Mat Siems
AI Workflow Automation
AI that ships, not slides. Founder-grade AI consulting from London.
Open pagePricing
POMs · Managed Agents · API.
Three ways to work together. First POM free.
Open pageTalk to Mat
LinkedIn is the fast lane.
Bring a problem. 30 minutes. No agenda.
Open pageClaude
Anthropic's frontier model family.
Claude is the model I reach for first. Opus for hard reasoning, Sonnet for production, Haiku for speed — all built around extended thinking, tool use, and the safest pairing for agentic work.
Open pageClaude Code
Agentic coding in your terminal.
The CLI I live in. Claude Code reads your repo, runs your tools, edits your files, and ships PRs — all on your machine, all with your hands on the wheel. The 10x for a senior engineer is real.
Open pageClaude Cowork
Multi-agent cooperative work.
A pattern for getting several Claude instances to plan, build, review, and ship in parallel — with humans steering the meta. Where solo-agent work hits a ceiling, cowork doubles throughput.
Open pageChatGPT
OpenAI's general-purpose assistant.
The household name for AI. Useful as a sparring partner for ideation, longform writing, and image generation — and a known baseline every client already understands.
Open pageCodex
OpenAI's coding agent.
OpenAI's cloud-side coding agent. Strong on long-horizon refactors and back-end work; pairs well with Claude Code for cross-checking on ambitious projects.
Open pageGemini
Google's multimodal model family.
Gemini's strength is multimodal — long video, long audio, and tight integration with the Google stack (Drive, Workspace, NotebookLM). I use it in pipelines that need to ingest the messy real world.
Open pageModels
Opus · Sonnet · Haiku · ChatGPT · OpenRouter.
The model choice matters more than the prompt. I run a routing layer — Opus for hard thinking, Sonnet for production defaults, Haiku for fast/cheap, GPT-5 for second opinions, OpenRouter for everything else.
Open pageAI Agents
Self-driving software that does the work, not just the planning.
I build agents that pick up tickets, edit code, run pipelines, write daily content, and ship deployable output — not chatbots. Pricing is per hour of agent runtime, not per seat.
Open pageCoding
Ship 10x the software with the same team.
Production code, not vibes. I rebuild your delivery pipeline around Claude Code, plan-mode discipline, skills, and subagents — so a senior engineer ships a week of work before lunch.
Open pageSecurity
AI-assisted code review and threat modeling.
Security work is review-heavy and pattern-heavy — exactly where AI agents thrive. I run automated SAST + LLM-augmented review against PRs, with humans signing the high-impact findings.
Open pageEducation
Personalised learning, instrumented teachers.
I build learning experiences where every student gets a tutor and every teacher gets a co-pilot — grounded in source materials, assessed in real time, exportable to any LMS.
Open pageVideos
Daily video at scale, without an editor on payroll.
Pipelines that turn a topic into a finished short — research → script → voiceover → cuts → render — every day, on a schedule. Used internally for VAD and CapCut renamer flows.
Open pageCustomer Support
First-line agents that actually resolve tickets.
Most support AI is window-dressing. I build agents grounded in your real docs, your real ticket history, and your real CRM — that close tickets, not just deflect them.
Open pageFinancial Services
Compliance-aware automation that survives audit.
Regulated environments are AI's hardest mode — audit trails, model-risk frameworks, no hallucinations on numbers. I ship pipelines that operate inside those rails, not around them.
Open pageHealthcare
Clinician co-pilots, not replacements.
AI that drafts the note, summarises the chart, surfaces the guideline — and steps out of the way when the clinician decides. Built with the boring-but-essential parts (audit, traceability, PHI handling).
Open pageNon-Profits
Punch above your weight with managed agents.
Small teams, big missions. Managed Agents handle the donor research, the content calendar, the report drafting — for less than the cost of a part-time hire.
Open pageStartup
Founder velocity — ship the product, not the org chart.
I pair with technical founders to bend the engineering curve — your team of 3 ships like 30. Claude Code, managed agents, and a brutal focus on what moves the metric.
Open pageAppadday
One shipped app per day.
Idea on Monday, prod URL on Monday. Appadday is the rhythm — every weekday I take a concept, scaffold it with Claude Code, wire it to Supabase + Vercel, and ship a real app the same day.
Open pagePAD — Page A Day
365 daily context pages a year.
A daily generated page that fuses personal usage data, project context, industry news, AI updates, and world context into one structured digest. Builds a yearly journal as a side effect.
Open pageLAD — List A Day
A fresh ranked top-100 list every day.
Each morning, /lad picks a never-done-before topic, researches it via NotebookLM, ranks 100 items, generates the markdown + HTML + infographic, and persists to AIDB.DAILY. ~810 lists and counting.
Open pageIAD — Image A Day
One illustrated image per day, on theme.
Daily image generation via Nano Banana / Gemini Imagen — keyed off PAD topics or LAD lists. Builds a year of consistent on-brand visuals.
Open pageAAD — Audio A Day
Daily TTS audio, transcribable + searchable.
Spoken renditions of PAD pages and LAD lists — TTS via Voxtral / ElevenLabs / OpenAI / local, Whisper transcripts for search. Browsable in the AUDIO gallery.
Open pageVAD — Video A Day
Daily short-form video, on autopilot.
Topic → research → script → voiceover → cuts → caption → render → publish. Drives daily YouTube Shorts / TikTok pipelines for client sites.
Open pageBAD — Brief A Day
Client-ready briefs from raw inputs in one POM.
A voice memo, a transcript, a PDF, a one-liner — turn into a structured Project Card, sector classification, technical architecture doc, and proposal pack. Daily, on demand.
Open pageAIOS
The agent operating system.
Where the agents live. AIOS is the runtime — schedulers, queues, model routing, audit logs, secrets, and a UI for humans to step in. Every Mat Siems agent runs on top of it.
Open pageAPPAI
Master control plane for the fleet.
One cockpit for 100 apps. APPAI orchestrates every app in ~/APPS — backlog → build → preview → push → deploy — so scaling the fleet doesn't mean scaling the chaos.
Open pageBOAI
AI-first backoffice.
The boring-but-vital layer — orgs, personas, assets, leads, billing, audit, deploys. Built so every workflow has a place to live and every output gets logged.
Open pageIMAGAI
Image generation pipelines.
Generate, organise, version, and publish images at scale. Nano Banana + Gemini Imagen + DALL·E + Flux behind one interface, with brand-consistent style packs.
Open pageAUDIOAI
Audio generation + transcription.
All audio Mat owns, in one brain — TTS generation, Whisper transcripts, podcast outputs, daily journal recordings, AAD pipelines, master index.
Open pageVIDAI
Video generation + editing.
Veo, Sora, Runway, CapCut — wrapped in one pipeline. From topic to finished short, daily, on a schedule, on every channel.
Open pageScroller
Infinite-scroll feed engine.
The reusable JSON-driven scroller that powers media feeds across the fleet. Randomised feed, modal details, JSON import/export, infinite scroll over any normalised dataset.
Open page