsure. Every step logged.
- 03
Computer vision
Detection and recognition on camera feeds and images, from model choice to the alert someone actually sees.
- 04
Data plumbing
The pipelines, queues, and background jobs that make AI features reliable instead of a demo that works on Tuesdays.
What we reach for
- OpenAI, Anthropic, Gemini, open-weight models
- Python, TypeScript
- Queues and background workers
- YOLO-family detectors, face embeddings
- Evaluation harnesses and tracing
From production
How we ship AI, from systems we run.
A second, adversarial pass
SigmaCPMSigmaCPM re-checks bond terms with a second model run. When the two disagree, the value lands in a disputed state for a person to resolve, instead of going into a bid.
Local inference, one sighting per vehicle
Acasa SentinelAcasa Sentinel runs YOLOv8 with ByteTrack, then plate detection and OCR per track with a multi-frame vote, so each vehicle produces one sighting instead of hundreds. All inference runs locally, and no footage leaves the network.
Fidelity before creativity
KraftrKraftr feeds structured garment, model, and scene descriptions into a prompt builder that keeps the wardrobe and the model's identity locked across every angle, with moderation on every generation.
Candidates, not verdicts
Acasa SentinelWatchlist hits in Acasa Sentinel go to a queue for an operator to confirm or reject, with evidence crops and an audit trail. The system proposes; a person decides.
Honest trade-offs
When you should not use AI
If a rule, a form, or a database query can do the job reliably, use that: it is cheaper, faster, and easier to test. We use a model when the input is messy, like documents, images, video, or free text, and when a person can check what it produces.
Proof
Shipped, not pitched.
01Case study
Acasa Sentinel
Give it a number plate. It returns the vehicle's route across a CCTV network.
A vendor-neutral video analytics system built for the Gujarat Police Hackathon Innovation Challenge 2026. It reads number plates, re-identifies vehicles by appearance across cameras, reconstructs routes, and raises watchlist alerts for an operator to confirm.
Operator console (web) · API
02Our product
Kraftr
One flat-lay photo becomes catalog-ready 4K on-model imagery.
An AI catalog studio for fashion and jewelry brands. Upload one flat-lay or hanger shot per SKU and get on-model imagery at up to 4K, in ten aspect ratios, sized for Amazon, Myntra, Flipkart, and Shopify.
Web · iOS · Android
03Our product
SigmaCPM
Construction project management where AI reads the tender and shows its work.
One project record for civil contractors, from tender to closeout: bids, contracts, RFIs, vendor quotes, and drawings. The AI pulls line items and bond terms out of tender PDFs and cites the page for every value.
Web · API · Chrome extension
04Our product
NoteTube
A lecture becomes a study system: chat, flashcards, quizzes, all grounded in the source.
An AI learning platform that turns PDFs, YouTube lectures, recordings, and notes into a structured notebook with grounded chat, spaced-repetition flashcards, and graded quizzes.
Web · iOS · Android · Chrome extension · MCP server
How we work
Build, then keep building.
- 01
Sit with the operation
We watch how the work actually happens. Inboxes, tools, handoffs. No workshop theater.
- 02
Ship the product
A working web or mobile app. Something your people can open on Monday, not a prototype graveyard.
- 03
Wire the automation
Agents and workflows on the product and the systems you already have.
- 04
Stay on it
We do not vanish after launch. The operation changes. The software should too.
Questions
AI questions we get asked.
We do not have an ML team. Is that a problem?
No. That is who we build for. We own the models, the pipelines, and the product around them, and we document it so your team can run it.
Which model provider do you use?
Whichever fits the job on quality, cost, latency, and data rules. We design so the model can be swapped without rewriting the product.
Can you automate a workflow that spans several tools?
Yes. Most of our automation work is exactly that: inboxes, spreadsheets, CRMs, and internal tools that do not talk to each other.
AI development
Tell us what you need built.
Thirty minutes on how the work happens today. You leave with a written plan: scope, timeline, and a number.