# How I work. > How a project runs: Discovery, a four-week Build Sprint, then Run & Grow. What I decide, what my AI agents do, and what they never do. APPROACH I decide what gets built, my AI agents carry out the execution, and I check every result before it reaches you. Here is how a project runs, what I do myself, what my agents do, and what they are never allowed to do. ## I make the decisions. My AI agents do the execution. ### My part - What to build, and what to leave out - How it should work, and why - Reviewing every change before it ships - Talking with you and your team ### The agents' part - Writing and testing the code - Checking every page after a change - Drafting migrations, docs, and runbooks - Watching logs, uptime, and dependencies The speed comes from the agents. The quality comes from twenty years of knowing what good looks like. ## What my agents never do - Deploy to production without my review. - Touch customer data outside the task they were given. - Change a price, a DNS record, a payment setting or an access right on their own. - Send anything to you or your customers that I haven't read. HOW I WORK ## A simple way to remove the bottlenecks slowing your team down ### Discovery Fixed price I map the workflow, the tools, and the manual steps behind the bottleneck. You leave with a clear plan, the highest-return opportunities ranked by ROI, and a practical scope you can act on. ### Build Sprint Four weeks In four weeks, I build the first working version of the fix. It is scoped to one high-impact bottleneck, tested with the people who use it, and ready to prove value quickly. ### Run & Grow Flat monthly, after a first project After launch, one flat monthly fee covers everything that comes next: bug fixes, integrations, data quality, new features, and new builds. It all runs through one transparent queue, with updates shipped every week and no new project to scope. See what Run & Grow covers THE ENGAGEMENT ## Start with one high-impact bottleneck. Ship the fix. Measure the result. You work directly with the person building it. I improve the systems you already have, avoid unnecessary complexity, and keep every step tied to hours saved, revenue unlocked, or both. 01 Plan First, I map how the work really happens today: systems, data, hand-offs, and manual effort. Then we choose the one to three outcomes most worth fixing and turn them into a focused backlog. 02 Build Next, I design and build the smallest version that solves the problem properly. You see progress every week, decisions happen in real time, and everything ships into your systems under your ownership. 03 Ship Then I roll it out, support adoption, and measure the change. Each cycle ends with a live improvement, a clear record of what changed, and the next best opportunity already scoped. 4 weeks TO A FIRST WORKING VERSION Weekly IMPROVEMENTS SHIPPED AFTER LAUNCH Direct ACCESS TO THE BUILD, NO ACCOUNT MANAGER ## What the work involves Two kinds of work sit behind every service. ### Screens people can use without training I design the flow first, then build the portal, checkout, or internal tool it runs in, and test it with the people who will use it. For design and front-end I bring in specialists from my network. Everything still passes through my review. - Onboarding, checkout, case handling, and approvals mapped end to end - Clickable prototypes in days, tested with the people who use them - Customer portals that answer the questions support keeps getting - Checkout and payments, including several sellers in one basket - The admin, queue, and reporting screens your team works in all day ### Systems that talk to each other, and agents that do the repetitive work I connect the tools you already use so they share the same data, and put AI agents where they carry real work. Every agent has one job, rules for what it may do alone, and a person on anything ambiguous. - Work that passes from one system to the next without anyone copying it - CRM, billing, support, and shop kept in sync automatically - The same customer record correct everywhere - Agents that read emails and documents, draft replies, and hand back to a person - Every change tried on your real cases before it goes live - An alert on anything that breaks, before your customers notice ## Run & Grow Run & Grow is available after a first project, because I only take on systems I know. Launching a system isn't the end of the work. Forms stop delivering, integrations change, and someone has to keep an eye on security and performance. Meanwhile the next idea is already waiting: a new workflow, a portal, an automation. I offer Run & Grow to clients I've already worked with, as one flat monthly subscription. It covers both: looking after what exists and building what's next, new projects included. I do the work myself, with my own AI agents carrying out the execution, and I bring fixes for problems before they land on your desk. ### One plan. Everything included. A flat monthly fee for support, improvements, and new builds, with the strategic side included. No change requests, no new project for every bigger idea. Pause it when things are quiet, stop it when you don't need it. The only thing you choose is how we communicate. - Unlimited requests, new builds included Send everything that comes up, from a typo to a new system. There's no ticket count, no hour budget, and no separate quote for bigger work. - One task at a time Each task gets finished, checked, and shipped before the next one starts. Bigger work gets split into steps that ship one by one. You set the order. - Context that carries over I know the system, the decisions behind it, and last month's work. You never explain it twice. - Shadowing Copy me on the threads where problems come up. Each week I bring a fix for what I've read, so the same pain doesn't follow you into the next week. - Proactive, not only on request Updates, security, speed, and improvements I plan ahead, on top of what you send. - Your own client portal One login with a username and PIN. Every task on a Kanban board, and your goals and progress on one page, always up to date. - Everything stays yours Code, hosting, domains, and data stay in your accounts, with or without the subscription. - Pause or stop anytime Billed monthly. Pause for up to two months when things are quiet. Cancel by the 15th and it ends with that month. ### Choose how we communicate - Async Send requests by email or through your client portal. I reply within 24 hours. - Live A shared chat channel for a quicker conversation while I'm working. Email and your portal are still available. The monthly fee follows my assessment of what your business needs: the systems involved, how much you want to build, and how fast you need answers. We agree on it in a free consultation. [Book a free call](https://calendar.app.google/QTaGVNQLgWooXE5x8) [Message me on WhatsApp](https://wa.me/4915679819490) My AI assistant answers in seconds. I take over the conversation. ### Partner level Included, not an add-on This is a premium service, so the strategic side comes with it by default. You get someone thinking about where the system should go, not only fixing what broke. - Strategic product guidance Where the system should go next, tied to hours saved and revenue. - Proactive recommendations Improvements I spot in how the system gets used, brought to you with a proposal. - Technical consultation A second opinion on any technical decision, even outside what I built. - Regular strategic check-ins A monthly call to review what shipped and decide what's worth doing next. ### Why a subscription, not hours or projects An hourly freelancer starts from zero every time you call. A project ends at launch, the context leaves with the team, and the next idea needs a new quote. A subscription keeps the context and the momentum from one month to the next. #### Run & Grow - Yes: Knows your system and its history - Yes: Context carried month to month - Yes: Acts before you ask - Yes: Senior on every task - Yes: Strategy included - Yes: Predictable monthly cost - Yes: New builds without a new quote #### Hourly freelancer - No: Knows your system and its history - No: Context carried month to month - No: Acts before you ask - Yes: Senior on every task - No: Strategy included - No: Predictable monthly cost - No: New builds without a new quote #### Project-based agency - Yes: Knows your system and its history - No: Context carried month to month - No: Acts before you ask - No: Senior on every task - No: Strategy included - Yes: Predictable monthly cost - No: New builds without a new quote ### How requests work 1. 01 Send it, or bring me into the conversation Email, your client portal, or chat on Live. You can also copy me on a relevant thread. 2. 02 We set the priority I'll tell you where it sits and when I expect to get to it. Bigger work gets split into steps. 3. 03 I make the change I decide the approach, my AI agents carry it out, and I check the result. One task at a time. 4. 04 We look ahead Each month, we review what changed and decide what's worth doing next. THE PRINCIPLES ## What clients can count on The rules behind every project. ### You own what I build Code, data, and infrastructure live in your accounts from day one. No lock-in, no black box. ### I fix one bottleneck at a time I ship, measure, and only then move to the next constraint. That keeps momentum high and complexity low. ### I use the simplest tool that will hold up If a low-cost no-code tool solves it well, I use it. If the problem needs custom software, I build it. ### Clarity Clear scopes, clear priorities, and clear reasons behind what gets built next. ### Ownership You keep the code, the data, and the infrastructure, so the value stays with your business. ### Pragmatism I choose the fastest approach that will still hold up under real use, instead of overengineering for show. [See the services](https://automatics.dev/#services) --- Markdown version of https://automatics.dev/approach/ Agent guide: https://automatics.dev/llms.txt ยท Sitemap: https://automatics.dev/sitemap.xml