{"schemaVersion":"2026-07-02","title":"Field Experience from the Edge","description":"How builders are adapting AI in real work","sessionCount":14,"generatedAt":"2026-07-11T16:21:34.170Z","entrypoints":{"discovery":"/.well-known/agent-context.json","context":"/api/agent-context","manifest":"/api/agent-context/manifest","openapi":"/api/agent-context/openapi.json","llmsTxt":"/llms.txt"},"instructions":{"audience":"Autonomous agents, research copilots, and retrieval pipelines consuming the Fireside Field Notes module.","recommendedUse":"Load the manifest first for a lightweight map, then load the full context when answering questions about guests, themes, links, or the journey.","traversal":"Use journey.segments as the canonical reading order. Follow hero segments for guest/session context and cross-cut segments for synthesis.","followThreadsBy":["guestName","eventId","slug","tags","shared links","Portal posts","cross-cut question id"],"citationGuidance":"Prefer cited URLs from links.posts, links.shared, media.fullInterviewURL, and entrypoints. Distinguish interview-derived summaries from external links."},"indexes":{"guests":[{"eventId":49,"guestName":"Spencer Graham","slug":"spencer-graham","tags":["personal-crm","agentic-coding","context-systems","open-source"],"links":{"eventURL":"https://portal.raidguild.org/events/49","posts":["https://portal.raidguild.org/posts/field-notes-spencer-graham-atomized-developer","https://portal.raidguild.org/posts/agentic-coding-doer-reviewer-pipeline"],"shared":["https://spengrah.com/","https://x.com/spengrah","https://www.linkedin.com/in/spencergraham/","https://github.com/slgraham"]}},{"eventId":53,"guestName":"Adam Kerpelman","slug":"adam-kerpelman","tags":["ai-in-education","proxy-collapse","taste","assessment"],"links":{"eventURL":"https://portal.raidguild.org/events/53","posts":["https://portal.raidguild.org/posts/when-execution-gets-cheap-taste-gets-expensive","https://portal.raidguild.org/posts/proxy-collapse-came-for-the-reflection-paper"],"shared":["https://www.linkedin.com/in/adamkerpelman/","https://x.com/thekerp","https://jurisproject.io","https://hypercontext.com","https://www.coursetools.ai/"]}},{"eventId":50,"guestName":"Justice Conder","slug":"justice-conder","tags":["engineering","friction","agents","product"],"links":{"eventURL":"https://portal.raidguild.org/events/50","posts":["https://portal.raidguild.org/posts/engineering-did-not-end-it-moved-up-the-stack","https://portal.raidguild.org/posts/without-friction-there-is-no-moat"],"shared":["https://www.linkedin.com/in/justiceconder","https://x.com/singularityhack"]}},{"eventId":51,"guestName":"Elco","slug":"elco","tags":["agent-orchestration","shared-state","coordination"],"links":{"eventURL":"https://portal.raidguild.org/events/51","posts":["https://portal.raidguild.org/posts/hard-part-agent-orchestration-shared-state"],"shared":["http://www.elco.work/","https://x.com/elementcomplex"]}},{"eventId":67,"guestName":"Cj Miller","slug":"cj-miller","tags":["creative-work","human-edit","judgment"],"links":{"eventURL":"https://portal.raidguild.org/events/67","posts":["https://portal.raidguild.org/posts/fireside-cj-miller-premium-human-edit"],"shared":["https://tmco.io/"]}},{"eventId":65,"guestName":"Andrej Berlin","slug":"andrej-berlin","tags":["community","alignment","taste"],"links":{"eventURL":"https://portal.raidguild.org/events/65","posts":["https://portal.raidguild.org/posts/ai-can-shorten-alignment-but-it-cannot-choose-the-taste"],"shared":["https://deepwork.studio/","https://x.com/kischiman","https://de.linkedin.com/in/andrejberlin","https://andrejberlin.substack.com/"]}},{"eventId":58,"guestName":"Victor Ginelli","slug":"victor-ginelli","tags":["developer-workflows","review","cli","orchestration"],"links":{"eventURL":"https://portal.raidguild.org/events/58","posts":["https://portal.raidguild.org/posts/the-middle-is-easy-now-the-edges-are-the-work"],"shared":["https://github.com/youngkidwarrior","https://x.com/send2vic","https://she.energy/"]}},{"eventId":69,"guestName":"0xHunter","slug":"0xhunter","tags":["context-transfer","consulting","self-hosting","codex"],"links":{"eventURL":"https://portal.raidguild.org/events/69","posts":["https://portal.raidguild.org/posts/context-transfer-is-the-new-agency-bottleneck"],"shared":["https://ddls.consulting/","https://hunter.cafe/"]}},{"eventId":68,"guestName":"Jake Winckowski","slug":"jake-winckowski","tags":["support-workflows","coworker-agent","qa"],"links":{"eventURL":"https://portal.raidguild.org/events/68","posts":["https://portal.raidguild.org/posts/the-first-useful-ai-support-agent-is-a-coworker-not-a-replacement"],"shared":["https://www.linkedin.com/in/jakewinckowski","https://www.hackerone.com/"]}},{"eventId":52,"guestName":"Graven Prest","slug":"graven-prest","tags":["product-management","public-goods","testing","bottlenecks"],"links":{"eventURL":"https://portal.raidguild.org/events/52","posts":["https://portal.raidguild.org/posts/product-bottlenecks-are-moving"],"shared":["https://flowstate.network","https://x.com/GravenPrest","https://www.linkedin.com/in/gravenprest","https://octant.app"]}},{"eventId":66,"guestName":"Travis McCutcheon","slug":"travis-mccutcheon","tags":["agent-readiness","cleanup","operations"],"links":{"eventURL":"https://portal.raidguild.org/events/66","posts":["https://portal.raidguild.org/posts/agent-readiness-starts-with-cleanup-not-magic"],"shared":["https://missions.md","https://farcaster.xyz/earth2travis","https://x.com/earth2travis"]}},{"eventId":76,"guestName":"Bill Warren","slug":"bill-warren","tags":["product-workflows","prototyping","judgment","scope"],"links":{"eventURL":"https://portal.raidguild.org/events/76","posts":["https://portal.raidguild.org/posts/after-the-prototype-judgment-does-the-work"],"shared":["https://www.linkedin.com/in/bill-warren-product","https://x.com/0xBillW","https://www.summon.xyz/","https://game7.io/"]}},{"eventId":59,"guestName":"Sara Brown","slug":"sara-brown","tags":["ad-ops","qa","domain-agents"],"links":{"eventURL":"https://portal.raidguild.org/events/59","posts":["https://portal.raidguild.org/posts/first-useful-ad-ops-agent-qa-auditor"],"shared":["https://www.linkedin.com/in/sara-b-876693120/"]}},{"eventId":78,"guestName":"Josh Tyson","slug":"josh-tyson","tags":["agent-first-design","creative-labor","source-of-truth","scope-creep","content-workflows","onereach-ai","agentic-companion"],"links":{"eventURL":"https://portal.raidguild.org/events/78","posts":["https://portal.raidguild.org/posts/the-hard-part-of-ai-production-is-stopping"],"shared":["https://uxmag.com/podcasts","https://www.linkedin.com/in/josh-tyson-1132a27","https://uxmag.com/contributors/josh-tyson-ux-magazine","https://onereach.ai/invisible-machines/"]}}],"synthesis":[{"id":"working-on-now","question":"What are you working on right now?","insertAfterEncounterIndex":1},{"id":"workflow-changed","question":"How has AI or automation changed your workflow?","insertAfterEncounterIndex":3},{"id":"useful-tools","question":"What tools, prompts, agents, or processes are actually useful?","insertAfterEncounterIndex":5},{"id":"broken-risky","question":"What still feels broken, risky, or confusing?","insertAfterEncounterIndex":7},{"id":"builders-should-learn","question":"What should other builders learn from your experience?","insertAfterEncounterIndex":9},{"id":"future-outlook","question":"Where do you see the future — optimistic or doomer?","insertAfterEncounterIndex":11}],"journey":[{"index":0,"kind":"hero","id":"spencer-graham","label":"Spencer Graham","eventId":49},{"index":1,"kind":"hero","id":"adam-kerpelman","label":"Adam Kerpelman","eventId":53},{"index":2,"kind":"cross-cut","id":"working-on-now","label":"What are you working on right now?","afterEncounterIndex":1},{"index":3,"kind":"hero","id":"justice-conder","label":"Justice Conder","eventId":50},{"index":4,"kind":"hero","id":"elco","label":"Elco","eventId":51},{"index":5,"kind":"cross-cut","id":"workflow-changed","label":"How has AI or automation changed your workflow?","afterEncounterIndex":3},{"index":6,"kind":"hero","id":"cj-miller","label":"Cj Miller","eventId":67},{"index":7,"kind":"hero","id":"andrej-berlin","label":"Andrej Berlin","eventId":65},{"index":8,"kind":"cross-cut","id":"useful-tools","label":"What tools, prompts, agents, or processes are actually useful?","afterEncounterIndex":5},{"index":9,"kind":"hero","id":"victor-ginelli","label":"Victor Ginelli","eventId":58},{"index":10,"kind":"hero","id":"0xhunter","label":"0xHunter","eventId":69},{"index":11,"kind":"cross-cut","id":"broken-risky","label":"What still feels broken, risky, or confusing?","afterEncounterIndex":7},{"index":12,"kind":"hero","id":"jake-winckowski","label":"Jake Winckowski","eventId":68},{"index":13,"kind":"hero","id":"graven-prest","label":"Graven Prest","eventId":52},{"index":14,"kind":"cross-cut","id":"builders-should-learn","label":"What should other builders learn from your experience?","afterEncounterIndex":9},{"index":15,"kind":"hero","id":"travis-mccutcheon","label":"Travis McCutcheon","eventId":66},{"index":16,"kind":"hero","id":"bill-warren","label":"Bill Warren","eventId":76},{"index":17,"kind":"cross-cut","id":"future-outlook","label":"Where do you see the future — optimistic or doomer?","afterEncounterIndex":11},{"index":18,"kind":"hero","id":"sara-brown","label":"Sara Brown","eventId":59},{"index":19,"kind":"hero","id":"josh-tyson","label":"Josh Tyson","eventId":78}]},"sourceMaterial":{"meta":{"title":"Field Experience from the Edge","subtitle":"How builders are adapting AI in real work","sessionCount":14,"threadURL":"https://portal.raidguild.org/threads/how-to-raidguild-field-experience-from-the-edge"},"throughLines":[{"id":"bottleneck-shift","title":"The bottleneck moved upstream","summary":"Execution got cheaper. Taste, judgment, QA, and production readiness became the scarce work."},{"id":"role-change","title":"Developer identity is changing","summary":"From hand-coding to orchestrating agents, reviewing output, and building CLI surfaces agents can use."},{"id":"context-crisis","title":"Context is the unsolved layer","summary":"Shared memory, scoped context, and atomized private agent loops threaten coordination."},{"id":"verification","title":"Verification everywhere","summary":"Frontend QA, oral exams, PR review, and coworker-not-replacement support agents."}],"encounters":[{"eventId":49,"guestName":"Spencer Graham","slug":"spencer-graham","startsAt":"2026-06-10T16:30:00.000Z","heroArchetype":"Scribe","spriteSlug":"scribe","hook":"Personal CRM, doer/reviewer agentic coding, and the risk that private agent loops thin shared context.","highlights":[{"text":"Agentic coding works better as a doer/reviewer pipeline with deterministic pre-push checks.","source":"summary"},{"text":"A self-hosted personal CRM ingests messages and transcripts without handing full history to third parties.","source":"summary"},{"text":"More work mediated by personal agents could unbundle project work from the relationships that make collaboration meaningful.","source":"summary"},{"text":"Open source may get stronger in the AI era when bug-finding is paired with repair and verification loops.","source":"summary"}],"tags":["personal-crm","agentic-coding","context-systems","open-source"],"media":{"recapYouTubeId":"0GrJo-mb70k","fullInterviewURL":"https://youtu.be/GAT4oMhf8Tk"},"links":{"eventURL":"https://portal.raidguild.org/events/49","posts":[{"title":"The Atomized Developer","url":"https://portal.raidguild.org/posts/field-notes-spencer-graham-atomized-developer"},{"title":"Agentic Coding Works Better as a Doer/Reviewer Pipeline","url":"https://portal.raidguild.org/posts/agentic-coding-doer-reviewer-pipeline"}],"shared":[{"title":"Spencer Graham site","url":"https://spengrah.com/"},{"title":"Spencer Graham X","url":"https://x.com/spengrah"},{"title":"Spencer Graham LinkedIn","url":"https://www.linkedin.com/in/spencergraham/"},{"title":"Spencer Graham GitHub","url":"https://github.com/slgraham"}]},"bio":"RaidGuild veteran and DAOhaus alum building a self-hosted personal CRM. Daily driver: Claude Code with doer/reviewer agent loops."},{"eventId":53,"guestName":"Adam Kerpelman","slug":"adam-kerpelman","startsAt":"2026-06-11T17:30:00.000Z","heroArchetype":"Wizard","spriteSlug":"wizard","hook":"Proxy collapse in education, AI as patient tutor, and the shift from execution scarcity to taste and judgment.","highlights":[{"text":"Reflection papers used to proxy for understanding — AI makes those proxies unreliable.","source":"summary"},{"text":"Oral exams and interview-style assessment move evaluation closer to live explanation.","source":"summary"},{"text":"When execution gets cheap, taste, sequencing, and distribution get expensive.","source":"summary"},{"text":"Model orchestration and context management remain daily pain points even for power users.","source":"summary"}],"tags":["ai-in-education","proxy-collapse","taste","assessment"],"media":{"recapYouTubeId":"PqHWbPfOZmw","fullInterviewURL":"https://youtu.be/LNOk_voJNkg"},"links":{"eventURL":"https://portal.raidguild.org/events/53","posts":[{"title":"When Execution Gets Cheap, Taste Gets Expensive","url":"https://portal.raidguild.org/posts/when-execution-gets-cheap-taste-gets-expensive"},{"title":"Proxy Collapse Came For The Reflection Paper","url":"https://portal.raidguild.org/posts/proxy-collapse-came-for-the-reflection-paper"}],"shared":[{"title":"Adam Kerpelman LinkedIn","url":"https://www.linkedin.com/in/adamkerpelman/"},{"title":"Adam Kerpelman X","url":"https://x.com/thekerp"},{"title":"Juris Project","url":"https://jurisproject.io"},{"title":"Hypercontext","url":"https://hypercontext.com"},{"title":"Course Tools","url":"https://www.coursetools.ai/"}]},"bio":"UVA McIntire professor teaching applied AI and entrepreneurship. Running live experiments on assessment after proxy collapse."},{"eventId":50,"guestName":"Justice Conder","slug":"justice-conder","startsAt":"2026-06-11T18:30:00.000Z","heroArchetype":"Warrior","spriteSlug":"warrior","hook":"Engineering did not end — it moved up the stack. Friction still matters.","highlights":[{"text":"Agents change product and engineering work without removing the need for judgment at the edges.","source":"summary"},{"text":"Without friction, there is no moat — some resistance still shapes durable work.","source":"summary"},{"text":"The grind of hand-coding feels less central, but specification and review gain leverage.","source":"summary"}],"tags":["engineering","friction","agents","product"],"media":{"recapYouTubeId":"exVWxi4vSBE","fullInterviewURL":"https://youtu.be/-zAISJjU2RQ"},"links":{"eventURL":"https://portal.raidguild.org/events/50","posts":[{"title":"Engineering Did Not End. It Moved Up the Stack.","url":"https://portal.raidguild.org/posts/engineering-did-not-end-it-moved-up-the-stack"},{"title":"Without Friction, There Is No Moat","url":"https://portal.raidguild.org/posts/without-friction-there-is-no-moat"}],"shared":[{"title":"Justice Conder LinkedIn","url":"https://www.linkedin.com/in/justiceconder"},{"title":"Justice Conder X","url":"https://x.com/singularityhack"}]},"bio":"Engineer and builder thinking about agents, product friction, and where judgment still lives in the stack."},{"eventId":51,"guestName":"Elco","slug":"elco","startsAt":"2026-06-12T16:30:00.000Z","heroArchetype":"Alchemist","spriteSlug":"alchemist","hook":"The hard part of agent orchestration is shared state — not the prompt.","highlights":[{"text":"Multi-agent workflows break when state and handoffs are implicit.","source":"summary"},{"text":"Coordination harnesses need observable checkpoints humans can inspect.","source":"summary"},{"text":"Local agent loops are powerful but can hide process from the team.","source":"summary"}],"tags":["agent-orchestration","shared-state","coordination"],"media":{"recapYouTubeId":"jBbT-nxAwX4","fullInterviewURL":"https://youtu.be/TYrT7HPUbPg"},"links":{"eventURL":"https://portal.raidguild.org/events/51","posts":[{"title":"The Hard Part Of Agent Orchestration Is Shared State","url":"https://portal.raidguild.org/posts/hard-part-agent-orchestration-shared-state"}],"shared":[{"title":"Elco site","url":"http://www.elco.work/"},{"title":"Elco X","url":"https://x.com/elementcomplex"}]},"bio":"RaidGuild builder working on agent orchestration, shared state, and coordination harnesses for multi-agent work."},{"eventId":67,"guestName":"Cj Miller","slug":"cj-miller","startsAt":"2026-06-16T16:30:00.000Z","heroArchetype":"Monk","spriteSlug":"monk","hook":"When production gets cheap, the premium is the human edit.","highlights":[{"text":"AI-assisted production still needs a human gate for taste and final cut.","source":"summary"},{"text":"Speed without editorial judgment produces volume, not signal.","source":"summary"}],"tags":["creative-work","human-edit","judgment"],"media":{"recapYouTubeId":"lGj9YsVIruw","fullInterviewURL":null},"links":{"eventURL":"https://portal.raidguild.org/events/67","posts":[{"title":"The Premium Is the Human Edit","url":"https://portal.raidguild.org/posts/fireside-cj-miller-premium-human-edit"}],"shared":[{"title":"TMCO","url":"https://tmco.io/"}]},"bio":"Creative producer focused on editorial craft — the human edit as premium when AI accelerates production."},{"eventId":65,"guestName":"Andrej Berlin","slug":"andrej-berlin","startsAt":"2026-06-17T17:00:00.000Z","heroArchetype":"Paladin","spriteSlug":"paladin","hook":"AI can shorten alignment, but it cannot choose the taste.","highlights":[{"text":"Community tools may need to matchmake contributors, not just host chat.","source":"summary"},{"text":"Alignment work compresses with AI; aesthetic and values choices do not.","source":"summary"}],"tags":["community","alignment","taste"],"media":{"recapYouTubeId":"0k59966ivMw","fullInterviewURL":null},"links":{"eventURL":"https://portal.raidguild.org/events/65","posts":[{"title":"AI Can Shorten Alignment, But It Cannot Choose The Taste","url":"https://portal.raidguild.org/posts/ai-can-shorten-alignment-but-it-cannot-choose-the-taste"}],"shared":[{"title":"Deep Work Network","url":"https://deepwork.studio/"},{"title":"Andrej Berlin X","url":"https://x.com/kischiman"},{"title":"Andrej Berlin LinkedIn","url":"https://de.linkedin.com/in/andrejberlin"},{"title":"Andrej Berlin Substack","url":"https://andrejberlin.substack.com/"}]},"bio":"Community and alignment builder exploring taste, contributor matchmaking, and tools beyond chat rooms."},{"eventId":58,"guestName":"Victor Ginelli","slug":"victor-ginelli","startsAt":"2026-06-17T22:00:00.000Z","heroArchetype":"Ranger","spriteSlug":"ranger","hook":"The new senior dev job is reviewing the machines — building CLIs and wrappers agents can run.","highlights":[{"text":"Trust in AI coding built gradually; early output needed as much verification as manual work.","source":"summary"},{"text":"The middle of a project is easier now; edges like deploy, infra, and governance remain human-heavy.","source":"summary"},{"text":"Small command surfaces and internal CLIs become high-leverage agent interfaces.","source":"summary"}],"tags":["developer-workflows","review","cli","orchestration"],"media":{"recapYouTubeId":"Ds4EZ4Jpt9I","fullInterviewURL":"https://youtu.be/Ds4EZ4Jpt9I"},"links":{"eventURL":"https://portal.raidguild.org/events/58","posts":[{"title":"The Middle Is Easy Now. The Edges Are The Work.","url":"https://portal.raidguild.org/posts/the-middle-is-easy-now-the-edges-are-the-work"}],"shared":[{"title":"Victor Ginelli GitHub","url":"https://github.com/youngkidwarrior"},{"title":"Victor Ginelli X","url":"https://x.com/send2vic"},{"title":"SHE","url":"https://she.energy/"}]},"bio":"Experienced developer moving from hand-coding toward orchestrating AI coworkers and CLI surfaces agents can run."},{"eventId":69,"guestName":"0xHunter","slug":"0xhunter","startsAt":"2026-06-23T16:00:00.000Z","heroArchetype":"Hunter","spriteSlug":"hunter","hook":"Context transfer is the new agency bottleneck — teams still copy-paste between agent sessions.","highlights":[{"text":"Custom software gets cheaper; medium businesses may own more of their stack.","source":"summary"},{"text":"Codex-heavy workflows work when scope is clear; frontend QA still needs human review.","source":"summary"},{"text":"Scoped context sharing between clients, projects, and teammates remains unsolved.","source":"summary"},{"text":"Some clients now open PRs on internal tools that the consultancy reviews.","source":"summary"}],"tags":["context-transfer","consulting","self-hosting","codex"],"media":{"recapYouTubeId":"Y-Xlv84NbTA","fullInterviewURL":"https://youtu.be/iTCxYD5pM4A"},"links":{"eventURL":"https://portal.raidguild.org/events/69","posts":[{"title":"Context Transfer Is The New Agency Bottleneck","url":"https://portal.raidguild.org/posts/context-transfer-is-the-new-agency-bottleneck"}],"shared":[{"title":"DDLS Consulting","url":"https://ddls.consulting/"},{"title":"hunter.cafe","url":"https://hunter.cafe/"}]},"bio":"Founder of DDLS Consulting — custom software, self-hosting, and Codex-heavy workflows for operational clients."},{"eventId":68,"guestName":"Jake Winckowski","slug":"jake-winckowski","startsAt":"2026-06-23T17:00:00.000Z","heroArchetype":"Healer","spriteSlug":"healer","hook":"The first useful AI support agent is a coworker, not a replacement.","highlights":[{"text":"Reviewed automation in support workflows beats magic all-purpose agents.","source":"summary"},{"text":"Domain-specific tooling and human checkpoints matter more than generic prompts.","source":"summary"}],"tags":["support-workflows","coworker-agent","qa"],"media":{"recapYouTubeId":"nzEe5-Rm3Ic","fullInterviewURL":"https://youtu.be/GbUmUQtjli4"},"links":{"eventURL":"https://portal.raidguild.org/events/68","posts":[{"title":"Reviewed AI Support Agents Start As Coworkers","url":"https://portal.raidguild.org/posts/the-first-useful-ai-support-agent-is-a-coworker-not-a-replacement"}],"shared":[{"title":"Jake Winckowski LinkedIn","url":"https://www.linkedin.com/in/jakewinckowski"},{"title":"HackerOne","url":"https://www.hackerone.com/"}]},"bio":"Builder of reviewed support workflows — useful AI agents start as coworkers with human checkpoints."},{"eventId":52,"guestName":"Graven Prest","slug":"graven-prest","startsAt":"2026-06-24T17:00:00.000Z","heroArchetype":"Druid","spriteSlug":"druid","hook":"Product bottlenecks are moving — speed and ambition rose, but testing and should-we decisions got harder.","highlights":[{"text":"Small teams can ship overnight what used to take a week — bottlenecks shift to QA and judgment.","source":"summary"},{"text":"Public goods and builder-relations work still needs distribution and production readiness.","source":"summary"},{"text":"Maybe yes we could, but should we? becomes the recurring product question.","source":"summary"}],"tags":["product-management","public-goods","testing","bottlenecks"],"media":{"recapYouTubeId":"m02tfKU3azw","fullInterviewURL":"https://youtu.be/U-9XKvwsNrU"},"links":{"eventURL":"https://portal.raidguild.org/events/52","posts":[{"title":"Product Bottlenecks Are Moving","url":"https://portal.raidguild.org/posts/product-bottlenecks-are-moving"}],"shared":[{"title":"Flow State Network","url":"https://flowstate.network"},{"title":"Graven Prest X","url":"https://x.com/GravenPrest"},{"title":"Graven Prest LinkedIn","url":"https://www.linkedin.com/in/gravenprest"},{"title":"Octant Dashboard","url":"https://octant.app"}]},"bio":"Product and builder-relations work around public goods, Octant, and Flow State. Claude Code in the terminal."},{"eventId":66,"guestName":"Travis McCutcheon","slug":"travis-mccutcheon","startsAt":"2026-06-24T20:30:00.000Z","heroArchetype":"Cleric","spriteSlug":"cleric","hook":"Agent readiness starts with cleanup — not magic prompts.","highlights":[{"text":"Repos and docs need hygiene before agents can operate reliably.","source":"summary"},{"text":"Realistic automation gates beat aspirational full-agent replacement.","source":"summary"}],"tags":["agent-readiness","cleanup","operations"],"media":{"recapYouTubeId":"3W6HqjzB6TE","fullInterviewURL":"https://youtu.be/ReRgmFTJtQE"},"links":{"eventURL":"https://portal.raidguild.org/events/66","posts":[{"title":"Agent Readiness Starts With Cleanup, Not Magic","url":"https://portal.raidguild.org/posts/agent-readiness-starts-with-cleanup-not-magic"}],"shared":[{"title":"missions.md","url":"https://missions.md"},{"title":"Earth2Travis Farcaster","url":"https://farcaster.xyz/earth2travis"},{"title":"Earth2Travis X","url":"https://x.com/earth2travis"}]},"bio":"Operator focused on repo hygiene, cleanup, and realistic gates before agents can do reliable work."},{"eventId":76,"guestName":"Bill Warren","slug":"bill-warren","startsAt":"2026-06-25T16:30:00.000Z","heroArchetype":"Tavern Keeper","spriteSlug":"tavern-keeper","hook":"After the prototype, judgment does the work — PRDs compress into testable builds fast.","highlights":[{"text":"Brain dump → PRD scaffold → agent prompts → prototype is the new product loop.","source":"summary"},{"text":"AI can bloat products unless someone keeps scope simple and the experience delightful.","source":"summary"},{"text":"Bullish on democratized creation; cautious about hype cycles and AGI timelines.","source":"summary"}],"tags":["product-workflows","prototyping","judgment","scope"],"media":{"recapYouTubeId":"mYHv8Dcjb64","fullInterviewURL":"https://youtu.be/Fi28v2nRzTE"},"links":{"eventURL":"https://portal.raidguild.org/events/76","posts":[{"title":"After the Prototype, Judgment Does the Work","url":"https://portal.raidguild.org/posts/after-the-prototype-judgment-does-the-work"}],"shared":[{"title":"Bill Warren LinkedIn","url":"https://www.linkedin.com/in/bill-warren-product"},{"title":"Bill Warren X","url":"https://x.com/0xBillW"},{"title":"Summon","url":"https://www.summon.xyz/"},{"title":"Game7","url":"https://game7.io/"}]},"bio":"Product lead with Protocol Labs background — brain dump to PRD to prototype is the new loop."},{"eventId":59,"guestName":"Sara Brown","slug":"sara-brown","startsAt":"2026-06-25T19:00:00.000Z","heroArchetype":"Archer","spriteSlug":"archer","hook":"The first useful ad ops agent is a QA auditor — domain judgment beats generic automation.","highlights":[{"text":"Programmatic ad workflows need audited checks, not blind generation.","source":"summary"},{"text":"Domain-specific QA agents outperform all-purpose copilots in operations work.","source":"summary"}],"tags":["ad-ops","qa","domain-agents"],"media":{"recapYouTubeId":null,"fullInterviewURL":null},"links":{"eventURL":"https://portal.raidguild.org/events/59","posts":[{"title":"The First Useful Ad Ops Agent Is A QA Auditor","url":"https://portal.raidguild.org/posts/first-useful-ad-ops-agent-qa-auditor"}],"shared":[{"title":"Sara Brown LinkedIn","url":"https://www.linkedin.com/in/sara-b-876693120/"}]},"bio":"Ad ops practitioner building domain QA agents — audited checks beat blind generation in programmatic workflows."},{"eventId":78,"guestName":"Josh Tyson","slug":"josh-tyson","startsAt":"2026-06-26T18:30:00.000Z","heroArchetype":"Rogue","spriteSlug":"rogue","hook":"Creative labor after LLMs — synthesis and taste over raw drafting, with agent-first design and stopping rules.","highlights":[{"text":"Writer value shifted to rumination, research, and synthesis — not competing on raw drafting volume.","source":"summary"},{"text":"Individual AI productivity is often wasted without organizational source-of-truth and scoped data access.","source":"summary"},{"text":"Agent-first experiences and trusted personal agents may mediate how people interact with other systems.","source":"summary"},{"text":"When everything is possible, the production skill is knowing when to stop — scope creep and token maxing dilute the work.","source":"summary"},{"text":"Built the Fuzzy novel digital companion on OneReach.ai — an agentic reader experience, not just a static ebook.","source":"summary"}],"tags":["agent-first-design","creative-labor","source-of-truth","scope-creep","content-workflows","onereach-ai","agentic-companion"],"media":{"recapYouTubeId":"_YVeBGhbY_s","fullInterviewURL":null},"links":{"eventURL":"https://portal.raidguild.org/events/78","posts":[{"title":"The Hard Part Of AI Production Is Stopping","url":"https://portal.raidguild.org/posts/the-hard-part-of-ai-production-is-stopping"}],"shared":[{"title":"UX Magazine Podcasts","url":"https://uxmag.com/podcasts"},{"title":"Josh Tyson LinkedIn","url":"https://www.linkedin.com/in/josh-tyson-1132a27"},{"title":"Josh Tyson UX Magazine","url":"https://uxmag.com/contributors/josh-tyson-ux-magazine"},{"title":"Age of Invisible Machines","url":"https://onereach.ai/invisible-machines/"}]},"bio":"UX Magazine editor and co-author of Age of Invisible Machines. Rebuilt the magazine with Cursor and Claude; built the Fuzzy novel companion on the OneReach.ai platform."}],"synthesis":[{"id":"working-on-now","insertAfterEncounterIndex":1,"question":"What are you working on right now?","eyebrow":"Cross-interview synthesis","body":["The cohort was not a single kind of builder. Spencer is wiring a self-hosted personal CRM that ingests messages, transcripts, and relationship context without handing a full life archive to a third party. Kerp is running applied-AI teaching experiments at UVA — course tooling, oral exams, and classroom workflows that assume AI is already in the room. Hunter runs a consultancy building custom, often self-hosted operational software for medium-sized businesses. Graven sits at the intersection of public goods, builder relations, and protocol work around Octant and Flow State.","Bill Warren compresses product loops inside Protocol Labs-style network experiments. Sara Brown works ad ops and programmatic QA. Jake Winckowski builds reviewed support automation. Elco, Justice, Victor, Travis, Cj, and Andrej each brought a different edge: orchestration harnesses, engineering identity, senior review workflows, repo hygiene, human editorial craft, and community alignment tools.","What repeats is not one stack or one job title. People are building relationship memory, classroom infrastructure, client-owned software, product prototypes, domain QA, and coordination surfaces — all while AI is becoming part of the delivery path, not a side experiment."],"voices":["Spencer Graham","Adam Kerpelman","0xHunter","Graven Prest","Bill Warren","Sara Brown"]},{"id":"workflow-changed","insertAfterEncounterIndex":3,"question":"How has AI or automation changed your workflow?","eyebrow":"Cross-interview synthesis","body":["Nearly everyone reported a shift in where time goes — not necessarily less work, but different work. Product and engineering loops compressed: brain dump to PRD scaffold to prototype arrived repeatedly, from Bill and Graven to Jake and Justice. Small teams described shipping overnight what used to take a week, then hitting harder stops at testing, scope, production readiness, and \"should we?\" decisions.","Developers consistently described an identity move. Victor and Spencer talked about orchestrating and reviewing rather than typing every line. Justice framed engineering as moving up the stack. Kerp said execution got cheaper while taste, sequencing, and assessment got more expensive. Cj argued the premium moved to the human edit when production sped up.","The emotional texture varied. Some guests felt more leverage and enjoyment once tooling improved; others flagged existential unease when hand-coding felt less central. Nobody described a clean replacement story. The through-line is a workflow rewrite: faster middle, heavier edges, more judgment."],"voices":["Bill Warren","Graven Prest","Victor Ginelli","Adam Kerpelman","Justice Conder","Cj Miller"]},{"id":"useful-tools","insertAfterEncounterIndex":5,"question":"What tools, prompts, agents, or processes are actually useful?","eyebrow":"Cross-interview synthesis","body":["Useful patterns were more about process than magic prompts. Spencer's doer/reviewer pipeline — coordinate, implement, review, repeat, with deterministic pre-push checks — was the clearest agentic coding pattern. Hunter praised Codex for reliable full-stack and browser-facing workflows. Victor emphasized small CLIs and command surfaces agents can actually run. Travis argued agent readiness starts with repo cleanup, not a better prompt.","Kerp got mileage from AI as a patient tutor and from course infrastructure like Canvas-integrated tooling, not from one omnibus model call. Jake and Sara both pointed at narrow, domain-reviewed agents: support coworkers and ad-ops QA auditors rather than general copilots. Elco's lesson was orchestration with observable shared state, not more autonomous actors.","What did not earn trust: one-shot frontend polish, unreviewed support automation, giant-context models without real project awareness, and all-purpose agents asked to own entire workflows. The useful layer is bounded tools, review gates, and repeatable internal interfaces."],"voices":["Spencer Graham","0xHunter","Victor Ginelli","Travis McCutcheon","Jake Winckowski","Sara Brown","Elco"]},{"id":"broken-risky","insertAfterEncounterIndex":7,"question":"What still feels broken, risky, or confusing?","eyebrow":"Cross-interview synthesis","body":["Context transfer was the loudest shared pain. Hunter said teams still copy-paste between agent sessions; Spencer worried about atomized developers and thinning shared context; Elco named shared state as the hard part of orchestration. Everyone had a version of: private agent loops work locally, but the team cannot see or inherit the process.","Verification showed up everywhere in different costumes. Frontend QA remains distrusted without human review. Kerp described proxy collapse in education — assignments that no longer prove understanding. Sara wanted audited checks instead of blind generation in operations work. Graven and Bill both warned that speed invites bloated products without judgment gates.","Model orchestration is still guesswork for power users: which model for which task, when to escalate, how to preserve memory across machines and tools. Andrej and Victor added social risk — coordination, taste, governance, and the edges of projects (deploy, infra, maintenance) remain stubbornly human-heavy."],"voices":["0xHunter","Spencer Graham","Elco","Adam Kerpelman","Sara Brown","Graven Prest","Victor Ginelli"]},{"id":"builders-should-learn","insertAfterEncounterIndex":9,"question":"What should other builders learn from your experience?","eyebrow":"Cross-interview synthesis","body":["The cohort advice was practical, not futurist. Clean up repos and docs before expecting agents to be dependable. Build review into the workflow — second model, human checkpoint, oral exam, QA auditor, or coworker-not-replacement support pattern. Keep scope small enough that someone with taste can still say no.","Invest in interfaces agents can use: CLIs, narrow commands, internal tools, and explicit handoffs beat giant prompts. Treat relationship and organizational context as infrastructure, not chat history. For teams, shared memory and scoped context matter as much as model choice.","Several guests pushed back on hype. Friction can be a moat. Open source may strengthen if bug-finding is paired with repair loops. Community tools may need to matchmake contributors, not just host more chat. The lesson for other builders: AI amplifies whatever process you already have — strong or weak."],"voices":["Travis McCutcheon","Spencer Graham","Jake Winckowski","Justice Conder","Andrej Berlin","Victor Ginelli"]},{"id":"future-outlook","insertAfterEncounterIndex":11,"question":"Where do you see the future — optimistic or doomer?","eyebrow":"Cross-interview synthesis","body":["Most firesides ended with a futures question, and the answers were mixed without being nihilistic. Bill was bullish on democratized creation but cautious about AGI hype cycles and cultural fallout from layoff narratives. Spencer rejected \"open source is dead\" and argued security could improve if discovery is paired with repair. Kerp pushed back on simple job-collapse stories while acknowledging real student anxiety.","Optimism clustered around leverage for small teams, cheaper custom software, better tutoring, faster alignment on prototypes, and stronger OSS repair loops. Doom-leaning notes clustered around atomized work, lost shared context, proxy collapse in institutions, frontend trust gaps, and vendors overclaiming autonomy.","The balanced take across sessions: roles shift rather than vanish; taste, judgment, distribution, and verification become the durable professions inside the stack. The future is not one camp. It is a rearrangement where builders who can coordinate agents, review output, and preserve human context do better than builders who only type faster."],"voices":["Bill Warren","Spencer Graham","Adam Kerpelman","Graven Prest","0xHunter","Andrej Berlin"]}]},"journey":{"segments":[{"index":0,"kind":"hero","encounterIndex":0,"encounter":{"eventId":49,"guestName":"Spencer Graham","slug":"spencer-graham","startsAt":"2026-06-10T16:30:00.000Z","heroArchetype":"Scribe","spriteSlug":"scribe","hook":"Personal CRM, doer/reviewer agentic coding, and the risk that private agent loops thin shared context.","highlights":[{"text":"Agentic coding works better as a doer/reviewer pipeline with deterministic pre-push checks.","source":"summary"},{"text":"A self-hosted personal CRM ingests messages and transcripts without handing full history to third parties.","source":"summary"},{"text":"More work mediated by personal agents could unbundle project work from the relationships that make collaboration meaningful.","source":"summary"},{"text":"Open source may get stronger in the AI era when bug-finding is paired with repair and verification loops.","source":"summary"}],"tags":["personal-crm","agentic-coding","context-systems","open-source"],"media":{"recapYouTubeId":"0GrJo-mb70k","fullInterviewURL":"https://youtu.be/GAT4oMhf8Tk"},"links":{"eventURL":"https://portal.raidguild.org/events/49","posts":[{"title":"The Atomized Developer","url":"https://portal.raidguild.org/posts/field-notes-spencer-graham-atomized-developer"},{"title":"Agentic Coding Works Better as a Doer/Reviewer Pipeline","url":"https://portal.raidguild.org/posts/agentic-coding-doer-reviewer-pipeline"}],"shared":[{"title":"Spencer Graham site","url":"https://spengrah.com/"},{"title":"Spencer Graham X","url":"https://x.com/spengrah"},{"title":"Spencer Graham LinkedIn","url":"https://www.linkedin.com/in/spencergraham/"},{"title":"Spencer Graham GitHub","url":"https://github.com/slgraham"}]},"bio":"RaidGuild veteran and DAOhaus alum building a self-hosted personal CRM. Daily driver: Claude Code with doer/reviewer agent loops."}},{"index":1,"kind":"hero","encounterIndex":1,"encounter":{"eventId":53,"guestName":"Adam Kerpelman","slug":"adam-kerpelman","startsAt":"2026-06-11T17:30:00.000Z","heroArchetype":"Wizard","spriteSlug":"wizard","hook":"Proxy collapse in education, AI as patient tutor, and the shift from execution scarcity to taste and judgment.","highlights":[{"text":"Reflection papers used to proxy for understanding — AI makes those proxies unreliable.","source":"summary"},{"text":"Oral exams and interview-style assessment move evaluation closer to live explanation.","source":"summary"},{"text":"When execution gets cheap, taste, sequencing, and distribution get expensive.","source":"summary"},{"text":"Model orchestration and context management remain daily pain points even for power users.","source":"summary"}],"tags":["ai-in-education","proxy-collapse","taste","assessment"],"media":{"recapYouTubeId":"PqHWbPfOZmw","fullInterviewURL":"https://youtu.be/LNOk_voJNkg"},"links":{"eventURL":"https://portal.raidguild.org/events/53","posts":[{"title":"When Execution Gets Cheap, Taste Gets Expensive","url":"https://portal.raidguild.org/posts/when-execution-gets-cheap-taste-gets-expensive"},{"title":"Proxy Collapse Came For The Reflection Paper","url":"https://portal.raidguild.org/posts/proxy-collapse-came-for-the-reflection-paper"}],"shared":[{"title":"Adam Kerpelman LinkedIn","url":"https://www.linkedin.com/in/adamkerpelman/"},{"title":"Adam Kerpelman X","url":"https://x.com/thekerp"},{"title":"Juris Project","url":"https://jurisproject.io"},{"title":"Hypercontext","url":"https://hypercontext.com"},{"title":"Course Tools","url":"https://www.coursetools.ai/"}]},"bio":"UVA McIntire professor teaching applied AI and entrepreneurship. Running live experiments on assessment after proxy collapse."}},{"index":2,"kind":"cross-cut","afterEncounterIndex":1,"question":{"id":"working-on-now","insertAfterEncounterIndex":1,"question":"What are you working on right now?","eyebrow":"Cross-interview synthesis","body":["The cohort was not a single kind of builder. Spencer is wiring a self-hosted personal CRM that ingests messages, transcripts, and relationship context without handing a full life archive to a third party. Kerp is running applied-AI teaching experiments at UVA — course tooling, oral exams, and classroom workflows that assume AI is already in the room. Hunter runs a consultancy building custom, often self-hosted operational software for medium-sized businesses. Graven sits at the intersection of public goods, builder relations, and protocol work around Octant and Flow State.","Bill Warren compresses product loops inside Protocol Labs-style network experiments. Sara Brown works ad ops and programmatic QA. Jake Winckowski builds reviewed support automation. Elco, Justice, Victor, Travis, Cj, and Andrej each brought a different edge: orchestration harnesses, engineering identity, senior review workflows, repo hygiene, human editorial craft, and community alignment tools.","What repeats is not one stack or one job title. People are building relationship memory, classroom infrastructure, client-owned software, product prototypes, domain QA, and coordination surfaces — all while AI is becoming part of the delivery path, not a side experiment."],"voices":["Spencer Graham","Adam Kerpelman","0xHunter","Graven Prest","Bill Warren","Sara Brown"]}},{"index":3,"kind":"hero","encounterIndex":2,"encounter":{"eventId":50,"guestName":"Justice Conder","slug":"justice-conder","startsAt":"2026-06-11T18:30:00.000Z","heroArchetype":"Warrior","spriteSlug":"warrior","hook":"Engineering did not end — it moved up the stack. Friction still matters.","highlights":[{"text":"Agents change product and engineering work without removing the need for judgment at the edges.","source":"summary"},{"text":"Without friction, there is no moat — some resistance still shapes durable work.","source":"summary"},{"text":"The grind of hand-coding feels less central, but specification and review gain leverage.","source":"summary"}],"tags":["engineering","friction","agents","product"],"media":{"recapYouTubeId":"exVWxi4vSBE","fullInterviewURL":"https://youtu.be/-zAISJjU2RQ"},"links":{"eventURL":"https://portal.raidguild.org/events/50","posts":[{"title":"Engineering Did Not End. It Moved Up the Stack.","url":"https://portal.raidguild.org/posts/engineering-did-not-end-it-moved-up-the-stack"},{"title":"Without Friction, There Is No Moat","url":"https://portal.raidguild.org/posts/without-friction-there-is-no-moat"}],"shared":[{"title":"Justice Conder LinkedIn","url":"https://www.linkedin.com/in/justiceconder"},{"title":"Justice Conder X","url":"https://x.com/singularityhack"}]},"bio":"Engineer and builder thinking about agents, product friction, and where judgment still lives in the stack."}},{"index":4,"kind":"hero","encounterIndex":3,"encounter":{"eventId":51,"guestName":"Elco","slug":"elco","startsAt":"2026-06-12T16:30:00.000Z","heroArchetype":"Alchemist","spriteSlug":"alchemist","hook":"The hard part of agent orchestration is shared state — not the prompt.","highlights":[{"text":"Multi-agent workflows break when state and handoffs are implicit.","source":"summary"},{"text":"Coordination harnesses need observable checkpoints humans can inspect.","source":"summary"},{"text":"Local agent loops are powerful but can hide process from the team.","source":"summary"}],"tags":["agent-orchestration","shared-state","coordination"],"media":{"recapYouTubeId":"jBbT-nxAwX4","fullInterviewURL":"https://youtu.be/TYrT7HPUbPg"},"links":{"eventURL":"https://portal.raidguild.org/events/51","posts":[{"title":"The Hard Part Of Agent Orchestration Is Shared State","url":"https://portal.raidguild.org/posts/hard-part-agent-orchestration-shared-state"}],"shared":[{"title":"Elco site","url":"http://www.elco.work/"},{"title":"Elco X","url":"https://x.com/elementcomplex"}]},"bio":"RaidGuild builder working on agent orchestration, shared state, and coordination harnesses for multi-agent work."}},{"index":5,"kind":"cross-cut","afterEncounterIndex":3,"question":{"id":"workflow-changed","insertAfterEncounterIndex":3,"question":"How has AI or automation changed your workflow?","eyebrow":"Cross-interview synthesis","body":["Nearly everyone reported a shift in where time goes — not necessarily less work, but different work. Product and engineering loops compressed: brain dump to PRD scaffold to prototype arrived repeatedly, from Bill and Graven to Jake and Justice. Small teams described shipping overnight what used to take a week, then hitting harder stops at testing, scope, production readiness, and \"should we?\" decisions.","Developers consistently described an identity move. Victor and Spencer talked about orchestrating and reviewing rather than typing every line. Justice framed engineering as moving up the stack. Kerp said execution got cheaper while taste, sequencing, and assessment got more expensive. Cj argued the premium moved to the human edit when production sped up.","The emotional texture varied. Some guests felt more leverage and enjoyment once tooling improved; others flagged existential unease when hand-coding felt less central. Nobody described a clean replacement story. The through-line is a workflow rewrite: faster middle, heavier edges, more judgment."],"voices":["Bill Warren","Graven Prest","Victor Ginelli","Adam Kerpelman","Justice Conder","Cj Miller"]}},{"index":6,"kind":"hero","encounterIndex":4,"encounter":{"eventId":67,"guestName":"Cj Miller","slug":"cj-miller","startsAt":"2026-06-16T16:30:00.000Z","heroArchetype":"Monk","spriteSlug":"monk","hook":"When production gets cheap, the premium is the human edit.","highlights":[{"text":"AI-assisted production still needs a human gate for taste and final cut.","source":"summary"},{"text":"Speed without editorial judgment produces volume, not signal.","source":"summary"}],"tags":["creative-work","human-edit","judgment"],"media":{"recapYouTubeId":"lGj9YsVIruw","fullInterviewURL":null},"links":{"eventURL":"https://portal.raidguild.org/events/67","posts":[{"title":"The Premium Is the Human Edit","url":"https://portal.raidguild.org/posts/fireside-cj-miller-premium-human-edit"}],"shared":[{"title":"TMCO","url":"https://tmco.io/"}]},"bio":"Creative producer focused on editorial craft — the human edit as premium when AI accelerates production."}},{"index":7,"kind":"hero","encounterIndex":5,"encounter":{"eventId":65,"guestName":"Andrej Berlin","slug":"andrej-berlin","startsAt":"2026-06-17T17:00:00.000Z","heroArchetype":"Paladin","spriteSlug":"paladin","hook":"AI can shorten alignment, but it cannot choose the taste.","highlights":[{"text":"Community tools may need to matchmake contributors, not just host chat.","source":"summary"},{"text":"Alignment work compresses with AI; aesthetic and values choices do not.","source":"summary"}],"tags":["community","alignment","taste"],"media":{"recapYouTubeId":"0k59966ivMw","fullInterviewURL":null},"links":{"eventURL":"https://portal.raidguild.org/events/65","posts":[{"title":"AI Can Shorten Alignment, But It Cannot Choose The Taste","url":"https://portal.raidguild.org/posts/ai-can-shorten-alignment-but-it-cannot-choose-the-taste"}],"shared":[{"title":"Deep Work Network","url":"https://deepwork.studio/"},{"title":"Andrej Berlin X","url":"https://x.com/kischiman"},{"title":"Andrej Berlin LinkedIn","url":"https://de.linkedin.com/in/andrejberlin"},{"title":"Andrej Berlin Substack","url":"https://andrejberlin.substack.com/"}]},"bio":"Community and alignment builder exploring taste, contributor matchmaking, and tools beyond chat rooms."}},{"index":8,"kind":"cross-cut","afterEncounterIndex":5,"question":{"id":"useful-tools","insertAfterEncounterIndex":5,"question":"What tools, prompts, agents, or processes are actually useful?","eyebrow":"Cross-interview synthesis","body":["Useful patterns were more about process than magic prompts. Spencer's doer/reviewer pipeline — coordinate, implement, review, repeat, with deterministic pre-push checks — was the clearest agentic coding pattern. Hunter praised Codex for reliable full-stack and browser-facing workflows. Victor emphasized small CLIs and command surfaces agents can actually run. Travis argued agent readiness starts with repo cleanup, not a better prompt.","Kerp got mileage from AI as a patient tutor and from course infrastructure like Canvas-integrated tooling, not from one omnibus model call. Jake and Sara both pointed at narrow, domain-reviewed agents: support coworkers and ad-ops QA auditors rather than general copilots. Elco's lesson was orchestration with observable shared state, not more autonomous actors.","What did not earn trust: one-shot frontend polish, unreviewed support automation, giant-context models without real project awareness, and all-purpose agents asked to own entire workflows. The useful layer is bounded tools, review gates, and repeatable internal interfaces."],"voices":["Spencer Graham","0xHunter","Victor Ginelli","Travis McCutcheon","Jake Winckowski","Sara Brown","Elco"]}},{"index":9,"kind":"hero","encounterIndex":6,"encounter":{"eventId":58,"guestName":"Victor Ginelli","slug":"victor-ginelli","startsAt":"2026-06-17T22:00:00.000Z","heroArchetype":"Ranger","spriteSlug":"ranger","hook":"The new senior dev job is reviewing the machines — building CLIs and wrappers agents can run.","highlights":[{"text":"Trust in AI coding built gradually; early output needed as much verification as manual work.","source":"summary"},{"text":"The middle of a project is easier now; edges like deploy, infra, and governance remain human-heavy.","source":"summary"},{"text":"Small command surfaces and internal CLIs become high-leverage agent interfaces.","source":"summary"}],"tags":["developer-workflows","review","cli","orchestration"],"media":{"recapYouTubeId":"Ds4EZ4Jpt9I","fullInterviewURL":"https://youtu.be/Ds4EZ4Jpt9I"},"links":{"eventURL":"https://portal.raidguild.org/events/58","posts":[{"title":"The Middle Is Easy Now. The Edges Are The Work.","url":"https://portal.raidguild.org/posts/the-middle-is-easy-now-the-edges-are-the-work"}],"shared":[{"title":"Victor Ginelli GitHub","url":"https://github.com/youngkidwarrior"},{"title":"Victor Ginelli X","url":"https://x.com/send2vic"},{"title":"SHE","url":"https://she.energy/"}]},"bio":"Experienced developer moving from hand-coding toward orchestrating AI coworkers and CLI surfaces agents can run."}},{"index":10,"kind":"hero","encounterIndex":7,"encounter":{"eventId":69,"guestName":"0xHunter","slug":"0xhunter","startsAt":"2026-06-23T16:00:00.000Z","heroArchetype":"Hunter","spriteSlug":"hunter","hook":"Context transfer is the new agency bottleneck — teams still copy-paste between agent sessions.","highlights":[{"text":"Custom software gets cheaper; medium businesses may own more of their stack.","source":"summary"},{"text":"Codex-heavy workflows work when scope is clear; frontend QA still needs human review.","source":"summary"},{"text":"Scoped context sharing between clients, projects, and teammates remains unsolved.","source":"summary"},{"text":"Some clients now open PRs on internal tools that the consultancy reviews.","source":"summary"}],"tags":["context-transfer","consulting","self-hosting","codex"],"media":{"recapYouTubeId":"Y-Xlv84NbTA","fullInterviewURL":"https://youtu.be/iTCxYD5pM4A"},"links":{"eventURL":"https://portal.raidguild.org/events/69","posts":[{"title":"Context Transfer Is The New Agency Bottleneck","url":"https://portal.raidguild.org/posts/context-transfer-is-the-new-agency-bottleneck"}],"shared":[{"title":"DDLS Consulting","url":"https://ddls.consulting/"},{"title":"hunter.cafe","url":"https://hunter.cafe/"}]},"bio":"Founder of DDLS Consulting — custom software, self-hosting, and Codex-heavy workflows for operational clients."}},{"index":11,"kind":"cross-cut","afterEncounterIndex":7,"question":{"id":"broken-risky","insertAfterEncounterIndex":7,"question":"What still feels broken, risky, or confusing?","eyebrow":"Cross-interview synthesis","body":["Context transfer was the loudest shared pain. Hunter said teams still copy-paste between agent sessions; Spencer worried about atomized developers and thinning shared context; Elco named shared state as the hard part of orchestration. Everyone had a version of: private agent loops work locally, but the team cannot see or inherit the process.","Verification showed up everywhere in different costumes. Frontend QA remains distrusted without human review. Kerp described proxy collapse in education — assignments that no longer prove understanding. Sara wanted audited checks instead of blind generation in operations work. Graven and Bill both warned that speed invites bloated products without judgment gates.","Model orchestration is still guesswork for power users: which model for which task, when to escalate, how to preserve memory across machines and tools. Andrej and Victor added social risk — coordination, taste, governance, and the edges of projects (deploy, infra, maintenance) remain stubbornly human-heavy."],"voices":["0xHunter","Spencer Graham","Elco","Adam Kerpelman","Sara Brown","Graven Prest","Victor Ginelli"]}},{"index":12,"kind":"hero","encounterIndex":8,"encounter":{"eventId":68,"guestName":"Jake Winckowski","slug":"jake-winckowski","startsAt":"2026-06-23T17:00:00.000Z","heroArchetype":"Healer","spriteSlug":"healer","hook":"The first useful AI support agent is a coworker, not a replacement.","highlights":[{"text":"Reviewed automation in support workflows beats magic all-purpose agents.","source":"summary"},{"text":"Domain-specific tooling and human checkpoints matter more than generic prompts.","source":"summary"}],"tags":["support-workflows","coworker-agent","qa"],"media":{"recapYouTubeId":"nzEe5-Rm3Ic","fullInterviewURL":"https://youtu.be/GbUmUQtjli4"},"links":{"eventURL":"https://portal.raidguild.org/events/68","posts":[{"title":"Reviewed AI Support Agents Start As Coworkers","url":"https://portal.raidguild.org/posts/the-first-useful-ai-support-agent-is-a-coworker-not-a-replacement"}],"shared":[{"title":"Jake Winckowski LinkedIn","url":"https://www.linkedin.com/in/jakewinckowski"},{"title":"HackerOne","url":"https://www.hackerone.com/"}]},"bio":"Builder of reviewed support workflows — useful AI agents start as coworkers with human checkpoints."}},{"index":13,"kind":"hero","encounterIndex":9,"encounter":{"eventId":52,"guestName":"Graven Prest","slug":"graven-prest","startsAt":"2026-06-24T17:00:00.000Z","heroArchetype":"Druid","spriteSlug":"druid","hook":"Product bottlenecks are moving — speed and ambition rose, but testing and should-we decisions got harder.","highlights":[{"text":"Small teams can ship overnight what used to take a week — bottlenecks shift to QA and judgment.","source":"summary"},{"text":"Public goods and builder-relations work still needs distribution and production readiness.","source":"summary"},{"text":"Maybe yes we could, but should we? becomes the recurring product question.","source":"summary"}],"tags":["product-management","public-goods","testing","bottlenecks"],"media":{"recapYouTubeId":"m02tfKU3azw","fullInterviewURL":"https://youtu.be/U-9XKvwsNrU"},"links":{"eventURL":"https://portal.raidguild.org/events/52","posts":[{"title":"Product Bottlenecks Are Moving","url":"https://portal.raidguild.org/posts/product-bottlenecks-are-moving"}],"shared":[{"title":"Flow State Network","url":"https://flowstate.network"},{"title":"Graven Prest X","url":"https://x.com/GravenPrest"},{"title":"Graven Prest LinkedIn","url":"https://www.linkedin.com/in/gravenprest"},{"title":"Octant Dashboard","url":"https://octant.app"}]},"bio":"Product and builder-relations work around public goods, Octant, and Flow State. Claude Code in the terminal."}},{"index":14,"kind":"cross-cut","afterEncounterIndex":9,"question":{"id":"builders-should-learn","insertAfterEncounterIndex":9,"question":"What should other builders learn from your experience?","eyebrow":"Cross-interview synthesis","body":["The cohort advice was practical, not futurist. Clean up repos and docs before expecting agents to be dependable. Build review into the workflow — second model, human checkpoint, oral exam, QA auditor, or coworker-not-replacement support pattern. Keep scope small enough that someone with taste can still say no.","Invest in interfaces agents can use: CLIs, narrow commands, internal tools, and explicit handoffs beat giant prompts. Treat relationship and organizational context as infrastructure, not chat history. For teams, shared memory and scoped context matter as much as model choice.","Several guests pushed back on hype. Friction can be a moat. Open source may strengthen if bug-finding is paired with repair loops. Community tools may need to matchmake contributors, not just host more chat. The lesson for other builders: AI amplifies whatever process you already have — strong or weak."],"voices":["Travis McCutcheon","Spencer Graham","Jake Winckowski","Justice Conder","Andrej Berlin","Victor Ginelli"]}},{"index":15,"kind":"hero","encounterIndex":10,"encounter":{"eventId":66,"guestName":"Travis McCutcheon","slug":"travis-mccutcheon","startsAt":"2026-06-24T20:30:00.000Z","heroArchetype":"Cleric","spriteSlug":"cleric","hook":"Agent readiness starts with cleanup — not magic prompts.","highlights":[{"text":"Repos and docs need hygiene before agents can operate reliably.","source":"summary"},{"text":"Realistic automation gates beat aspirational full-agent replacement.","source":"summary"}],"tags":["agent-readiness","cleanup","operations"],"media":{"recapYouTubeId":"3W6HqjzB6TE","fullInterviewURL":"https://youtu.be/ReRgmFTJtQE"},"links":{"eventURL":"https://portal.raidguild.org/events/66","posts":[{"title":"Agent Readiness Starts With Cleanup, Not Magic","url":"https://portal.raidguild.org/posts/agent-readiness-starts-with-cleanup-not-magic"}],"shared":[{"title":"missions.md","url":"https://missions.md"},{"title":"Earth2Travis Farcaster","url":"https://farcaster.xyz/earth2travis"},{"title":"Earth2Travis X","url":"https://x.com/earth2travis"}]},"bio":"Operator focused on repo hygiene, cleanup, and realistic gates before agents can do reliable work."}},{"index":16,"kind":"hero","encounterIndex":11,"encounter":{"eventId":76,"guestName":"Bill Warren","slug":"bill-warren","startsAt":"2026-06-25T16:30:00.000Z","heroArchetype":"Tavern Keeper","spriteSlug":"tavern-keeper","hook":"After the prototype, judgment does the work — PRDs compress into testable builds fast.","highlights":[{"text":"Brain dump → PRD scaffold → agent prompts → prototype is the new product loop.","source":"summary"},{"text":"AI can bloat products unless someone keeps scope simple and the experience delightful.","source":"summary"},{"text":"Bullish on democratized creation; cautious about hype cycles and AGI timelines.","source":"summary"}],"tags":["product-workflows","prototyping","judgment","scope"],"media":{"recapYouTubeId":"mYHv8Dcjb64","fullInterviewURL":"https://youtu.be/Fi28v2nRzTE"},"links":{"eventURL":"https://portal.raidguild.org/events/76","posts":[{"title":"After the Prototype, Judgment Does the Work","url":"https://portal.raidguild.org/posts/after-the-prototype-judgment-does-the-work"}],"shared":[{"title":"Bill Warren LinkedIn","url":"https://www.linkedin.com/in/bill-warren-product"},{"title":"Bill Warren X","url":"https://x.com/0xBillW"},{"title":"Summon","url":"https://www.summon.xyz/"},{"title":"Game7","url":"https://game7.io/"}]},"bio":"Product lead with Protocol Labs background — brain dump to PRD to prototype is the new loop."}},{"index":17,"kind":"cross-cut","afterEncounterIndex":11,"question":{"id":"future-outlook","insertAfterEncounterIndex":11,"question":"Where do you see the future — optimistic or doomer?","eyebrow":"Cross-interview synthesis","body":["Most firesides ended with a futures question, and the answers were mixed without being nihilistic. Bill was bullish on democratized creation but cautious about AGI hype cycles and cultural fallout from layoff narratives. Spencer rejected \"open source is dead\" and argued security could improve if discovery is paired with repair. Kerp pushed back on simple job-collapse stories while acknowledging real student anxiety.","Optimism clustered around leverage for small teams, cheaper custom software, better tutoring, faster alignment on prototypes, and stronger OSS repair loops. Doom-leaning notes clustered around atomized work, lost shared context, proxy collapse in institutions, frontend trust gaps, and vendors overclaiming autonomy.","The balanced take across sessions: roles shift rather than vanish; taste, judgment, distribution, and verification become the durable professions inside the stack. The future is not one camp. It is a rearrangement where builders who can coordinate agents, review output, and preserve human context do better than builders who only type faster."],"voices":["Bill Warren","Spencer Graham","Adam Kerpelman","Graven Prest","0xHunter","Andrej Berlin"]}},{"index":18,"kind":"hero","encounterIndex":12,"encounter":{"eventId":59,"guestName":"Sara Brown","slug":"sara-brown","startsAt":"2026-06-25T19:00:00.000Z","heroArchetype":"Archer","spriteSlug":"archer","hook":"The first useful ad ops agent is a QA auditor — domain judgment beats generic automation.","highlights":[{"text":"Programmatic ad workflows need audited checks, not blind generation.","source":"summary"},{"text":"Domain-specific QA agents outperform all-purpose copilots in operations work.","source":"summary"}],"tags":["ad-ops","qa","domain-agents"],"media":{"recapYouTubeId":null,"fullInterviewURL":null},"links":{"eventURL":"https://portal.raidguild.org/events/59","posts":[{"title":"The First Useful Ad Ops Agent Is A QA Auditor","url":"https://portal.raidguild.org/posts/first-useful-ad-ops-agent-qa-auditor"}],"shared":[{"title":"Sara Brown LinkedIn","url":"https://www.linkedin.com/in/sara-b-876693120/"}]},"bio":"Ad ops practitioner building domain QA agents — audited checks beat blind generation in programmatic workflows."}},{"index":19,"kind":"hero","encounterIndex":13,"encounter":{"eventId":78,"guestName":"Josh Tyson","slug":"josh-tyson","startsAt":"2026-06-26T18:30:00.000Z","heroArchetype":"Rogue","spriteSlug":"rogue","hook":"Creative labor after LLMs — synthesis and taste over raw drafting, with agent-first design and stopping rules.","highlights":[{"text":"Writer value shifted to rumination, research, and synthesis — not competing on raw drafting volume.","source":"summary"},{"text":"Individual AI productivity is often wasted without organizational source-of-truth and scoped data access.","source":"summary"},{"text":"Agent-first experiences and trusted personal agents may mediate how people interact with other systems.","source":"summary"},{"text":"When everything is possible, the production skill is knowing when to stop — scope creep and token maxing dilute the work.","source":"summary"},{"text":"Built the Fuzzy novel digital companion on OneReach.ai — an agentic reader experience, not just a static ebook.","source":"summary"}],"tags":["agent-first-design","creative-labor","source-of-truth","scope-creep","content-workflows","onereach-ai","agentic-companion"],"media":{"recapYouTubeId":"_YVeBGhbY_s","fullInterviewURL":null},"links":{"eventURL":"https://portal.raidguild.org/events/78","posts":[{"title":"The Hard Part Of AI Production Is Stopping","url":"https://portal.raidguild.org/posts/the-hard-part-of-ai-production-is-stopping"}],"shared":[{"title":"UX Magazine Podcasts","url":"https://uxmag.com/podcasts"},{"title":"Josh Tyson LinkedIn","url":"https://www.linkedin.com/in/josh-tyson-1132a27"},{"title":"Josh Tyson UX Magazine","url":"https://uxmag.com/contributors/josh-tyson-ux-magazine"},{"title":"Age of Invisible Machines","url":"https://onereach.ai/invisible-machines/"}]},"bio":"UX Magazine editor and co-author of Age of Invisible Machines. Rebuilt the magazine with Cursor and Claude; built the Fuzzy novel companion on the OneReach.ai platform."}}]}}