Real problems, shipped systems.
How teams use SoftmaxAI to get agents, voice AI, and grounded GenAI into production — described the way engineers do, by outlining the problem, the specification, and what was ultimately shipped.
A voice agent that onboards patients and books appointments
A conversational voice agent that welcomes callers, tells whether the patient is new or returning, leads new patients through the onboarding process, and arranges appointments from start to finish — so routine calls are handled in a natural way and staff can concentrate on the interactions that require a human.
Read case study →Retail & E-commerce · Chatbots & VoiceA voice agent that answers customer calls and recommends the right product
A production-grade conversational voice agent that customers can access through a regular phone number in order to check the status of their orders and receive personalized product recommendations — an actual telephony system designed for concurrent, natural conversations rather than a scripted IVR.
Read case study →Technology & SaaS · GenAI & RAGA knowledge-grounded assistant that speeds up support resolution
An internal support assistant which enables employees to resolve technical problems by asking questions in plain language and receiving answers based on the organization's own documentation, rather than having to go through error logs and refer to dispersed sources.
Read case study →Media · GenAI & RAGAutomatic video metadata that creators can rely on
Production-quality AI incorporated into a media platform's product to automatically convert video content into summaries, titles, and timestamped chapters, thereby enabling creators to spend less time on manual cataloguing.
Read case study →Technology & SaaS · GenAI & RAGTurning raw user feedback into actionable insight
A production-grade pipeline on Google Cloud which automatically summarizes rating-based comments and identifies their sentiment, thereby converting large amounts of unprocessed user feedback into insights that teams can act on.
Read case study →Manufacturing · ML & VisionVideo analytics that reveal and remove assembly-line bottlenecks
LineSense is an AI-powered video analytics platform which constantly monitors workstations and converts motion data into objective and unbiased insights, providing manufacturing teams with a factual, up-to-date view of where time is being lost on the production line.
Read case study →Industrial Safety · ML & VisionReal-time hazard detection that keeps workers safer on site
Safety+ is a computer vision safety system which integrates existing CCTV to detect hazards and send alerts to the teams before incidents occur; it is a scalable and production-quality platform for proactive risk management.
Read case study →Retail · ML & VisionAutomated shelf intelligence that keeps stores stocked and compliant
A computer vision system which automates the process of checking inventory in stores and provides real-time data on the shelves — replacing manual and error-prone methods of tracking stock and compliance with continuous, objective supervision.
Read case study →Retail · ML & VisionTurning existing camera feeds into workflow insight
A computer vision system which analyzes existing camera feeds in order to gain an understanding of how work is carried out in stores — converting ordinary footage into clear, consolidated insights regarding tasks and workflow so that managers can plan staffing, not for the purpose of monitoring individuals.
Read case study →Finance · ML & VisionTurning market signals into research-grade decision support
A research and decision-support system of production quality which combines technical indicators with market sentiment in order to enable an investment research team to reason about equities in a more systematic way — a tool intended to assist and inform human judgment, not a source of financial advice or a guarantee of any outcomes.
Read case study →Security & Surveillance · ML & VisionLive CCTV surveillance that detects, reasons, and alerts in real time
A real-time surveillance platform which applies AI to live camera feeds as they are received, allowing it to detect, track, and recognize what it sees frame by frame, trigger alerts within seconds, and index every detection so that it can be searched for using text, image, or face queries. Operators specify in simple language what they are interested in — there being no need for configuration files or model training.
Read case study →Education & EdTech · GenAI & RAGA question-generation system that turns any source material into a question bank
A learning tool which takes a passage — whether the passage is typed, photographed, spoken, or taken from a video — and produces a gradeable question bank from it, including multiple choice, true/false, fill-in-the-blank, and short answer questions, each of which can be traced back to the sentence it was derived from.
Read case study →Healthcare & Eldercare · ML & VisionOn-device fall detection that alerts carers without recording the room
A self-contained edge device, consisting of a single-board computer and a camera, which monitors a living space for signs of a human fall and sends an alert to carers within seconds. It is able to tell the difference between people and pets, and it stores no footage — the only thing that ever leaves the device is the fact that a fall has taken place.
Read case study →Education · ML & VisionA well-being scoring tool that helps schools notice the students who need support
A behavioral scoring tool that combines structured questionnaires with the engagement signals a school already gathers, generates physical, mental, and emotional well-being scores over time, and informs staff when a student's trend calls for a human conversation – it brings the signal to their attention, and it is then up to people to decide how to proceed.
Read case study →Real Estate · ML & VisionA property recommendation engine that learns what a buyer actually wants
A recommender system for a property portal that blends the things a user says they want — gathered via a structured preference survey — with the things that people similar to them actually interact with, providing the properties in a ranked list together with a transparent match score for each stated requirement.
Read case study →Retail & Home Improvement · ML & VisionAn augmented-reality previewer that puts real tile designs on a customer’s own floor
Software which takes the place of the tile showroom's trial-and-error sample board allows you to aim your phone at a room or to load a photo of it and then see the different tile designs from the catalogue rendered onto the real floor in perspective, enabling you to swap between patterns, compare the various options and share the result before making a purchase.
Read case study →Logistics · AI AgentsFreight-quote agents that turn inbox chaos into booked shipments
Quote requests arrived as unstructured emails in every format imaginable. We built a multi-agent workflow that reads, extracts, prices against carrier rules, and drafts responses — with the ops team approving every send.
Read case study →Every engagement runs spec-first, agent-accelerated, human-owned.
A written specification that has been approved by the client prior to any coding taking place — covering the architecture, the edge cases, and the acceptance criteria.
The engineers arrange for the AI agents to carry out the specification in parallel, taking weeks rather than quarters.
The evaluations, tests, and security checks are carried out together with the build, and humans examine every decision that is important.
The system has been deployed to your GCP or AWS, complete with monitoring, production evaluations, and retraining loops.
Have an AI idea? Let’s pressure-test it — free.
A 30-minute call with our engineers, not a salesperson. You'll receive an honest assessment of feasibility, a preliminary architecture, and a clear next step.