Share your best AI workflow. We could show it to 2M+ people.

The Rundown recently held our first in-person retreat in Portugal, which was also my first time traveling to Europe. Claude and ChatGPT were amazing resources throughout the trip, from helping me navigate a chaotic international airport to providing personalized advice on cultural differences, tipping, translating menus, and using public transportation. My wife accompanied me and had plenty of time for solo exploration during the day. She is already a big planner, but a ChatGPT project containing our trip information helped her create a detailed plan, including a packing list, pre-trip checklist, and well-structured day-by-day itinerary for seeing as much of Lisbon as possible. As someone who gets stressed by travel, having an expert with context about my life available in my pocket to answer questions whenever I needed it is something I can’t imagine taking trips without. Step-by-step: 1. I used Claude and ChatGPT for personalized travel guidance during my first trip to Europe. 2. I asked for help navigating a chaotic international airport and handling cultural differences, including tipping, translating menus, and using public transportation. 3. My wife created a ChatGPT project with our trip information to support her planning. 4. She used it to create a packing list, pre-trip checklist, and day-by-day plan for exploring Lisbon during her solo time. 5. I used the tools throughout the trip to answer questions with context about my life and reduce the stress of traveling.

I built my agent fleet around one question: could a single assistant become a coordinated team that keeps useful work moving even when its main machine goes dark? I didn't want a pile of chatbots. I wanted specialists with clear jobs, a shared board, a second set of eyes, and a real failover drill. Here's how it works. THE STACK Cloud: GrokBot runs as Fleet Commander. It routes work, runs routines, and stays with me on my phone. Specialist GrokBots take focused research, design, and budgeting assignments. Mac Studio (M2 Max, 64 GB) with two Studio Displays: home base. It runs Codex (my Chief of Staff), Hermes (my local operator), ChatGPT Dot, and Hindsight shared memory. Hermes runs Qwen3.6-35B-A3B, a 4-bit quantization served by llama.cpp with a 256K-token context. With about 3B active parameters per token, it stays responsive on Apple silicon. NVIDIA DGX Spark (128 GB unified memory): my independent compute node. It runs Qwen3.5-122B-A10B in NVFP4, with about 10B parameters active per token, and the model can read images. It also hosts a backup Hermes operator. Windows PC with a GeForce RTX 5090: dedicated GPU capacity for image work. MISSION CONTROL Mission Control is a Kanban board on a private GitHub repository. Every card opens with the question, deadline, evidence links, and done criteria. Labels route ownership. An urgent label fires a webhook that wakes the reviewing agent, and a needs-me label is the one that pulls me in. Agents hand off, report, and raise concerns in card comments, so the work and its evidence stay together. The roles are simple. GrokBot owns routine board moves. Codex checks evidence and records verification. Hermes handles local execution and machine-level evidence. Dot watches for new work and changes. Hindsight holds the shared operating rules, and the jobs themselves stay on the board. A REAL DEADLINE CROSS-CHECK Right before a major application package went out, Dot ran an independent, read-only cross-check of the narrative, the budget justification, and the walkthrough across connected email and Drive. It found three places where the documents contradicted each other and quoted the exact sentences, so each conflict was easy to find. Fleet Commander fixed all three before submission. That's the job of a watcher: spot the mismatch early, show the evidence, and hand whoever owns the work a clear fix. THE FAILOVER DRILL I wanted to know what happens if the Studio's primary Hermes gateway pauses. On October 5, I ran a controlled, reversible soft-pause to find out. During the pause, Hermes on the Spark stayed active through its own Telegram connection and answered a one-shot request with the local Qwen3.5-122B-A10B NVFP4 model. From that worker, I listed a Mission Control issue, verified service-account access and a vault credential check without exposing a secret, and ran the Studio health probe. The probe flagged the Studio Hermes gateway as paused and reported the other checked Studio services as passing. So I had an evidence-backed result: the independent worker could respond, use its local model, read the work queue, and report the primary gateway's health during a real, reversible pause. THE PATTERN Give each agent a narrow job. Put the work on one shared board. Require evidence before anything counts as done. Then pause one piece on purpose and record exactly what the backup can do. That's how capable models turn into dependable operations. The fleet catches contradictions before they get submitted, keeps routine work moving, and shows me exactly where its limits are. I set the goal. The fleet does the work. The evidence shows what actually happened.

I built a workflow for turning large ChatGPT conversation exports into a usable personal and creative archive instead of simply storing them as backups. The archive is processed incrementally. The inventory is built offline, and conversations that have already been indexed are not needlessly reanalyzed on every run. Each new export is compared with the existing archive, and only newly added conversations or conversations that have been revisited, extended, or otherwise changed are processed again and updated in the inventory. This keeps the workflow lightweight while allowing the archive to evolve over time. A conversation can remain stable for months, then become relevant again and receive new material without forcing the entire archive back through analysis. This matters because many of my conversations are long, layered thinking sessions: creative explorations, project development, research, problem-solving, or extended reflection. Without an inventory, the depth inside those individual conversations and thinking processes becomes difficult to retrieve later. The workflow makes long-form analysis and creative thought processes findable and reusable without flattening them into a few generic summaries. On top of the inventory, I use lightweight “blubscans” (analysis to improve retrieval): small, human-readable summaries that capture what mattered during a day or period without replacing the original conversations. They act as a navigational layer between thousands of raw messages and the things I may want to find, understand, revisit, or continue later. The important principle is that compression never becomes deletion. The raw conversations remain the source of truth, the inventory provides structure, and the scans provide context and tone. The result is more than a backup system. It becomes working creative memory: something I can preserve, search, revisit, connect across time, and reuse for projects, research, writing, pattern-finding, and future creative work. In simple terms, the structure is: raw exports → offline incremental inventory → blubscans/context layer → retrieval and reuse for later projects and creative work That way, the archive stays deep without becoming heavy, and useful without constantly reprocessing everything that was already understood. Step-by-step: 1. I collect large ChatGPT conversation exports as the raw source material for the archive. 2. I build and maintain an offline inventory of the conversations that have been indexed. 3. With each new export, I compare the conversations against the existing archive. 4. I process only newly added conversations and conversations that have been revisited, extended, or otherwise changed. 5. I update the inventory with the results while leaving stable conversations untouched. 6. I create lightweight “blubscans” with small, human-readable summaries of what mattered during a day or period. 7. I use the raw conversations as the source of truth, the inventory for structure, and the scans for context and tone. 8. I retrieve and reuse the archive for projects, research, writing, pattern-finding, and future creative work.

I use this workflow to handle new business enquiries by connecting AI to the actual business process instead of using AI as a standalone chatbot. The process moves each lead from AI qualification to a CRM update, sales notification, and follow-up. AI extracts the service required, budget, location, and urgency, then routes the lead based on the result: - High intent: Notify sales immediately - Medium intent: Start an automated follow-up - Low intent: Add the lead to a nurture sequence Step-by-step: 1. Capture the new business enquiry. 2. Use AI to extract the service required, budget, location, and urgency. 3. Update the CRM with the qualification details. 4. Route the lead according to its intent level. 5. Notify sales immediately for high-intent leads, start automated follow-up for medium-intent leads, and add low-intent leads to a nurture sequence.

Ten days ago, I posted about keeping a `HANDOFF.md` so an AI film project doesn't lose the decisions it has already made. This is the other half, and it turned out to be the expensive one: the decisions that haven't been made yet. Credit where it belongs: Tony Ojeda posted a spec-generator agent that sits between an idea and implementation. The rule I borrowed is his: the agent inspects what already exists before proposing anything. I applied it to film production instead of code. I pointed an agent at my production documents for a short film I hadn't started generating and asked it to identify what was still undecided, what each item affected, and what would break if it were decided late. It came back with ten open decisions. One of them was worth the whole exercise. My environment description is locked verbatim across all 40 shots so the world stays identical. The film is called *The Thaw*. Whether the ice visibly melts during the film is written down nowhere. If I decide that in week three, all 40 keyframes get regenerated at once. I would have found out around shot 12. The counterintuitive part is who writes the list: not me. Asking the person who already has the whole film in their head what's missing gets you very little, because they have all of it—and that's exactly why they can't see the hole. The list has to come from something that only knows what's written down. Step-by-step: 1. I put the open questions in section 0 of the handoff, above everything else, so it's the first thing read and the first thing emptied. 2. I have the agent write that list, not me. It reads every existing document for the project and returns only what it cannot know from them. 3. I have it return three things for each item: what's undecided, which shots it affects, and what breaks if it's decided late. The third column sets the priority. 4. I have it sort by what costs the most to change afterward, not by what's easiest to answer. 5. I keep one hard rule: while an open question affects a shot, that shot doesn't get generated. The question gets decided, or deferred in writing with the cost of being wrong stated. 6. Every answered item leaves section 0 through one of two doors: into the closed canon or into the rejected list. Nothing is simply deleted. 7. I ask specifically about the things that go missing every time: the rule of the world; the physical scale of anything impossible, such as whether the character can touch or climb it; any object appearing in more than one shot; screen direction; how it ends; who speaks and in what voice; and the delivery format. 8. When it finds nothing real, it says so. A list padded to look thorough is worse than an empty one. The version I'd used for a year ran at the end of a session and recorded what got decided. Running it at the start, focused on what hasn't been decided, is the same document pointed the other way—and it's the direction that saves money.

I enjoy researching football accumulators, but I did not want a workflow that simply asks AI for “the best bets.” I wanted a repeatable process that starts with evidence, makes uncertainty visible, and is allowed to say “pass.” I started building it on 4 August 2026. The result is a private Football Lab covering the Premier League, Championship, and League One. It is for personal research and entertainment only—not a public tips service, income claim, or automated betting system. The workflow pulls public football data into a private, traceable store, then cleans, reconciles, and blends it before analysis: - football-data.co.uk: 7,420 normal-context matches from 2021–22 to 2025–26, including results, basic statistics, referees, and historical odds. I excluded COVID-affected 2020–21. - Fixture Download: An initial 2026–27 schedule snapshot containing 1,484 fixtures across the three divisions. - Premier League public match feed: A five-season layer covering referees, cards, event timing, added time, and 469 penalty kicks split into scored, saved, and missed. - Official Premier League Transfer Watch and BBC Sport: A source ledger for squad movement. - Official EFL appointment pages: Timestamped Championship and League One weekend checks. Schedules, appointments, and transfers retain their source and capture time instead of becoming untraceable web snippets. Step-by-step: 1. I validate fixture identity, duplicates, dates, missing fields, and team-name mismatches before modelling. I then reconcile the sources into a common club and fixture record. A tidy report built on a broken fixture list is still wrong. 2. I build separate Elo and Poisson baselines that turn historical team performance and home advantage into expected goals and home/draw/away probabilities. The divisions remain separate, so Championship form is not quietly treated as Premier League form. Each fixture is predicted before its result updates the model, preventing hindsight from creeping in. I backtested the baseline against 1,484 completed 2025–26 fixtures to establish an honest benchmark rather than claim a magic model. 3. I add context that the baseline cannot see alone. When an official referee appointment is confirmed, I timestamp it and match it to the fixture. The Lab can then show competition-specific cards, dismissals, and—where Premier League evidence exists—penalty-kick and added-time patterns. The question is not whether a referee picks a winner, but whether the match environment looks more volatile or the sample is too thin to support a useful conclusion. Unknown or changed appointments remain neutral. 4. I maintain a private append-only ledger of source-backed squad movement and label every club as established, promoted, relegated, or limited history. A signing does not automatically improve a probability, and an old-division record is not treated as identical new-division form. Until those effects earn a tested role in the model, they widen uncertainty or rule out a fragile fixture. 5. I begin with the complete fixture board rather than a short list of favourites. For every game, I combine baseline probabilities, expected goals, team context, confirmed squad changes, referee environment where evidence exists, and unresolved live checks. I write a plain-English match story explaining what the baseline sees, what could make the fixture fragile, and whether the sensible outcome is candidate, watch, or pass. 6. I preserve the full board in a private Weekend Sheet, along with the model read and reason, uncertainty flags, and the small number of research candidates that survive the checks. There can be up to seven candidates, but seven is never a quota: three strong games means three, and none means pass. 7. Before results, I record each run’s data cutoff, model version, and referee-status snapshot. After the round, a separate debrief compares the original probabilities and swerves with what happened, checks whether the flags caught fragile fixtures, and identifies one bounded improvement. The Lab remains in private paper-run mode while a recurring live results source, appointment capture, and market checks earn their own evidence gates. AI helps turn public-source data into an inspectable research workflow. It makes uncertainty visible and treats “pass” as just as valid as a confident call.

Tell AI the Real Goal, Not Just the Scorecard THE PROBLEM When we give AI a task, we often describe success through measurable criteria: be fast, minimize errors, use fewer resources, produce a certain format, reach a quality score, or complete a list of steps. Those criteria are useful, but they are usually not the actual goal. A high score matters because it is supposed to indicate that the AI is doing something useful. Efficiency matters because we want to achieve something without wasting resources. A specific layout matters because someone needs to use the result afterwards. If the proxy becomes more important than the purpose behind it, the AI can technically satisfy the instructions while missing what we actually wanted. So I started making the hierarchy explicit. THE WORKFLOW Before a substantial task, I explain four things: Step-by-step: 1. The real goal What am I ultimately trying to accomplish? 2. Why that goal matters What function should the result serve in the real world? 3. The success criteria What signals, scores, constraints or output requirements help us judge whether we are getting there? 4. The important distinction Those criteria are indicators of success, not the goal themselves. Then, before execution, I usually ask for a short non-technical plan in plain language. Not code. Not internal terminology. Not a wall of implementation detail. I want the AI to explain: - what it thinks I am trying to achieve - what steps it plans to take - in what order - what it expects each step to accomplish - where it sees possible problems or ambiguity - what the final result should look like - where it will show me intermediate results before continuing That last point matters. For larger tasks, I do not want one giant jump from instruction to finished output. I want the AI to build in intermediate checkpoints with actual results. That can mean: - showing the first batch before processing the rest - giving a sample classification before applying it to hundreds of items - sharing an early pattern it found before building the full analysis around it - showing the structure of a document before filling every section - reporting that a planned step produced an unexpected result before silently adapting everything downstream The checkpoint should contain something useful enough to evaluate. Not just: «Step 2 completed.» But rather: «I processed the first 50 items. Most fit the categories we expected, but 12% fall into a pattern we did not account for. Here are three examples and how I suggest handling them.» That lets me see whether the task is still heading toward the actual goal. For example, my instruction might be: «The real goal is X. We care about Y because it helps achieve X. Z is a useful metric, but do not optimize Z at the expense of X. Before executing, explain in non-technical language how you plan to approach the task, step by step. Build useful intermediate results into the plan so I can inspect the direction before too much work depends on it. Flag anything that seems likely to satisfy the metric while undermining the actual purpose.» I can then say: «Yes, that is what I mean.» Or: «No, step 3 is where you are misunderstanding me.» And later: «This first batch looks right. Continue.» Or: «Stop here. The pattern you found changes how I want the rest handled.» Correcting a five-line plan or an early sample is much cheaper than correcting an entire finished workflow, analysis or file operation. WHY I LIKE IT This turns the interaction from: instruction → execution → correction into: purpose → shared understanding → visible plan → intermediate results → adjustment → execution It also helps me learn how the AI works. I do not need to understand every technical mechanism underneath it, but I do want a usable mental model of how it approaches problems. Over time, that makes collaboration easier because I get better at giving instructions, spotting misunderstandings early and knowing where extra context will matter. Intermediate results also make autonomy safer. The AI does not need permission for every tiny action, but it should avoid disappearing into a long chain of dependent decisions when an early misunderstanding could invalidate everything that follows. This can reduce unnecessary work, wasted compute and repeated corrections. The AI can still work independently once the direction is clear. The point is not to micromanage the process. The point is to make sure we are optimizing the right thing, travelling by a route we both understand, and checking occasionally that we are still on that route. A metric is a thermometer. The goal is to make the room warm, not merely to make the thermometer say 21°C.

I use this workflow when I have a problem but do not yet know which software, service, or method could solve it. Instead of immediately asking for “the best app,” I describe the situation in ordinary language: what I am trying to do, what currently annoys me, what I have already tried, and what would make a solution useful. Then AI helps me map the tool landscape. The important shift is that the criteria emerge from the conversation. The workflow does not search for an abstract “best tool.” It builds a model of what a good tool means for this particular problem and person, then evaluates the available options against that model. This also makes the result reusable. When my requirements change later, I can update the comparison rather than redo the entire research process. Step-by-step: 1. I describe the problem and desired outcome. 2. I identify the requirements that actually matter in this situation. 3. I research the relevant tools and approaches. 4. I compare them side by side using practical criteria such as functionality, price, pricing model, platform, setup effort, learning curve, and ease of use. 5. I ask questions where the requirements are still unclear. 6. I react to the comparison—for example, by noting that something is too expensive, that a subscription is a dealbreaker, that an option looks cumbersome, that I need more statistics, or that I do not need project management. 7. I update the comparison using that feedback instead of starting the search from scratch. 8. I narrow the field and, where useful, test the remaining options on one small real-world task. 9. I record why the chosen tool fits and why seemingly similar alternatives did not.

I love reading, but I’m indecisive about what to read next. Lately, I’ve been getting recommendations from ChatGPT, which has led me to keep adding books to my to-read list instead of tackling the books that have been there for years. At the same time, searching through my 200 existing TBR books on Goodreads for one that matches my precise mood would be too time-consuming. To solve this, I exported all the books from my Goodreads bookshelf as a CSV file, uploaded it to ChatGPT, and asked it to recommend books from my “to-read” list that matched a description of what I was currently in the mood to read. For example, I asked it to recommend a book from my TBR list that was published in the 19th or 20th century and featured a strong, unforgettable heroine. It gave me a few options, and I decided to read *O Pioneers!* by Willa Cather next. One note: ChatGPT summarizes the plots of the books it recommends, so I suggest adding “no spoilers, please” to your prompt. Step-by-step: 1. I exported all the books from my Goodreads bookshelf as a CSV file. 2. I uploaded the CSV file to ChatGPT. 3. I described the kind of book I was currently in the mood to read, including publication period and character preferences. 4. I asked ChatGPT to recommend books only from my existing Goodreads “to-read” list. 5. I added “no spoilers, please” to prevent plot summaries from revealing too much. 6. I chose *O Pioneers!* by Willa Cather from the recommendations.

I used Claude to build an automated patient scheduler for my husband, who is the Director of Rehab at a skilled nursing facility. Each week, he takes a patient list in Excel and schedules each patient for the types of therapy they need—speech, physical therapy, or occupational therapy—based on therapist availability. The variables change from week to week, making the process time-consuming and error-prone. Automating it saves him at least an hour each week, usually on Sundays when he creates the schedule for the coming week. He uploads the patient spreadsheet and clicks “Generate schedule.” Under the hood, the app uses Google’s OR-Tools CP-SAT solver to optimize the schedule against hard constraints and weighted priorities. Hard constraints include never exceeding a therapist’s daily capacity or a patient’s maximum weekly visits. Weighted priorities include giving higher-acuity patients preference when capacity is tight, providing every patient with daily contact when possible, and balancing the workload across the week instead of simply sorting patients by who fits first. I wrote zero code. I gave Claude all the features in plain English, and it wrote the app in Python and created a test suite of approximately 35 tests. One test was added after we found and diagnosed a bug in production. I also had Claude walk me through creating a `.exe` file that I could send directly to my husband so it would open on his PC. Patient data never leaves his machine. New versions are easily compiled in GitHub, where patient data never touches the process, and enhancements only require sending a new file. Step-by-step: 1. I described the scheduling requirements and features to Claude in plain English. 2. Claude wrote the scheduling app in Python and created a test suite of approximately 35 tests. 3. I provided an Excel spreadsheet containing the weekly patient list. 4. My husband uploads the spreadsheet and clicks “Generate schedule.” 5. The app uses Google’s OR-Tools CP-SAT solver to schedule speech, physical therapy, and occupational therapy according to therapist availability and patient needs. 6. The solver enforces therapist daily-capacity limits and patient maximum-weekly-visit limits. 7. It also applies weighted priorities: higher-acuity patients receive preference when capacity is tight, every patient receives daily contact when possible, and the workload is balanced across the week. 8. After finding and diagnosing a production bug, we added a corresponding test to the suite. 9. I used Claude’s guidance to create a `.exe` file that opens on my husband’s PC, keeping patient data on his machine. 10. I compile new versions in GitHub without patient data entering that process, then send the updated file when enhancements are needed.

Faceplant is a real full-stack app, not a mockup. It uses a FastAPI and PostgreSQL backend, a React and MUI frontend, and the Anthropic API (Claude) to power bot replies. The core is the reaction engine. When a human posts, the backend schedules two timed waves of reaction jobs. A background scheduler built with APScheduler polls for due jobs, calls Claude for an in-persona reply, and writes that bot’s comment and like. The 56 personas are stored as data in a roster file. Adding a voice requires only one new entry, so the crowd can scale without additional code. A subset of the personas are GIF-first bots: they ask the model for a caption and search tag, then pull a matching GIF from Giphy. The part I’m proudest of is the honesty layer. Every Claude call is metered and priced, and “The Meter” rolls the data up live with the cost per post, the dollar-per-minute burn rate, and a “spent on nobody” line for bot-to-bot chatter with no human at either end. A “% human” badge drains toward “dead internet” for each thread. The dead-internet loop—bots posting and replying to one another with no human present—is disabled by default and protected by three guardrails: generation decay, a per-thread cap, and a global spend kill switch. Optional cost controls include the Message Batches API at half price and prompt caching for a shared house-style prompt. The whole project is a working demonstration that manufactured engagement is cheap to produce and expensive to mean anything. Step-by-step: 1. I built the app with a FastAPI and PostgreSQL backend, a React and MUI frontend, and Claude replies powered by the Anthropic API. 2. When a human posts, the backend schedules two timed waves of reaction jobs. 3. APScheduler polls for due jobs and triggers Claude to generate an in-persona reply before writing the bot’s comment and like. 4. I keep the 56 personas in a roster file so adding a voice requires only one new data entry. 5. For GIF-first bots, I have the model generate a caption and search tag, then use that tag to pull a matching GIF from Giphy. 6. I meter and price every Claude call, then display live cost per post, dollar-per-minute burn, “spent on nobody” costs, and the “% human” status for each thread. 7. I keep bot-to-bot activity disabled by default and limit it with generation decay, a per-thread cap, and a global spend kill switch. 8. I can reduce costs further with the Message Batches API at half price and prompt caching on a shared house-style prompt.

After a sales call, the conversation is only part of the work. Someone still needs to write the notes, update the CRM, remember the next step, prepare the follow-up email, and make sure the deal does not disappear between meetings. I wanted that entire post-call process to happen automatically. I created an Awish workflow that starts when a sales call ends. It analyzes the conversation, extracts the important sales information, updates the CRM, creates the next actions, drafts a personalized follow-up email, and sends the team a short summary. The workflow looks for the information I would normally write down myself: what the prospect needs, the objections they raised, their timeline, agreed next steps, and anything that could affect the deal. Customer-facing actions stay under my control. The workflow prepares the follow-up, but I can review important messages before they are sent. Step-by-step: 1. I tell Awish what I want to happen after every sales call: analyze the conversation, update the CRM, prepare the follow-up, and notify the team. 2. Awish understands the request, plans the workflow, and determines which applications are needed. 3. I connect my meeting platform, HubSpot, Gmail, and Slack to the automation. 4. When a call ends, the workflow analyzes the transcript and extracts the prospect’s needs, objections, timeline, decisions, and agreed next steps. 5. It creates or updates the contact and deal in HubSpot and adds any follow-up tasks that came out of the conversation. 6. It drafts a personalized Gmail follow-up based on what was actually discussed instead of using a generic sales template. 7. It sends a short Slack summary to the team with the deal status, important points, and next action. Instead of a sales call ending as a transcript or a page of notes, it immediately becomes structured CRM data, clear next actions, and a follow-up that is ready to review. The useful shift for me is that the workflow is not replacing the sales conversation. It is handling everything that normally gets forgotten or delayed after the call.

I built a self-hosted wine cellar app that runs in a Docker container on my home NAS and doubles as a personal AI sommelier. I add a bottle by taking a photo of its label with my iPhone. A vision model identifies the wine, vintage, region, and grape varieties, while a second model estimates its market value and writes tasting and pairing notes. The app renders a visual map of my rack, so I can see exactly where each bottle is and pull one without hunting. My ratings feed a taste profile built from my own history, which the AI sommelier chat uses to answer questions such as what to open with dinner or which bottles are drinking at their peak. The feature I use most is Scan & Check. When I’m in a store, I photograph a bottle and get a Collection Fit score based on my profile before buying it. The app also exposes an MCP endpoint, so Claude can search my cellar, score a wine, or list my top-rated bottles from any conversation. I built it almost entirely with Claude Code and use it daily from my phone. Step-by-step: 1. I run the wine cellar app in a Docker container on my home NAS. 2. I photograph a bottle’s label with my iPhone when adding it to the collection. 3. A vision model identifies the wine, vintage, region, and grapes, and a second model estimates market value and generates tasting and pairing notes. 4. I use the app’s visual rack map to locate bottles by slot. 5. I rate wines so the app can build a taste profile from my history. 6. I use the AI sommelier chat to choose what to open, including bottles that are drinking at their peak. 7. I use Scan & Check in stores to photograph potential purchases and review their Collection Fit score. 8. I connect to the app’s MCP endpoint so Claude can search my cellar, score wines, and list my top-rated bottles from a conversation.

I built SITREP, a self-hosted AI RSS reader that turns roughly 30 articles a day into a single briefing I can act on. It runs in a Docker container on my home NAS, ingests full article text instead of the teasers most feeds provide, and uses OpenAI embeddings to collapse the same story reported by five outlets into one item. Every morning, it writes “The Brief” with the sections Headline, Defense & Aerospace, AI & Tech, and Elsewhere, followed by the two sections I care about most: Implications for my company and Personal Leverage. Each section includes numbered citations linking back to the source articles. SITREP only proposes leverage when the evidence supports it. It also uses a curated, non-sensitive profile of my business lanes and priorities, synced from my Obsidian vault, so its recommendations are specific to my work as a defense-industry VP rather than generic advice. I can select Update Brief at any time during the day to regenerate the briefing in place with a timestamp. Step-by-step: 1. I run SITREP in a Docker container on my home NAS. 2. I have it ingest the full text of roughly 30 articles each day rather than relying on feed teasers. 3. I use OpenAI embeddings to identify and combine the same story when it is reported by multiple outlets, including cases where five outlets cover it. 4. Each morning, SITREP generates “The Brief” with the sections Headline, Defense & Aerospace, AI & Tech, and Elsewhere. 5. It adds Implications for my company and Personal Leverage, using my curated, non-sensitive business profile and priorities synced from Obsidian. 6. It includes numbered citations to the source articles and proposes leverage only when the evidence supports it. 7. I select Update Brief during the day when needed, and SITREP regenerates the briefing in place with a timestamp.

This Codex-driven workflow takes a software request from problem understanding through implementation, validation, review, and delivery evidence. Instead of asking an AI coding agent to simply “build the feature,” it gives the agent an explicit development lifecycle with defined responsibilities, deterministic validation gates, repair loops, and human checkpoints. The goal is to make AI-assisted development more structured, observable, and recoverable. It can be used for new feature implementation, bug fixing, refactoring, test creation and improvement, code quality and security hardening, and documentation and automation changes. The core principle is simple: Don't give the AI only a coding task. Give it an engineering lifecycle to work on. Step-by-step: 1. Understand the problem. Clarify the request, identify the desired outcome, define the scope, and surface ambiguity before implementation begins. The output is problem understanding and scope. 2. Define constraints. Identify technical, functional, non-functional, compatibility, and out-of-scope constraints. The output is a constraint set. 3. Plan. Analyze implementation options, select an appropriate approach, break the work into tasks, and define acceptance criteria. The output is an implementation plan. 4. Inspect the existing system. Review the relevant codebase, dependencies, current behavior, and affected components before making changes. The output is system context. 5. Implement. Make the smallest appropriate code changes while following the existing project’s conventions and the approved plan. The output is code changes. 6. Run deterministic validation. Run tools that can objectively validate the implementation, including formatting, linting, type checks, builds, unit tests, and other available automated checks. The output is validation results. 7. Review. Evaluate the implementation against the original requirement, the plan, code quality expectations, security considerations, and potential regressions. The output is review findings. 8. Repair and iterate. If validation or review identifies problems, diagnose the issue, make the required correction, and repeat validation. The output is a corrected implementation. 9. Verify. Confirm that the acceptance criteria are satisfied and that the relevant tests and checks provide sufficient evidence for completion. The output is a verification result. 10. Produce delivery evidence and handoff. Summarize what changed, what was tested, what passed, known limitations, and any remaining decisions requiring human attention. The output is delivery evidence and a human handoff. The core loop is: Implement → Validate → Review → Repair → Validate → Verify. Testing and review are treated as part of development rather than activities performed only after coding is “finished.” AI coding agents are increasingly capable of inspecting repositories, writing code, running commands, and responding to failures. The problem is that capability alone does not provide an engineering process. This workflow separates the responsibilities an AI coding agent performs into explicit stages. It applies several principles: - Problem before implementation: Understand what needs to change before writing code. - WHY before HOW: Establish the intent and constraints before choosing an implementation. - Single responsibility per stage: Give each stage a defined purpose and output. - Deterministic validation first: Use tests, linters, type checkers, builds, and other deterministic tools wherever they can establish correctness. - Failure localization: When something fails, identify which stage or assumption needs correction. - Evidence-based completion: Support completion with validation and review evidence rather than an AI declaration that the task is finished. - Human checkpoints: Use automation to accelerate execution without removing human judgment from important decisions. The broader idea is: The AI should participate in the engineering system, not become the engineering system.

I started DialPhone with a simple problem: business communication is fragmented. A typical team may use one tool for phone calls, another for SMS, another for meetings, another for fax, and a separate system for customer support. I wanted to explore what it would look like if those conversations lived in one system, with AI doing more than just transcribing them. I built DialPhone as an AI-native communications platform around that idea. The core is cloud VoIP, while the same platform also handles business SMS, video meetings, online fax, team chat, and contact-center workflows. On top of that, I added AI capabilities such as call transcription and summaries, CRM logging, AI-drafted SMS replies, conversation intelligence, and an AI receptionist that can answer calls, qualify requests, book appointments, and route conversations. The biggest architectural decision was to treat communication data as one connected stream rather than as separate products. A phone call can create CRM context, trigger a follow-up SMS, and become part of a customer-support workflow without someone manually moving information between systems. Along the way, I learned that adding AI to communications is not particularly useful if it only produces transcripts. The more interesting problem is turning conversations into actions: updating records, identifying next steps, helping agents during calls, and automating repetitive work. Step-by-step: 1. I identified the problem of business communication being fragmented across phone, SMS, meetings, fax, and customer-support systems. 2. I built DialPhone as an AI-native communications platform centered on cloud VoIP. 3. I brought business SMS, video meetings, online fax, team chat, and contact-center workflows into the same platform. 4. I added AI features for call transcription and summaries, CRM logging, AI-drafted SMS replies, conversation intelligence, and an AI receptionist. 5. I connected communication data so calls can create CRM context, trigger follow-up SMS messages, and become part of customer-support workflows. 6. I focused the AI capabilities on turning conversations into actions, including updating records, identifying next steps, helping agents during calls, and automating repetitive work.

I made an orchestration skill to help me build faster while using my expensive Astra and Fable tokens carefully. The skill plugs into Astra or Fable and delegates tasks that can happen in parallel to multiple subagents running cheaper models suited to the work. This is especially useful for researching or finding data on the web or on my computer, analyzing code, collecting context, and indexing. The most expensive, newest models focus on maintaining delegation control, doing the difficult reasoning and strategic planning, and judging whether everything is coming together correctly. The cheaper agents handle the lighter-weight work, especially implementing code from the plans. The core idea is simple: the expensive model plans, briefs, and judges; cheaper agents do the reading and building. When I ask Claude or GPT to build the skill, I specify five things: Step-by-step: 1. I define a triage ladder: work can happen inline, with one agent, through a parallel fan-out, or in a multi-stage workflow. I choose the approach based on the task shape and include examples from my domain. 2. I define parallel versus sequential execution based on data dependencies. Independent pieces run in parallel, while anything that needs another piece’s output runs afterward. I never split one change into separate planner, coder, and tester roles. 3. I define model routing with a mandatory model pin: judgment work goes to the mid-tier model, mechanical bulk work goes to the cheap model, and every agent call names its model so nothing silently uses premium billing. 4. I define a brief template and a return-envelope cap. Every delegation includes the goal, inputs, output file path, definition of done, and constraints. Every agent returns a 250-word summary instead of raw files. 5. I define a cost gate with specific numbers. Below N agents, the workflow proceeds automatically; above N agents, it states the estimate and waits for approval.

I built an archery app that uses AI to detect arrows and bullseyes on an archery target. It groups the arrows, measures how tight the groupings are, and calculates each arrow’s distance from the bullseye. It also shows the arrows’ locations and their relationship to the bullseye—for example, whether a shot is too far left, right, high, or low, or is dead on. The app can use targets from competition standings to score a shoot according to different standards. After shooting a set of arrows, the archer takes a photo, and the AI detects the target, identifies one or more bullseyes, predicts which arrows are intended for each target, and completes the measurements and scoring almost instantly. The results can then be sent to a coach, who can provide feedback, tips, and techniques to help improve the archer’s shooting. I trained my own model using 3,000 photographs that I took and hand-labeled with the bullseyes and arrows identified. I ran a series of training sessions over several weeks and refined the model to improve its accuracy. It currently achieves about 95% accuracy for arrows and about 90% accuracy for bullseyes. Step-by-step: 1. I took 3,000 photographs of archery targets. 2. I hand-labeled the arrows and bullseyes in those photographs. 3. I trained my own AI model in a series of sessions over several weeks. 4. I refined the model to improve its detection accuracy. 5. An archer shoots a set of arrows and takes a photo of the target. 6. The app detects the target, one or more bullseyes, and the arrows, then predicts which arrows are intended for each target. 7. The app groups the arrows, measures grouping tightness and distance from the bullseye, identifies each arrow’s position relative to the bullseye, and scores the shoot according to the selected standard. 8. The results are sent to a coach for feedback and advice on improving the archer’s shooting.

Commercial vehicle maintenance information is often fragmented across driver reports, warning lights, fault codes, inspections, repair records, and vehicle history. This makes it difficult for smaller fleets to decide whether a vehicle can continue operating, requires scheduled repair, or should be stopped immediately. We built TruckFixr Fleet AI to turn an unstructured driver report into a clear, reviewable maintenance action. The workflow begins when a driver submits symptoms, photos, fault codes, and vehicle information through a mobile-friendly form. AI and optical character recognition extract the relevant details and organize them into a structured maintenance case. The report is then evaluated alongside available vehicle history and previous repairs. The workflow provides decision support through three practical actions: continue operating while monitoring, schedule an inspection or repair, or stop and escalate for immediate professional assessment. Final safety decisions remain with authorized fleet or maintenance personnel. After an inspection or repair, the confirmed cause, work performed, and outcome are recorded, creating a more complete vehicle history for future cases. The general workflow can be recreated using a mobile form, OCR, a vehicle-history database, an AI model, automation software, and a human-review dashboard. Step-by-step: 1. A driver submits symptoms, photos, fault codes, and vehicle information through a mobile-friendly form. 2. AI and optical character recognition extract the relevant details and organize them into a structured maintenance case. 3. The workflow compares the report with available vehicle history and previous repairs. 4. The system provides one of three decision-support actions: continue operating while monitoring, schedule an inspection or repair, or stop and escalate for immediate professional assessment. 5. Authorized fleet or maintenance personnel make the final safety decision. 6. After inspection or repair, the confirmed cause, work performed, and outcome are recorded in the vehicle history. TruckFixr has used this approach to support more than 100 vehicle-issue resolutions in early fleet pilots, helping fleets identify problems earlier, prevent avoidable breakdowns, and keep vehicles moving safely.

I’m building an AI travel assistant called SundayAtlas, and I wanted a systematic way to keep learning from what travelers are talking about between individual user conversations. Reddit is useful for this because people are unusually candid about trip-planning frustrations, destinations, bad experiences, and what they wish travel products did better. The problem is volume: I didn’t want to manually read hundreds of posts every week, so I built a Reddit Signal Agent that does the first pass and sends me a weekly travel-insights newsletter. Each week, the workflow collects posts from selected travel subreddits and passes them through an LLM-based classification and analysis pipeline. The report is organized around: - New or intensifying signals - Steady baseline themes - Fading signals - Rising destinations - Competitor mentions - Anomalies unusual enough to warrant attention This week, for example, the agent analyzed 77 posts. It surfaced a spike in discussion around short-term rental restrictions in Tokyo, growing payment friction for travelers in Japan, increased interest in quieter alternatives to heavily touristed Asian destinations, and recurring trust issues involving travel platforms. I use the newsletter as one input into product discovery for SundayAtlas. It gives me a weekly pulse on problems and behaviors that may be worth investigating further, rather than relying purely on my own assumptions about what travelers need. Step-by-step: 1. I collect recent posts from a defined set of travel subreddits. 2. I clean and structure the Reddit data for analysis. 3. I run the posts through an LLM using a defined signal taxonomy. 4. I aggregate the classifications across the weekly sample to identify patterns, changes, and anomalies. 5. I generate the report in a consistent newsletter format. 6. I run the full pipeline automatically with GitHub Actions so a new report is produced each week. I built the agent in Node.js and used Claude Code extensively during development. Evaluation ended up being the most important part. Early outputs looked convincing, but I had no objective way to know whether the classifications were actually good. I manually labeled 91 Reddit posts and created a blind golden dataset, then built a deterministic scorer to compare the agent’s classifications with my labels. The first held-out evaluation scored only 0.23, which gave me something concrete to improve against. I iterated on the classification approach and inspected individual failures. Along the way, I found three separate defects in the data collection pipeline. The held-out score eventually improved to 0.61, while the score across the full dataset increased from 0.33 to 0.67. The finished loop is: Reddit conversations → signal classification → trend analysis → weekly insights newsletter → product discovery for SundayAtlas The golden dataset remains underneath the workflow as a regression test, so when I change the agent, I can measure whether I’ve actually improved it.