BlackBox AI | Portfolio & Consulting

AI-driven engineering, research, and consulting. We create practical solutions for real-world business challenges.

9Research Projects

107+Articles

UKBased Firm

HOW WE HELP

What We Do

Most businesses know AI matters. Far fewer know how to make it actually work. BlackBox AI bridges that gap — combining strategic consultancy with hands-on engineering to take you from first idea to production-ready system. No vague reports, no vendor juggling. Just end-to-end delivery from a team that builds what it recommends.

Our Focus

RESEARCH

Stay ahead of the curve

ENGINEERING

From prototype to production

We build working AI systems end-to-end — machine learning models, APIs, cloud infrastructure, and integrations — delivered in agile sprints with a full handover your team can run independently.

CONSULTING

Strategy without the fluff

Portfolio

Research Projects

Each project trials a specific AI use case or emerging technology — so we understand what's coming before it hits the mainstream, and our clients benefit from real-world insight, not just what's already here.

Drag to explore

Hardware

EdgeSense

A comprehensive IoT sensor integration system for EdgeSense, a native macOS AI desktop application. The update enables real-time data streaming from edge devices—including Raspberry Pi, Arduino (ESP32/ESP8266), and Nvidia Orin Nano—directly into LLM conversations. By leveraging WebSocket communication, mDNS zero-configuration discovery, and an efficient in-memory ring buffer architecture, the system provides seamless contextual awareness to AI assistants about physical world conditions.

RustTauri 2.0Axumtokiomdns-sdparking_lotserdereact18TypeScriptZustandTauri IPCTailwind CSS

Explore Project Source ↗

AI Use Case

What are the limitation of using Local AI with real-time sensor data?

Desktop Software

H I V E - Synthetic Environments

HIVE is a DIS-compliant simulation platform for unmanned swarm operations. Built on Tauri, React, and Rust, it generates standards-compliant IEEE 1278.1 entities with full NATO APP-6D symbology, simulates multi-sensor payloads across configurable threat environments, and streams live telemetry to EdgeSense for AI-driven operational picture simplification. Dozens of drones carrying CBRN, EO/IR, thermal, and RF sensors produce thousands of data points per second. HIVE bridges that complexity to a local LLM through EdgeSense, distilling noisy sensor feeds into one actionable paragraph a commander can act on. Designed for extensibility — supporting additional PDU types, HLA federation, waypoint guidance, and third-party simulator interoperability — HIVE serves as both a standalone simulation tool and a synthetic environment for testing AI-enabled C2 concepts.

ROS 2AirSimUnreal EngineDISPX4SITLHITL

Explore Project Source ↗

AI Use Case

How can we generate synthetic sensor data for model training?

Machine Learning

Edge AI using LLMs

This project explores the deployment of Large Language Models (LLMs) on the NVIDIA Orin Nano Super — a powerful yet compact edge computing platform. The goal is to assess performance, latency, and real-world usability of AI inference workloads in constrained environments. It combines hardware acceleration with software optimization to demonstrate intelligent edge solutions capable of processing natural language locally, without relying on cloud infrastructure. This research provides foundational insight for building secure, low-latency AI systems for robotics, IoT, and offline applications.

LLMNVIDIA Jetson Orin NanoONNX Runtime / TensorRTPythonDockerTinyLLMGCUFJavaScriptReact

Explore Project Source ↗

AI Use Case

What are the capabilities and limitations of Edge AI Hardware?

Data Analysis

Hardware Development TestBed

This custom-designed hardware testbed supports the experimental development and validation of solid rocket motor technologies. Engineered to integrate load cells, relay systems, and signal-conditioning electronics, the platform enables safe, repeatable testing of thrust, ignition control, and circuit reliability. The modular truss structure is optimized for mechanical strength and sensor accessibility, serving as a robust base for iterative R&D in propulsion and aerospace subsystems. The platform also acts as a proving ground for embedded software and real-time telemetry systems ahead of full motor integration.

Precision Load CellsAmplifiersRaspberry PiArduinoPCB DesignI2CSPIUARTData Logging3D PrintingHX711

Explore Project Source ↗

AI Use Case

How can physical hardware be used to test AI capabilities?

AI/ML

Host Buddy

HostBuddy aims to resolve the lost applications/deep research/tools created using AI. HostBuddy is locally deployed onto the users workstation and enables the user to copy and paste AI generated code into a run time environment so they can utilise tools created across multiple AI platforms, making them more accessible and manageable for the user. HostBuddy supports .html, React and Python applications and is deployed in a lightweight Electron application that can be distributed to Apple and Microsoft environments and run locally.

ElectronReactPythonNodeJShtml

Explore Project Source ↗

AI Use Case

How can hosting and sharing AI apps be made simpler?

Web App

Brain

Brain is a local-first knowledge system that turns your files, documents, and web sources into a structured, searchable knowledge graph — then exposes it to any LLM via a Model Context Protocol (MCP) server. Most tools for managing knowledge were designed for humans. They optimise for readability, navigation, and manual organisation. Project Brain explores a different view: it is a knowledge system designed primarily for machines — specifically, for large language models operating in agentic and semi-agentic contexts.

FastAPIReact + D3.jsChromaDBSQLiteClaude APIMCP Protocolsentence-transformers

Explore Project Source ↗

AI Use Case

How can large volumes of data be efficiently utilised by AI?

Web App

BlackBoxAI Website

The BlackBoxAI Website: A platform for sharing research projects with the goal of automating the Development workflow using coding agents taking User Feedback to Completed, Reviewed and Tested code commits.

Next.jsReactTypeScriptTailwindPostCssCustom CSS

Explore Project Source ↗

AI Use Case

How can user feedback be automated using agentic development ?

Web App

Network Influencer - Micro SaaS

A professional network visualisation and relationship management tool designed for maintaining and analysing LinkedIn connections. This enterprise-grade application helps users track influence, interest, and relationships within their professional network.

NodeJSReactNext.jsTypeScriptPostgreSQLGPT3.5 TurboSupaBase

Explore Project Source ↗

AI Use Case

How effective is AI at cleaning and enriching large datasets?

Web App

PlanPilot

PlanPilot is an AI-powered trip planning web application that transforms travel research from hours of manual work into seconds of automated intelligence. Users simply fill out a quick form with their destination, duration, budget preferences, and interests, and the application's AI agent researches and generates a complete, interactive trip itinerary. The platform leverages Gemini 2.5 Flash and Pro with web search capabilities to find real accommodations with current prices and booking links, curated attractions with entry fees, and accurate GPS coordinates for every location. All results are visualised on an interactive Leaflet map with colour-coded markers: purple for key destinations, green for accommodations, and red for attractions. Built with a server-less architecture on Vercel, PlanPilot combines a vanilla JavaScript frontend with Node.js server-less functions that proxy requests to Gemini 2.5. The application supports both AI-generated trips and manual JSON import/export, with automatic browser storage for persistence across sessions. The tech stack includes Gemini 2.5, Vercel for deployment, Leaflet.js for mapping, and OpenStreetMap tiles. With an estimated cost of just £0.01-0.05 per trip generation, PlanPilot makes professional-grade travel planning accessible and affordable for everyone, eliminating the need for spreadsheets, multiple browser tabs, and endless research sessions.

.htmlpythonvercelserverlessgithub

Explore Project Source ↗

AI Use Case

How effective is prompt injection requiring strict AI responses?

Activity

Recent Activity

Updates, features & bugs across all projects — last 90 days

43 total32 active11 completed

FeatureBugUpdateCompleted

MarAprMay

90d agoToday

Research

Research Knowledge

How We Help

Expert Consulting

Good AI strategy starts with honest advice from people who've actually built real systems. Whether you need a Discovery Workshop to find your footing or an ongoing retainer to keep things moving, we help you make confident decisions — without the guesswork or the jargon.

Drag to explore

Consulting

Service

Build Sprint: 4 to 6 Weeks

Build Sprint — 4 to 6 Weeks Take your prototype to production. Deployed, documented, and ready to run. A successful prototype proves the idea works. The Build Sprint turns it into something your business can actually rely on — a productionready system, integrated with your real environment, tested against real conditions, and handed over to your team with everything they need to own it. This is the natural next step after a Workflow Automation Prototype or Discovery Sprint. Fixed scope. Deliveryfocused. No drawnout engagements. What you get A productionready system — fully built, tested, and deployed in your environment, not a dev sandbox Integration with your existing tools and data sources — connecting to the systems your team already uses Automated tests and quality checks — so you know it works and can verify it keeps working Operational documentation — architecture notes, a runbook, and setup guides your team can actually follow Knowledge transfer sessions — structured handover so your team understands, owns, and can extend the system Postdeployment support window — we stay close for the first two weeks after golive to catch any issues fast How it works Week 1 — Design and Plan We take your prototype or a wellscoped brief and turn it into a production design. We finalise the architecture, agree on integration points, define acceptance criteria, and set the delivery plan for the weeks ahead. Weeks 2–4 — Build Iterative, sprintbased development with weekly demos. You see progress every week and can feed back in real time — no waiting until the end to find out it went in the wrong direction. Week 5 — Test and Harden Endtoend testing against your real data and environment. We resolve issues, run security and quality checks, and prepare for deployment. Stakeholder signoff happens here. Week 6 — Deploy and Hand Over Production deployment, knowledge transfer sessions with your team, and documentation handover. We stay on hand for the postgolive support window to ensure a smooth landing. Who it's for Teams who have validated an idea through a prototype or discovery engagement and are ready to build for real Organisations that need a working system in weeks, not months — without the overhead of a lengthy procurement or delivery process SMEs and innovation teams who want a trusted engineering partner to own delivery endtoend, then hand it over cleanly Why BlackBoxAI Most build projects fail not because the technology is wrong, but because scope creeps, handovers are rushed, and nobody ends up owning the outcome. The Build Sprint is designed to avoid exactly that: fixed scope agreed upfront, weekly visibility throughout, and a structured handover that leaves your team genuinely selfsufficient. Fully remote. Fixed price. Clean handover. No retainer required. Need ongoing support after golive? We offer flexible retainer packages to keep your system healthy and evolving.

View Details

Team AI Enablement Workshop - Half Day or Full Day

Consulting

Service

Team AI Enablement Workshop - Half Day or Full Day

Team AI Enablement Workshop — Half Day or Full Day Get your team using AI practically and confidently — in a single session. There's no shortage of AI tools. What most teams lack is the understanding of how to use them well: what they're genuinely good at, where they fail, how to prompt effectively, and how to integrate them into real daytoday work without creating new risks. The BlackBoxAI Team AI Enablement Workshop closes that gap in a single, focused session — tailored to your team, your tools, and your actual workflows. No theory for its own sake. No vendor sales pitches. Just practical, applicable skills your team can use the next morning. What you get A tailored session built around your team's role, tools, and daytoday work — not a generic AI overview Handson exercises using real examples from your workflow so the learning sticks A practical AI toolkit — a curated reference guide of tools, prompts, and use cases relevant to your function A clear picture of where AI helps and where it doesn't — so your team uses it confidently and responsibly A followup Q&A window — two weeks of async questions via email so the learning carries forward Session summary and takeaways delivered after the workshop for sharing with the wider organisation What we cover The session is tailored to your team, but typically includes: What AI can and can't do — a grounded, honest framing that replaces hype with clarity Prompting that actually works — how to get useful, reliable outputs from AI tools AI for your function — specific use cases for your team's role ops, finance, support, engineering, commercial Workflow integration — how to slot AI tools into your existing processes without creating bottlenecks or quality risks Risk and responsibility — data privacy, accuracy limitations, and how to verify AI output before acting on it What to try this week — a short list of immediate, lowrisk actions each attendee can take straight away Format options HalfDay Workshop 3 hours Focused on one function or one core use case. Ideal for a small team up to 8 people who want a fast, targeted session with handson practice. FullDay Workshop 6 hours Broader coverage across multiple use cases and functions. Includes deeper handson exercises, group problemsolving, and a working session where we apply AI to a real challenge from your organisation. Best for crossfunctional teams or leadership groups. Who it's for Teams who have heard a lot about AI but aren't sure how to use it in their actual work Organisations rolling out AI tools who want to ensure adoption is effective, not superficial Leaders who want their team to engage with AI critically and responsibly — not just enthusiastically Any function: ops, finance, HR, customer support, sales, marketing, project delivery, engineering Why BlackBoxAI We're practitioners, not trainers. Everything we teach comes from handson experience building and deploying AI systems — which means we know where the tools genuinely save time and where they confidently produce nonsense. Your team will leave with a realistic, grounded understanding of AI that makes them more effective, not just more excited. Fixed price. Remote or onsite. Tailored to your team. Want to follow up with a prototype or automation project? The workshop often surfaces the perfect place to start.

View Details

AI Document & Data Intelligence Prototype - 5 Days

Consulting

Service

AI Document & Data Intelligence Prototype - 5 Days

AI Document & Data Intelligence Prototype — 5 Days Your documents and data already contain the answers. AI can surface them in seconds. Most organisations are sitting on a goldmine of untapped knowledge — policy documents, contracts, reports, support tickets, internal wikis, historical data — that nobody has time to read, search, or make sense of at scale. BlackBoxAI builds you a working AI intelligence prototype in five days: a system that understands your content, answers questions in natural language, and surfaces the right information at the right moment. This isn't a generic chatbot. It's an AI assistant trained on your specific documents and data, built for your team's actual questions. What you get A working AI prototype — query your documents and data in plain English and get accurate, sourced answers Custom knowledge base — built from your uploaded documents, data exports, or internal content Source attribution — every answer references where it came from, so your team can trust and verify the output A simple, usable interface — your team can start using it immediately, no technical knowledge required Handover documentation — how it works, how to add new content, and how to extend it as your needs grow What we can build this on Internal documents — policies, procedures, contracts, handbooks, reports, meeting notes Structured data — spreadsheets, databases, CRM exports, support ticket histories Knowledge bases — Notion, Confluence, SharePoint, or similar platforms Customerfacing content — FAQs, product documentation, support materials Not sure if your content is suitable? Tell us what you have — we'll tell you what's possible. How it works Day 1 — Input and Scope You share your documents or data sources and tell us the questions your team needs answered. We agree on the scope of the prototype and the success criteria. Day 2–3 — Build We ingest your content, set up the AI model, configure retrieval and response logic, and build the interface your team will use. Day 4 — Test and Refine We run the prototype against real questions from your team, assess the quality of responses, and tune the system to improve accuracy and relevance. Day 5 — Handover Live walkthrough with your team, documentation handover, and guidance on how to add new content and evolve the system over time. Who it's for Teams drowning in documents who need answers fast — legal, compliance, HR, operations, customer support Organisations with valuable institutional knowledge locked in files that nobody can find or search effectively Startups and product teams who want to build an AIpowered knowledge or support feature into their product Leaders who want to give their team a private, secure AI assistant trained on their own internal content Why BlackBoxAI We've built AI intelligence tools for our own platform — including the assistant that runs on this website. We know how to go from raw documents to a working, reliable AI system quickly, and we know the pitfalls that cause these projects to produce confident wrong answers instead of trustworthy useful ones. Source attribution, accuracy tuning, and responsible design are built into our approach from day one. Fully remote. Fixed price. Clean handover. No retainer required. Want to take it further? We can build this into a production product your customers or whole organisation can use.

View Details

Consulting

Service

AI Workflow Discovery Sprint

AI Workflow Discovery Sprint — 5 Days Find out exactly where AI and automation will move the needle — before you spend a penny on build. Most AI projects stall not because the technology is wrong, but because the problem was never properly defined. The BlackBoxAI Discovery Sprint gives you five focused days to map your workflows, identify the highestimpact opportunities, and walk away with a delivery roadmap you can actually execute — not a strategy deck that gathers dust. What you get A clear workflow map — what happens today, who owns what, and where time, money, or quality is being lost Prioritised opportunity list — AI and automation candidates scored by impact vs. effort, so you know what to tackle first Defined success measures — concrete metrics so improvement is measurable, not a matter of opinion Constraints and risks captured upfront — data gaps, compliance requirements, tooling limitations, and approval chains identified before they become blockers A delivery roadmap — recommended next steps, dependencies, and sequencing so the path forward is clear and defensible Optional: a scoped pilot — a readytorun 1–2 week proofofvalue engagement to validate the top opportunity immediately How it works Day 1–2 — Assess We map your workflow endtoend, align on what success actually looks like, and surface the constraints that will shape any solution. Day 3 — Prioritise We score every opportunity against impact and delivery effort, cutting through the noise to identify where to start and why. Day 4–5 — Plan We build your delivery roadmap: recommended next steps, dependencies, risk register, and — if you want to move fast — a scoped pilot ready to kick off immediately. Who it's for Teams with repetitive manual processes but no clear picture of where automation would actually help Leaders who need a defensible, evidencebased plan before committing budget to tooling or build work Ops, finance, customer support, project delivery, and engineering organisations under pressure to do more with less Why BlackBoxAI We're engineers, not strategists — which means everything we recommend is grounded in what can actually be built and delivered. The Discovery Sprint is deliberately fixedscope and timeboxed so you get a sharp, usable output in five days, not a rolling engagement that never quite finishes. Fully remote. Fixed price. Clean handover. No retainer required. And if you want to keep going — we can run the pilot ourselves.

View Details

Consulting

Service

AI-Ready KPI Dashboard

AIReady KPI Dashboard — 5 Days You can't improve what you can't measure — and you can't trust what you can't see. Most teams are flying blind. Reporting lives in spreadsheets, metrics mean different things to different people, and by the time the numbers are ready, the moment to act has passed. BlackBoxAI builds you an AIready KPI dashboard in five days: a single, reliable view of performance that baselines where you are today, tracks change over time, and gives you the decisiongrade data you need to prove improvement — and know what to fix next. What you get KPI definitions aligned to your workflow — clear, agreed measures of what "good" looks like, not vanity metrics Baseline and trend tracking — so you can demonstrate progress over time, not just pointintime snapshots Actionable insights built in — the dashboard surfaces what needs attention, not just what happened A data model that scales — structured from day one to support automation and AIassisted decisionmaking down the line KPIs we can deliver quickly Cycle Time — how long work takes endtoend, and where it slows down Throughput — how much your team completes per period, and whether that's improving Error Rate — quality failures, rework, and exceptions that drain capacity and cost Not on this list? We scope to your workflow. These are starting points, not limits. How it works Step 1 — Input You share your workflow, data sources, and the decisions you need the dashboard to support. We ask the right questions to make sure we're measuring what actually matters. Step 2 — Define We agree KPI definitions, calculation logic, and success criteria before anything is built — so there are no surprises at handover. Step 3 — Build Dashboard and data model built with sensible defaults, clear visualisations, and a layout designed for the people who'll use it daily. Step 4 — Validate Sanity checks against real sample data and a walkthrough with stakeholders to confirm the numbers look right and the insights make sense. Step 5 — Handover Full documentation and a live walkthrough so your team can maintain, update, and extend the dashboard without coming back to us. Who it's for Teams running repeatable processes who need proper visibility — not a oneoff report Ops, finance, customer service, project delivery, and engineering functions Anyone currently relying on manual reporting, inconsistent spreadsheets, or gut feel to run daytoday operations Why BlackBoxAI We build dashboards the way engineers build software: define the problem first, prototype fast, validate against real data, and hand over something your team actually owns. The result isn't just a pretty chart — it's a measurement foundation you can build on. Fully remote. Fixed price. Clean handover. No retainer required. And once you can see what's happening — we can help you automate it.

View Details

Consulting

Service

Workflow Automation Prototype

Workflow Automation Prototype — 5 Days Stop losing time to manual, errorprone admin. See it automated — in a week. Every team has at least one workflow that runs on copypaste, chased emails, and spreadsheets held together with hope. You know it should be automated. The problem is finding the time and confidence to actually do it. BlackBoxAI takes one of those workflows and delivers a working automation prototype in five days — built around your real steps, your real data, and your real constraints. Not a proof of concept in isolation. Something that actually runs. What you get A working prototype that demonstrates the endtoend flow against your actual process — not a slide deck or a diagram Basic validation confirming the automation behaves as expected against agreed success criteria A clear handover package — setup notes, a simple runbook, and a live walkthrough so your team can own, run, and extend it without us How it works Step 1 — Input You share the workflow steps and a sample of real data. We ask focused questions to understand the edge cases, exceptions, and integrations that matter. Step 2 — Build We prototype the automation quickly and pragmatically — focused on proving the core flow endtoend, not overengineering it before you've seen it work. Step 3 — Test Basic checks against the success criteria we agreed upfront. We surface any issues and resolve them before handover, not after. Step 4 — Handover Documentation and a live walkthrough with your team. You leave knowing how it works, how to maintain it, and how to extend it when you're ready to scale. Who it's for SMEs and teams who need a fast, lowrisk proof of value before committing to a full build Ops, finance, commercial, HR, and project delivery functions with repeatable, rulesbased processes Anyone currently running workflows through spreadsheets, inboxes, or manual copypaste that could — and should — be automated Why BlackBoxAI We follow a prototypefirst approach for a reason: prototype → validate → scale. You see something working before you invest in a full rollout, which means lower risk, faster buyin, and a much clearer picture of what scaling will actually take. Fully remote. Fixed price. Clean handover. No retainer required. Ready to scale it? We can take the prototype all the way to production.

View Details