FLUXUS / AI PRODUCT STUDIO

AI PRODUCT
STUDIO.

We design and build complete AI products.

From the interface and AI workflows to the integrations and infrastructure that run them. One partner, from your first brief to production.

BESPOKE AI / COMPLETE DELIVERY

AWS from
day one.

Your product in 8 - 16 weeks.
Fixed price for an agreed scope.

We agree the scope of your product around your users and business, then take responsibility for designing, building and launching it.

Building on AWS doesn’t have to mean months of infrastructure work. We’ve already built and tested the common components, and bring decades of experience delivering bespoke software for major enterprises and funded startups. We put that head start into your product, built on standard AWS services. We can deploy in your AWS account, or arrange hosting and ongoing support.

Define the product.

Agree the users, workflows and scope. Test the AI approach and work through the data, integrations and requirements for launch.

Design and build.

Develop the interface, AI behaviour and backend together. Build automated tests alongside the product, benchmark models against real tasks, and review working software throughout.

Launch and support.

Release through separate test and production environments, with repeatable deployments, monitoring and recovery procedures. Before launch, we agree how the product will be run and maintained, including any ongoing support from Fluxus.

The products below show the range of interfaces, AI capabilities and engineering we can bring to your project.

01 / PRODUCTS BUILT AT FLUXUS

Edition visual editor arranging the PERSONA art collection gallery
PERSONA · Gallery editor + published mobile site.
The PERSONA art collection website built in Edition, viewed on a phone

EDITION

AI website builder
& hosting platform

A multi-tenant SaaS product for creating websites with AI, editing them visually and hosting them on customers’ own domains. We built the web editor, CMS, publishing service and companion mobile app.

Multi-tenant SaaS · Web + mobile · AWSInside the Edition build
Deck presentation app with an editable sample growth-flywheel slide
Deck · An editable slide with individual shapes, labels and connectors.

Deck

Presentation & document editors

Create a presentation, then edit the text, diagrams and charts as individual objects. AI-assisted changes use the same structured slide model as the visual editor.

SaaS · Structured documents · In development
Browser game scene editor with a bookshelf selected and appearance and position controls
Daze · Selecting and adjusting an object in a playable 3D scene.

Daze

Browser games & 3D authoring

Build the tools for making a world as well as the experience of playing it. Select objects, change their appearance and placement, then switch from editing to play.

WebGL · Scene editor + runtime · In development

02 / ISO + TING - AGENTIC WORKFLOWS

Verification
built in.

Iso runs verified AI workflows. Ting applies that platform to software delivery.

Two products we’re building around the same principle: agents do the work, independent checks test the result, and failed checks send it back for correction.

Workflow builder showing a customer support process with drafting, checks, approval and sending
Workflow builder - a customer support process with checks, human approval and failure routes.

ISO

A platform for agentic workflows.

Iso is a multi-tenant SaaS platform for defining and running workflows that combine agents, tools and people. It provides isolated execution, verification gates, human approvals, and a record of outputs and costs.

Products built on Iso manage their own customer accounts. Iso separates each tenant’s work, credentials and usage.

SaaS · Workflow builder + API · Tenant isolation · In development

TING

A software factory, built on Iso.

Ting turns a brief into requirements, acceptance tests, implementation and a reviewed pull request. Tests are written before the implementation; independent reviews require evidence, and failures go back for correction.

Ting manages the software-specific experience: repositories, projects, tickets and delivery. Iso runs the workflows beneath it.

Software delivery · Test generation · Independent review · In development

03 / DATA PIPELINES, RAG & ANALYSIS

Put your data
to work.

Documents, conversations and live feeds. We build pipelines that collect and structure them, and products that make the results usable.

Clouseau investigation linking a finding in the public Enron email archive to its original source documents
Clouseau · Findings linked to original emails in the public Enron archive.

Clouseau

Document investigation

Extract facts from PDFs, email, spreadsheets and scans. Discover related material and ask questions with references back to the original evidence.

Desktop · Extraction · Retrieval · Analysis
Distill Ask screen: a bid question answered with three cited playbook rules
Distill · A question answered using relevant rules and their sources.

Distill

Company knowledge

Turn decisions in email, chat and documents into a company playbook. Ask about a new situation and retrieve the relevant rules and their sources.

SaaS + service · RAG · In development
Monitor sentiment report with positive, negative and neutral themes from social posts
Monitor · Sentiment and recurring themes in a social listening report.

Monitor

Social listening

Collect social posts, identify recurring subjects and sentiment, and produce reports that keep the underlying conversations close to the analysis.

Data pipelines · Classification · Reporting
Flowcast news view showing a story summary alongside a topic trends chart
Flowcast · A news summary alongside topic trends.

Flowcast

News & research feeds

Assemble incoming posts into summaries, track changing topics and show connections between stories. Readers can follow each summary back to its sources.

Data pipelines · Summarisation · Visualisation

04 / SPECIALIST PRODUCTS

Built for the job.

Centralised framework view showing demo AI systems grouped by EU AI Act, GDPR and ISO 42001 coverage
Centralised · Framework coverage across a demonstration AI portfolio.

Centralised

Governance & internal platforms

Register AI systems, assign owners and collect evidence against configurable frameworks. Approval workflows, auditor access and change history support ongoing review.

SaaS · Registries · Approvals · Audit history

Hoopsy

A mobile app with computer vision

Mobile app · Computer vision on AWS · Claude chat · Product analytics

We built the mobile app and AWS backend, with computer vision to analyse photographs of test strips. The app guides users through taking a photograph and presents the result with visual and spoken feedback. We integrated Anthropic Claude for streaming chat within the app.

We also built the client dashboard, which tracks key usage metrics and provides ongoing AI-generated recommendations for improving the product, based on how people actually use it.

Hoopsy client dashboard showing usage metrics and AI product recommendations with illustrative sample data
Client dashboard - usage metrics and AI recommendations. Illustrative sample data and recommendations.
Hoopsy mobile app demo reviewing a test-strip photograph before analysis
Mobile app - photo review.
Hoopsy mobile app demo showing the conversation about a test result
Mobile app - in-app chat.

A HEAD START ON THE ENGINEERING

Components we’ve built.
A head start for you.

We build on standard AWS serverless services, using components we’ve already developed and tested in our own products: customer accounts and permissions, billing, an AI model gateway, usage and cost controls, storage, monitoring and deployment.

We adapt and integrate the relevant components into your product. That reduces development time and cost, gets working software into your customers’ hands sooner, and brings existing tests, access controls and operational safeguards into the build.

For a team that already owns the product and needs specialist help, our AI consulting service ↗ brings that engineering experience into your team.

ONGOING MODEL EVALUATION

The right model
for your workload.

We benchmark models on your application’s real tasks, measuring quality, reliability, response time and cost.

As new models arrive, we rerun the same workloads against the model already in use. The results show where a change can improve the product or lower running costs, and where it would introduce a regression.

YOUR IDEA / OUR PRODUCT STUDIO

Tell us who the product is for, what it needs to do and whether you’re starting with an idea or an existing pilot.

Email us: info@fluxus.io

Built on edition.pm