Most teams I meet use AI like a faster search box. A little prompting, a bit of rewriting, back to the spreadsheet. The tools are capable. The gap is training, not software. I fix that, and where the job needs more than the tools give you, I build the system.
ChatGPT, Claude, Gemini, Copilot. If your team already has them, this is where the return is. Context about your business, reusable instructions, connecting it to your inbox and docs, and the habits that turn a lucky afternoon into an actual process.
Hermes as the agent layer, running open-source and open-weight models such as DeepSeek, MiniMax, Qwen and Llama, hosted on your hardware. No per-seat cost, your data stays in the building, and nothing gets switched off by someone else's pricing decision.
When the tools are not enough, the workflow gets built properly. That is where twenty years of running companies comes in: I know where the hours actually go, what your people will genuinely use, and what a process costs to run at volume.
Most people who train you use one or two of these. I use all of them, and I can tell you which is right for the job in front of you rather than selling you the one I happen to prefer.
Hermes and OpenCode running DeepSeek, MiniMax, Qwen, Llama, Mistral and the rest on local hardware or a box I set up for you. Setup, configuration, upgrades, and the training so your team is not frightened of it. For volume work, private data, and anyone who has been burned by a price change.
ChatGPT, Claude, Claude Code, Gemini and Google AI Studio, Grok, and Perplexity Computer. ChatGPT Business and Claude's team plans, structured properly with shared instructions, projects, and the admin side set up so it does not become a shadow IT problem.
Not sure which is right for you? That is the first conversation, and it is free. Twenty minutes on a workflow you are trying to improve, and you leave with a clear answer either way.
These are the real ones from the last few years of small businesses figuring out AI. If yours is not here, it is probably better than the ones that are.
This is the most common question I get. The answer is almost never a new tool. It is context, reusable instructions, and a decision about where the work lives so two people doing the same job stop getting two different results.
More than most people assume, less than vendors would like. Customer names, contracts and pricing need handling; a rewritten product description usually does not. I give you a plain rulebook your team can follow, which is the Your Data, Your Rules guide in short form.
Honest answer per task. Some of it is fully handled today. Some is a good first draft. Some is not worth automating yet, and I will say so rather than build something that quietly underperforms. The Where AI Actually Helps guide covers which is which.
That question gets a straight answer, because getting it wrong causes real damage. In practice AI removes the boring parts of a role and shifts what the role is. That conversation is part of the work, not a footnote to it.
Sometimes it is not, and I will tell you on the first call. Before spending anything, the AI automation ROI calculator sizes the hours, and the Is This Worth Doing guide walks the decision. Both are free.
That is the argument for self-hosting in one line, and it is why open source is my default. The AI vendor interrogation kit asks the questions that tell you how exposed you are.
Everything below is free and public. Reading these first is genuinely the right thing to do, whether or not you ever work with me. If they answer your question, you do not need me.
Six AI advisers argue your next decision from different angles before you spend on it. Useful on its own, and the fastest way to see what these tools are actually capable of.
Ten questions, one honest score. Where you actually are before you start, so the plan is built on the business you have rather than the one in the pitch deck.
Size the hours a workflow costs you in pounds. The number makes the internal conversation much shorter.
Which tasks are worth automating and which are not. Start here. Download the PDF
The shape of a first project that finishes in a month instead of a year. Download the PDF
What can go into these tools and what should not. Download the PDF
So everyone on the team gets the same result from the same prompt. Download the PDF
The questions that show you how exposed you are to a price change. Download the PDF
Setting up the named owner who makes the system work day to day. Download the PDF
Running an agent on your own machine, for when the data cannot leave. Download the PDF
The decision, before you commit money to it. Download the PDF
All twelve guides, including build-the-thing, get-found, get-the-conversation and close-the-deal →
Bring one workflow that eats your team's time. You leave with a clear yes or no on AI for it, what it would take, and one thing you can run yourself today. No pitch, no obligation.
A working session built around what your people actually do, on what you already use or the stack I set up for you. Practical, on your own examples, with the written material to keep afterwards.
The full system: workflow mapped, built, handed over, with your team trained to run it. Fixed price, and you own the system, the data and the roadmap when it is done. See the full service list.
Pick the question on this page that matters most to you, and ask it on the call. If the answer is that you do not need me yet, that is a genuinely useful twenty minutes and I will tell you.
What helps me help you: the workflow, who does it, roughly how many hours a week it takes, and what you have already tried.