Conference talk by Radek Sienkiewicz about running a personal AI agent continuously for months, granting it increasing permissions over time — from reading files to handling email, overnight cron-based automation, memory backups, self-monitoring, and drafting business replies. Covers incremental permission creep, integration with an Obsidian knowledge base, core agent job types (ambient operations, attention filtering, execution), Discord channel integrations, system architecture (LLMs, scripts, memory management), practical challenges (bad memory, brittle automations, noisy nodes), trust boundaries, and the value of giving an agent a personality that disagrees with its owner. # Channel Identity Observations: Radek Sienkiewicz (@velvetshark-com) ## Explicit Self-Description - **Open Claw maintainer identity**: The speaker opens with a direct claim: *"Hey. I'm Radek and one of the Open Claw maintainers."* - **User-to-maintainer journey claim**: He describes a progression — *"the simple user, then power user. Now I'm also a maintainer"* — originating from *"I just encountered some errors and just like submitted first PR."* - **Reluctance to share exact setup**: Explicitly stated: *"I'm not a big fan of sharing like my exact setup. Cuz that exact setup is like very specifically for me."* ## Intended Audience - **Open Claw community / conference attendees**: The talk references an event context — *"You'll have more talks about the insides of the OpenCL"* and *"Please welcome Radek, everybody."* - **Aspiring AI-agent users**: The takeaway framing targets beginners: *"start with one recurring pain, grow trusting incrementally, build the knowledge base."* - **People overwhelmed by agent setup complexity**: He addresses common fears — *"now it broke my computer... completely bricked during the update"* — as problems his approach avoids. ## Claimed Scope of Content - **Practical demonstration over internals**: He positions his content as complementary: *"I want to show how it looks from the other side, from the... user"* perspective, versus talks on "how the sausage is made." - **Personal workflow showcase**: Demonstrates his agent managing email, calendar, Obsidian knowledge base (~3,000 pages), overnight maintenance (backups, indexing, updates), Discord-based channels (inbox, consulting, video research, briefing, YouTube, playground). - **Adjacent content areas implied**: He references client consulting work and YouTube video production — a "Video research" channel *"for researching what's out there to help me with the next episode."* ## Demonstrated Presentation Patterns - **Step-by-step incremental methodology**: A repeated rhetorical structure — *"I see the need. I solve it in a very simple way... then I add more steps."* - **Concrete personal examples as evidence**: Specific anecdotes: Netflix payment failure fixed "within 5 minutes," domain renewal caught proactively, email drafts pre-written. - **Honest limitation-sharing**: Devotes time to failure modes — *"bad memory compounds," "brittle automations... it can break and it probably will break," "noisy nodes."* - **Philosophical framing**: Closes with a "past me / present me / future me" narrative: *"the job is to become friends with the future me... that's the job of the agent."* - **Cultural reference anchoring**: Cites Andrej Karpathy's viral tweet on LLM knowledge bases as validation of his own setup. ## Notes on Evidence Limits - The handle "@velvetshark-com" appears only in the title; no transcript content elaborates on the channel or site itself. - The talk describes *his setup*, not the channel's publishing cadence or content strategy; no explicit channel-level claims beyond the Open Claw maintainer role. # Delivery Observations — Radek Sienkiewicz ## Voice & Tone - **Conversational, talk-style delivery** with frequent filler words and self-corrections: "Uh and like it's almost like literally," "so so this happens like step by step." - **Casual, first-person framing** throughout: "I want to talk what happens like in my life with Open Claw." - **Occasional self-deprecating humor**, e.g., "the past me is just like this completely stupid guy. He does nothing. He's lazy." ## Recurring Vocabulary & Phrases - **"Step by step" / "one step at a time"** — repeated core motif: - "So so this happens like step by step. It wasn't all at the same time" - "because I felt that it's just like one small step at a time" - "So, it's also was just like one one step at a time." - **Hedge word "like"** used constantly as a discourse marker: "just like controls my life," "it's just like an example," "just like one simple workflow." - **Reduplicated words for emphasis**: "to to show," "is is what I just showed you," "that that's where it it it's starting to uh to be super super useful." ## Teaching Habits & Structure - **Enumerated frameworks**: "these are some like five areas or five types of jobs that my agent is doing," followed by first/second itemization (ambient operations, attention filtering, execution supports). - **Concrete examples after abstract claims**: after introducing attention filtering, gives three specific cases — "Netflix payment failure," "domain renewal coming up," and drafted email replies. - **Real artifacts as evidence**: repeatedly points to live material — "this is the real screenshot of my my vault," "these are uh like real screenshots from from today morning." - **Narrative arc from beginner to expert**: "from the first the simple user, then power user. Now I'm also a maintainer." - **Closes with actionable takeaways**: "start with one recurring pain, grow trusting incrementally, build the knowledge base," "optimize for the future you." ## Communication Patterns - **Addresses anticipated audience objection** early: "Or it just like completely bricked during the update. Because I have all these small steps." - **Rhetorical self-questioning** to advance steps: "Okay, so we are there. What's what's next that we can do?" - **Explicit reluctance to prescribe his exact setup**: "I'm not a big fan of sharing like my exact setup. Cuz that exact setup is like very specifically for me." - **Caveats about failure modes** near the end: "what gets harder bad memory compounds," "Brittle automations… it can break and it probably will break." - **Audience-relatable framing of tools** with plain-language definitions: "The scripts are just like, if this happens, do this." ## Uncertainty Notes - **Transcription artifacts** make some patterns uncertain: inconsistent spelling ("Open Claw" vs "OpenCL" vs "Open Cog") and "QMD" likely mis-transcribed (possibly "Grep/OMDb-style search" — cannot verify). - **"Soul MD" and "agent MD" file references** appear consistently and seem to be real project terminology, but exact names cannot be confirmed from the transcript alone. - Repeated **"[snorts]"** stage directions suggest audience/laughter reactions, but the exact nature of these sounds is not verifiable. # Observation Document: "I Gave an AI Agent the Keys to My Life (Here's What Happened)" **Speaker:** Radek Sienkiewicz (@velvetshark-com), self-identified OpenClaw maintainer. ## Central Thesis - Radek gradually gave OpenClaw ("the keys to my life") access to his emails, notes, files, calendars, tools, and operating system, building on memory of everything he does at the computer. - He explicitly stresses this happened **step by step**, not in one leap: > "it wasn't all at the same time... it would be silly to do or even silly to expect that this could even work." ## How He Got There (Stated Progression) - Started with one chat channel: **WhatsApp → Telegram → Discord**. - One simple workflow/task at a time, then added the next step. - Became a maintainer the same way: encountered errors while playing with a workflow, submitted first PR, then a second, joined Discord, and got involved. - His stated method: **see the need → solve it simply → add more steps**. - Claims this incrementalism is why he avoids big breakages: when something breaks, he takes one small step back, fixes it, understands why, and prevents recurrence. ## Knowledge Base (Obsidian Integration) - Gave the agent his Obsidian vault: ~**3,000 pages/notes** of markdown covering work, personal, tasks, projects, research, articles, and an inbox of links. - Search setup: normal search, QMD search for Obsidian, memory for workspace, all interlinked. - **Inbox workflow:** agent analyzes added links (tweet, thread, article, YouTube video), adds tags and context, relates them to existing vault content, and surfaces prior notes: > "yeah, I completely forgot about that." - References a viral **Andrej Karpathy tweet** about LLM knowledge bases as the moment he realized his setup wasn't as simple as he thought. ## Ambient Operations (Overnight, ~3–6 a.m., "4:00 a.m." given as example) - While sleeping, the agent: indexes everything, backs everything up (worst case loses "maybe couple hours of work"), refreshes QMD/memory/Obsidian indexes, prepares morning summary of emails and calendar, and updates OpenClaw itself with surrounding scripts that know what to do, what can break, and how to verify before restarting the gateway. ## Five Job Types (exemplars given; he declines to share his exact setup because it's specific to him) 1. **Ambient operations** – updating, plumbing, background tasks. 2. **Attention filtering** – with full context from Obsidian, it flags important/urgent items. Examples: Netflix payment failure fixed within 5 minutes; domain renewal email he would have missed, agent messaged him on Discord and renewed the domain. 3. **Execution support / draft-synthesize** – reads emails with project context, drafts replies into his draft folder for accept/delete/edit. ## Discord Channel Structure - **General** (everything; specific channels split off when a conversation type recurs), **inbox** (knowledge base), **consulting** (clients, quotes, deadlines, tasks), **video research**, **briefing**, **Instagram**, **YouTube**, **OpenClaw** (maintainer work), **playground** (testing different models/workspaces; promote what works, discard what doesn't). ## System Architecture & Memory - **LLMs** for judgment: understanding email, context, connections. - **Scripts** for deterministic rules: > "if this happens, do this... LLM is even skipped." - Memory optimization: SOUL MD and **critical-rules MD** (placed high in agent MD) because things in agent MD/SOUL MD were sometimes forgotten. - Memory evolved: one memory file → whole memory folder → "dreaming" with memory promotion. - Notes everything is inspectable — done files are editable and readable — which makes this easy in OpenClaw. ## Stated Failure Modes - **Bad memory compounds:** with thousands of nodes/memories, you must actively work on it. - **Brittle automations:** 10-step automations "can break and probably will break" — split into simpler ones or add guardrails. - **Noisy notes:** regularly cleaned/removed. - **Weak boundaries:** SOUL MD and other files must be optimized for your needs. ## Takeaways He Offers the Audience - Start with **one recurring pain**. - **Grow trust incrementally.** - Build the knowledge base; move as much as possible to markdown and make connections. - Inspect the system (easy with OpenClaw). - **"Optimize for the future you."** ## Closing Framing - References his earlier article about **past me / present me / future me**: past me is "completely stupid," future me is "some kind of god creature" — a rationalization he used to avoid doing things. - The job is to "become friends with the future me"; the agent's purpose in his setup is to help the future him so that when he wakes up, as much as possible has already been done by someone other than him. - Qualification: > "that's the whole purpose of this setup at least for me. I don't know, it could be different for you." # Speaker Attribution & Provenance Record ## Video Metadata - **Title:** I Gave an AI Agent the Keys to My Life (Here's What Happened) — Radek Sienkiewicz (@velvetshark-com) - **Video ID:** `01a15ca9-7f1a-74cf-8358-be4c086c8ce8` - **Format:** Conference/talk presentation (live audience, applause present) ## Speakers & Roles ### Primary Speaker: Radek Sienkiewicz - **Role:** Creator / keynote speaker - **Self-identified affiliations:** - One of the **Open Claw maintainers** - User of Open Claw (self-described journey: simple user → power user → maintainer) - **Role trajectory (per speaker):** Started as a user via WhatsApp, migrated to Telegram, then Discord; became a maintainer after submitting PRs and joining Discord. ### Secondary Voice (unnamed) - **Role:** Emcee / introducer - **Evidence:** Opens with "Ready to go? All good, yeah. Please welcome Radek, everybody." followed by audience applause. - **Identity:** Not stated in transcript or metadata. ### Audience - **Role:** Live audience — applause, snorts/laughter reactions noted in transcript. No individual speakers identified. ## Referenced Sources & Third-Party Claims | Source | Type | Claim Context | |---|---|---| | **Andrej Karpathy's tweet** | Social media post (described as viral) | Speaker says it addressed "LLM knowledge bases" and matched his own setup; he read it and recognized his setup in it. Exact tweet content not quoted. | | **Twitter threads, YouTube videos, other users' setups** | Community comparisons | Speaker cites these as benchmarks revealing his setup is more sophisticated than others he encountered. | | **Speaker's own article (a couple years prior)** | Personal blog article | Topic: "past me, present me, and the future me." No title or URL given. | ## Referenced Tools / Platforms (mentioned, not endorsed claims) - **Open Claw / OpenCL / Open Cog** (agent platform — speaker is a maintainer) - **Obsidian** — ~3,000 markdown notes vault - **QMD search**, memory workspaces - **WhatsApp, Telegram, Discord** — chat channels - **Netflix** (payment-failure anecdote), domain renewal, email/calendar tools ## Promotional Segments - **None explicitly flagged.** No sponsor reads, product promotions, or calls to purchase identified in the transcript. The talk promotes the speaker's own workflow and Open Claw project by description, but no commercial promotional segment is marked. ## Availability & Uncertainty Notes - **Comment availability:** Not indicated in supplied source material. - **No timestamps** present in transcript. - **Emcee identity:** Unknown. - **Karpathy tweet:** Referenced but not directly quoted — provenance is speaker paraphrase only. - **Terminology inconsistency:** Transcript alternates between "Open Claw," "OpenCL," "Open Cog" — treated as the same project per speaker context; identity not independently confirmed. \>\> Ready to go? All good, yeah. Please welcome Radek, everybody. \>\> \[applause\] \>\> Hey. I'm Radek and one of the Open Claw maintainers. And uh I want to talk what happens like in my life with Open Claw when I practically gave the keys to my life to to Open Claw. Uh and like it's almost like literally. And uh what that actually means. So so this happens like step by step. It wasn't all at the same time, but it can access my emails, it can access my notes, files, calendars, tools, my operating system. So automations and it builds on top of like memory of everything that I do uh at the computer. So uh it can do anything if with it that uh that is possible to do with the computer. But it didn't all happen in one big like leap. So I install Open Claw and now uh it just like controls my life and does everything for me. \>\> \[snorts\] \>\> Uh that that would be silly to do or like even silly to expect that this could even work. Uh so what happened is that I I tried installing uh just like like everybody does just like with one channel. I think it's at the beginning it was just WhatsApp. Then I migrated to Telegram. Now I'm on Discord. But it was just just WhatsApp, just one ability to do to just like chat. Okay, so we are there. What's what's next that we can do? Let's let's do some like one simple workflow or one very simple task that you can do. Once we are there, let's go to the next step. So, this is how it happened where I am today where I used to think that I have quite a simple setup with my OpenCL and what it does because I never did any big change. But when I encountered different I don't know Twitter threads, YouTube videos, or talking to other people how they have it set up I see that like my setup has everything that they have more on top of that and most also is just like more sophisticated than what what I see out there. Which was really surprising to me because I felt that it's just like one small step at a time. I have a pretty like simple setup, works for me, but uh that that's what I want to to show how that happened and how it looks like today. So, you you already had like a lot of talks about how the sausage is made how we are making it better. You'll have more talks about the insides of the OpenCL. I want to show how it looks from the other side, from the first the simple user, then power user. Now I'm also a maintainer. You don't have to go to the maintainer route, but when I was playing like with one of the workflows, I just encountered some errors and just like submitted first PR, then the second PR, then just looked into Discord, and then you just got involved. Now I'm a maintainer there. So, it's also was just like one one step at a time. Uh So, that that's the step of the These are the steps that uh it usually happens that I see I see the need. Uh I solve it in in a very simple way. Uh and and then I add more steps to it. And this is also why I usually don't have big issues that people have that, "Okay, now it broke my computer." Or it just like completely bricked during the update. Because I have all these small steps that I take. If something breaks, I just like step take one small step back, fix it, see what doesn't work, understand it why it didn't work, uh have a setup that it never happens again, and just like the take one step uh further again. So, \>\> \[snorts\] \>\> uh where it started being more and more helpful and kind of like running my life is when I gave it my knowledge base. So, I had a lot of stuff in my Obsidian, which I built up for years. So, right now I have like about 3,000 pages or notes, markdown files in my Obsidian. And this is everything. This is work stuff, personal stuff, tasks, projects, research, um what else? Articles, um kind of like an inbox of links that I'm just putting there, and it then finds the the connections, uh and puts it in in perspective and in context to to other stuff that I have. So, all of that is now accessible through my open claw with a very good search. I have search and memory. I have like normal search. I have QMD search for for Obsidian. I have different memory for for my workspace. Um and all of that is interlinked. And and and that that's where that magic happens. And when I saw recently And that that's where it hit me that I probably don't have a simple setup. When I recently saw Andrej Karpathy's uh tweet that went viral uh where he says about LLM knowledge bases. I was reading that and it's just like yeah, that's exactly what I have. Like, what's like super uh revolutionary about it. And then I I I understood that, okay, so I got there step by step. It works for me. So, it's probably probably worth I don't know sharing sharing telling more about it, uh showing how it works, showing how you can get to that point as well. \[snorts\] Uh and uh for example, for Obsidian uh Do we Yeah, this \[snorts\] is how the this is the real screenshot of my my vault and all the notes and these are different clusters. Some are probably uh project related like the big clusters. Some of the one off. Uh these are probably more uh kind of like bookmarks. And one of the tasks that I'm doing and that I have is that when I add something to inbox it then takes that link that I add there, looks what's there. It could be a tweet, it could be a thread, it could be an article, it could be a YouTube video, analyzes it, adds tags to it, adds context to it, looks at what's already there on this topic in my vault, how it could be helpful in other areas, and adds connections to it. So, what previously was just like Twitter bookmarks that you bookmark and you never go back to that, now it just adds more context, builds up my knowledge base, and is much more helpful, and even surfacing the things for me when I add a bookmark that okay, so you already had like this and this and this about this subject, and this is how it connects. Maybe you should look at those notes. And very often it's just like yeah, I completely forgot about that. And and that's a good source of of knowledge and of thinking about it. Uh Cuz there was the reason why I'm adding this bookmark. Uh so, that that's where it it it's starting to uh to be super super useful. On top of that, \>\> \[snorts\] \>\> also uh at 4:00 a.m. like 4:00 a.m. is just like uh an example that that I have is happens probably between 3:00 and 6:00, more or less. Um so, this this is what what is happening when I'm sleeping. So, when I'm sleeping, again, my agent does everything so that it runs well. It indexes everything. It backs everything so that worst case, I if I lose something, I lose maybe couple hours of work, of content, of anything else. Refreshes all the indexes for for QMD, for memory, for my Obsidian vault. and I start fresh in the morning with whatever waits for me. Maybe summary of the emails, of the calendar, everything updated, the latest the latest Open Claw version is waiting for me, which also took like step by step. I have some scripts around it, so that it knows what to do and what not to do when updating, what can break, why it breaks, how to verify it before updating or before restarting your gateway, so that it is able to come back online again. So, that that all is also automated, and as as I get up, it's it's already waiting for me fresh and ready for me to start the day. And each Open Claw is Like, I'm not a big fan of sharing like my exact setup. Cuz that exact setup is like very specifically for me, for what I need right now, for what I will need in the near future, for like the errors that I encountered, for issues that I want to be solved, but to give you some ideas, so that we can talk more also about specifics and not just like in general. So, these are some like five areas or five types of jobs that my agent is doing. The first one is is ambient operations. So, so this is what I just showed you. So, it it does all the updating, it does all the plumbing, it does all all the stuff that needs to happen, but I don't need and I don't want to think about. The the second is attention filtering. So, this is also super useful that because it has access to everything, and because it has all the content context actually so it knows that for example when an email comes and uh it's something important or urgent and it knows from Obsidian what's the context and the background behind it uh cuz yeah I I I keep everything in Obsidian all the projects about everything else so it then can proactively tell me that uh I think I have here so like these are like three very specific examples that I had recently that when the system notices that something is important and urgent it just lets me know so like Netflix payment failure for some reason didn't go through uh was fixed within 5 minutes when it happened domain renewal coming up I would probably miss that email uh but uh it picked it up uh gave me gave me a message on my Discord uh renewed my domain uh emails that can already be with enough context given about the project for example it can already uh give like read the email uh understand what's happening understand what's already done within the project and just draft the reply and it's already in in draft uh folder for me to uh accept or or delete or make some changes so so these are some examples of like attention filtering uh execution supports yeah so that's draft synthesize is that uh the the inbox and these are on the right these are the channels that I have in my Discord that more or less relate to these types of jobs so general is where I have everything uh I just start the conversations uh see where it goes and if enough times I have a type a certain type of conversation, I added a specific channel for it. So, the these are uh like real screenshots from from today morning. Uh the inbox is where I just like drop links and it builds the knowledge base for me. Consulting is for for for the clients and every all the backgrounds, it knows all the projects. Uh it's now knows all the quotes, deadlines, tasks, next steps, everything else. Video research is for for YouTube, for researching what's what's out there uh to help me uh with with the next episode. Uh briefing is for morning briefings, Instagram for social posting, YouTube is uh for for creating creating the the videos. Open claw is for maintainer stuff, and there's also one playground channel uh which it changes depending on day, month, or the need. Uh it's for testing. I usually test maybe a different model, maybe a different uh workspace, different way of setting up the the important files like memory and everything else. So, I just play there, see what works. If something worked, uh I promote it. If if it doesn't, I discard it. Uh and all of that works because it's not just uh it's a system that has many moving parts that work well together. So, LLM's is for judgment like understanding the email, understanding the context, making the connections. Then there are all the files, the the tools, the scripts that I have built. The scripts are just like, if this happens, do this. It's done. You don't even need judgment, so LLM is even skipped. Uh and important uh is also to optimize your memory file, your soul soul MD file. Uh I have also critical rules MD uh because even if I had something in agent MD uh or in soul MD, uh it it still managed to to forget something or not do something. Uh with critical rules, having critical rules helps and having it uh mentioned quite high in the agent MD file. Uh so that that's also an improvement. Uh I I went through a few different setups of memory where I had one memory file. Now now that uh I have like the whole memory folder, now you also have dreaming where uh we have like promoting the memories. So this is important to work on these files uh and but it's easy to do in Open Cog because everything is inspectable. These are our done files, editable, you can look at it, you can read it, you can understand it. Uh and it work works well. Uh what gets harder bad memory compounds if the memory is not set up correctly and your vault, your nodes, your memories grow to thousands, you're going to have an issue, so you need to actively work on that. Brittle automations, especially when it's like 10-step automations, uh it can break and it probably will break at some point. So it's again, either split it up into simpler ones or have uh some guardrails uh that are more effective. Uh noisy nodes, uh I'm getting rid of them, uh cleaning um cleaning regularly. And weak boundaries, so so those are all the soul MD and everything else uh that the files that that are important to optimize for your needs. Uh so, what I want you to take away from this is that like do what I did and then at some point you realize, "Yeah, this stuff is awesome and this stuff helps my life." Um start with one recurring pain, grow trusting incrementally, build the knowledge base, uh move everything or like move as much as you can or as you want to markdown files and and start making those connections. Uh inspecting system expectable is is easy for you done for with uh with open claw. Um and optimize for the future you. And this is what I want to close with. So, a couple \[snorts\] years ago I had an article about like the past me, the present me, and the future me. And the past me is just like this completely stupid guy. He does nothing. He's lazy. Uh he doesn't want to do anything. So, now I present me need to do everything for that like past me. And and the future me, the future me is just like some kind of like god creature. It can do anything. Uh that that that creature is like um all powerful. And just like if I don't do something today, it's fine that that other creature will do it for me. So, that that was the the issue and \>\> \[snorts\] \>\> uh the job for me is to to become friends with the future me. To to treat that as a person that I want to help with. And that's the job of the agent. So, I don't need to do as much as I used to because the agent just helps the future me as much as possible so that when I wake up tomorrow, it's like as much as could be done but in someone else other than me is done. So, that's that that's the whole purpose of of this setup at least for me. I don't know, it could be different for you. So, that's what I want to leave you with. Thank you. \>\> \[snorts and applause\]