AI Is Changing WordPress Agencies Faster Than We Think  🔗 **Source:** [AI Is Changing WordPress Agencies Faster Than We Think](https://www.youtube.com/watch?v=E4EoTMExhaw) > [!note]+ Document Notes > tags: [ai-agencies, wordpress-agencies, value-based-pricing, agency-hiring, ai-tool-sprawl] > author: Toby Cryns and Kurt von Ahnen (The WP Minute, Agency Action) > source: WP Minute Podcasts (YouTube) > reference: https://www.youtube.com/watch?v=E4EoTMExhaw > date created: 2026-10-01 13:48 > > > [!info] Generated analysis — not the author's words. Sections below are machine-written > > from the captured source. Only text inside `> [!quote]` blocks is verbatim from the author. > > Do not quote or attribute any other line to the named author. > > # 🏷️ Title: AI Is Changing WordPress Agencies Faster Than We Think > <!-- deep-read:sections start --> ## 🧭 Table of contents - [📝 Summary](#-summary) - [📖 Main content](#-main-content) - [🗒️ Notes](#️-notes) - [📚 References](#-references) - [🗣️ Paraphrase](#️-paraphrase) - [📰 Newsletter](#-newsletter) - [✅ Action items](#-action-items) - [🧪 Hypotheses](#-hypotheses) - [📊 Methodology](#-methodology) - [❓ Questions](#-questions) - [🔮 Future research](#-future-research) - [💡 Implications](#-implications) - [🧠 Learning styles assessment](#-learning-styles-assessment) - [📊 Accessibility assessment](#-accessibility-assessment) - [🌍 DEIB assessment](#-deib-assessment) - [✅ Logical validity](#-logical-validity) - [🛡️ DARVO analysis](#️-darvo-analysis) - [🎯 Conclusion](#-conclusion) - [🔄 Refactor appendix](#-refactor-appendix--ai-is-changing-wordpress-agencies-faster-than-we-think) ## 📝 Summary - **What happened.** WP Minute Podcasts published a 40:49 episode of its Agency Action show on 2026-09-29. Two agency owners talk through how AI tools change WordPress agency work: one rebuilt a personal test site without a page builder using Claude, both describe clients arriving with AI-built demos, and most of the second half covers pricing, hiring, accountability and what "success" means for a growing agency. The transcript runs 00:00:00 to 00:40:09 in 57 paragraphs. - **Names and captions.** The auto-captions garble the hosts' names as "Toby Crimes" and "Kurt Vonan." The show-notes page spells them Toby Cryns and Kurt von Ahnen, and this analysis uses those spellings in prose. Inside `> [!quote]` blocks, the transcript spelling stays. Near the close, the caption spells the surname "C R Y NS" letter by letter, which matches the page's "Cryns." - **Speaker labels.** The transcript has no speaker labels. Where this analysis attributes a point to one host, it infers that from self-references: the owner of "the Mighty Mo" is read as Toby Cryns, and the automotive-industry veteran who runs the "mñana" business and sells LifterLMS-based sites is read as Kurt von Ahnen. Treat each attribution as an inference, not a caption fact. - **The claim.** AI is compressing project hours (20 to 40 hours down to 3 to 5 in one client example), clients are building demos themselves, and the surviving agency value sits in judgment, consulting and value-based pricing, not hours. Both hosts say outright that the outcome is unknown. - **Why it matters.** The episode is first-hand practitioner testimony about pricing and hiring pressure at small WordPress agencies. It contains no survey, dataset or benchmark. - **Key stakeholders.** Small WordPress agency owners, their developers and project managers, marketing teams building AI apps, and WordPress site-builder vendors such as Elementor (named as a subscription one host decided not to renew). - **Risks and tradeoffs.** Every figure is anecdotal (400,000 Yoast entries removed, 47 duplicate episodes, a 10-hour project done in one prompt, $300 as an example price). Both hosts run services businesses and may favor "expertise" framings. A WP Engine link appears on the show-notes page, but the transcript never mentions WP Engine and no sponsor read occurs, so any sponsor relationship is not established by this source. - **Bottom line.** The strongest content is practical and candid: the pricing argument (price the outcome, not the hours), the judgment-versus-task hiring problem, and the admission that growth made the agency harder to run and less profitable for a year. The forecast that AI in WordPress core "changes the whole game" is opinion, stated at 00:04:17 and 00:04:58 with no evidence. ## 📖 Main content **The core argument.** AI makes the code-and-configuration part of agency work cheap. One host connected Claude to an Elementor-built site and asked it to convert the site to a simple build with no page builder, and the transcript reports that it also cleaned out roughly 400,000 bloated Yoast entries caused by a podcast-feed plugin that re-added the same episodes (one show appeared "47 times"). The same host plans page-speed comparisons the next day and wonders if "cleaner, no page builder" conversion is a packageable offer. From there the conversation widens to a question both keep returning to: if projects shrink from days to hours, how does an agency make money? **Where each host lands.** - **Hours versus expertise.** Using a brake-job analogy (00:15:50), Kurt von Ahnen argues that an experienced technician still bills the book time even when the work takes 12 minutes, so agencies should move toward value-based pricing. Toby Cryns agrees in principle and admits to giving work away: a Gravity Forms invoice-ID plugin built by Claude in one prompt during a phone call, which the same host calls a 10-hour project two years ago. - **Consulting as the refuge.** Kurt von Ahnen reports clients who bring "pages and pages" of AI chat transcripts and now pay for a weekly meeting to discard noise. Toby Cryns is more guarded: a consulting practice is a different business from the one currently run, and asks what clients will pay for it. - **Tool sprawl.** AI chat assistants keep recommending extra tools and "pro versions," which leaves users with forgotten subscriptions. A client who forgot a SiteGround account is the example. Hosting that bundles licenses is offered as the counter-model. - **Hiring.** Both say the hard part is judgment, not task execution. Toby Cryns describes a year of slipping communication, expensive reactive hires and staff turnover. Kurt von Ahnen describes validating every change on a development URL before production and an unfinished CRM install with no sending address. - **Accountability and success.** Toby Cryns is building weekly yes/no metrics per role (for example, whether every client email got a reply within half a business day). Kurt von Ahnen defines success as staying at the current level of comfort while adding apprentices and revenue. **Terms.** - **Vibe coding.** Producing working software by prompting an AI instead of writing code by hand. Used at 00:04:17 without definition. - **MCP.** Model Context Protocol, the integration layer the host says is "already built in" to WordPress. The transcript doesn't explain it or say which MCP implementation is meant. - **Adaptive versus technical challenge.** Used at 00:37:58 to 00:39:25. An adaptive challenge has no final solution and needs ongoing change; a technical challenge can be solved by a process. Kurt von Ahnen frames agency training as technical, and Toby Cryns calls the business an adaptive challenge. **Contested or unverified in this source.** - The forecast that WordPress-based design, plugin and API-integration jobs "are going to face trouble" is opinion with no timeline or data. - "Most people who are marketers... are 100% fine with the generic designs that AI pumps out" is introduced as "a fact that might or might not be true, but this is my experience." It is a hypothesis, and the analysis treats it as one. - The Elementor replacement is described as working ("everything worked") but no before/after page-speed result exists yet, because the comparison was planned for the next day. - Kurt von Ahnen's caption renders the theme as "Astro theme" at 00:00:42 and "Astra" at 00:03:35. Given the context (the Gutenberg block editor, replacing a page builder), the intended name is almost certainly the Astra theme, not the Astro web framework. Verify before citing. **Show-notes cross-check.** The page's takeaways compared against the transcript: | Show-notes takeaway | Transcript support | Strength | |---|---|---| | AI shrinks projects from days to hours | 00:09:15 (20 to 40 hours down to 3 to 5); 00:15:08 (10-hour project in one prompt) | Supported by two anecdotes. "Days to hours" is the page's paraphrase. | | Sell expertise, not time | 00:09:15 ("consultative expertise"), 00:15:50 (brakes), 00:17:15 | Supported, argued by analogy. | | AI-generated sites are good enough for many clients | 00:17:15 to 00:17:58 | Anecdotal and flagged by the speaker as possibly untrue. | | Developers as customer-facing problem solvers | 00:06:25 to 00:07:48 | Supported. One agency is hiring for this and reports trouble finding it. | | AI creates tech debt through tools and subscriptions | 00:12:54 to 00:14:23 | Supported in substance. The phrase "tech debt" doesn't appear in the transcript; the speakers say "baggage," "extra tools," "extra subscriptions." | | Consultants get more valuable as clients are overwhelmed | 00:10:00 to 00:11:29 | One client example. Toby Cryns questions willingness to pay at 00:11:29 and 00:14:23. | | Value-based pricing | 00:16:32 (said in those words) | Supported. | | Hire for judgment | 00:23:40 | Supported, in the words "you can't really hire the judgment." | | Agency growth adds complexity | 00:20:06, 00:29:25 | Supported. Profits "are down" for a year despite higher totals. | | Clear accountability | 00:30:07 to 00:31:32 | Supported. Weekly objective metrics per role. | All ten takeaways trace to the transcript. The page overstates the evidence only where it presents anecdote as trend. ## 🗒️ Notes #### ⏱️ [00:00:00](https://www.youtube.com/watch?v=E4EoTMExhaw&t=0) Claude replaces an Elementor build - Opening exchange: one host has not renewed an Elementor subscription and used Claude connected to WordPress to convert an Elementor site to a build without a page builder. - Claude inspected the site, reported seven pages, flagged Yoast bloat, and with permission removed about 400,000 entries (00:02:09). - The cause was a podcast RSS plugin that kept re-adding episodes. One show sat on the site 47 times; the plan kept one copy of each (00:02:52). > [!quote] [00:01:26] > I'm already seeing the the very small jobs disappearing, you know, like or disappeared. - The other host counters that familiarity with the expected result gives confidence, and notes this was a personal test site with backups. #### ⏱️ [00:04:17](https://www.youtube.com/watch?v=E4EoTMExhaw&t=257) AI in WordPress core and the developer role - Results so far: static pages rebuilt in Gutenberg and the Astra theme, same copy except outdated parts, and Claude asked permission before each step. Page-speed comparison is planned for the next day (00:03:35). - Toby Cryns predicts Claude-in-WordPress arrives soon because AI writes the integration code. This is a forecast with no evidence offered. > [!quote] [00:04:17] > high level AI is coming to WordPress. Vibe coding is coming to WordPress. - Kurt von Ahnen still wants a human in the loop and asks what a custom WordPress developer is for now (00:05:41). - The Mighty Mo is hiring a WordPress developer who is customer-facing, and says it is unsure it needs a project manager (00:06:25). #### ⏱️ [00:08:32](https://www.youtube.com/watch?v=E4EoTMExhaw&t=512) Clients arrive with AI-built apps - Technicians in the car business want to work on the car, not talk to customers. Kurt von Ahnen asks whether custom developers resemble that (00:07:48). - Two clients that week showed marketing-team demos hosted on a vendor URL. One chose to keep the vendor URL and link to it. The other is "50/50" (00:08:32 to 00:09:15). > [!quote] [00:09:15] > But at that point, we're taking it from a 20 to 40 hour project to like a 3 to 5 hour project. - Kurt von Ahnen says consultative expertise still has to be sold, and describes clients who bring pages of AI chat logs and now use the agency as a filter (00:10:00 to 00:11:29). #### ⏱️ [00:12:12](https://www.youtube.com/watch?v=E4EoTMExhaw&t=732) Tool sprawl and forgotten subscriptions - A client call showed what a real build needs beyond the demo: a database for partial form entries, a transactional-email API and a GitHub repo. The client said no (00:11:29 to 00:12:12). - Toby Cryns asks whether marketing teams will do the building and leave agencies to port and connect it. The transcript calls that an open question. > [!quote] [00:12:54] > You end up with extra steps or extra tools or extra subscriptions to things - Kurt von Ahnen says chat assistants keep pointing to paid tools. Bundled licensing through hosting is the counter-model offered (00:12:54 to 00:13:38). - Example: a client forgot a SiteGround account she'd been paying for years (00:14:23). #### ⏱️ [00:15:08](https://www.youtube.com/watch?v=E4EoTMExhaw&t=908) Pricing: brakes, free work, value - Toby Cryns built a Gravity Forms invoice-ID plugin with one prompt while on the phone, for free, and calls it a 10-hour project two years ago (00:15:08). - Kurt von Ahnen's brake analogy: a master technician does the job in 12 minutes and still bills two hours. > [!quote] [00:15:50] > he can knock out those brakes in 12 minutes. He's still going to get paid two hours to put those brakes on. - Proposed answer: price the outcome. Example from the transcript: "it's going to be 300 bucks. Is is it $300 important to you?" (00:16:32). - Both admit it's often easier to just do the work than to sell it. LifterLMS scaffolding is the example of expert speed (00:17:15). - The "generic designs" claim and the Lovable anecdote of a three-year result beaten by a prompt follow at 00:17:15 to 00:17:58. #### ⏱️ [00:19:22](https://www.youtube.com/watch?v=E4EoTMExhaw&t=1162) Growth made it harder - Hosts split the job-loss question: "whole industries" have vanished, but coal and oil still exist in new places, so the outcome is unknown (00:18:41). - Toby Cryns reports a year of growth in clients and team, three open hires (designer, developer, project manager), and rising totals with falling profit. > [!quote] [00:20:06] > It's not simpler. It's not easier and profits are down. - Staff left when told their work would change. A hire was likely holding about 10 other jobs. Communication, more than work product, slipped (00:21:31 to 00:22:15). - Reactive hires were expensive and were chosen for urgency and organizing ability (00:22:15). #### ⏱️ [00:23:40](https://www.youtube.com/watch?v=E4EoTMExhaw&t=1420) Judgment, validation and apprentices - Kurt von Ahnen hands out tasks and validates everything on a development URL before production, which adds steps. > [!quote] [00:23:40] > you can hire for the task, but you can't really hire the judgment. - Failure examples: a FluentCRM Pro install with no sending address and no SMTP, and a Lovable form that "doesn't go anywhere" (00:24:24). - Standard operating procedures are recorded on video and stored as a hidden course in the agency's own LMS, so a contractor gets a lesson for each task (00:25:06). - Fear: local apprentices won't free up time for sales (00:25:49). #### ⏱️ [00:27:16](https://www.youtube.com/watch?v=E4EoTMExhaw&t=1636) Screening for something unmeasurable - Toby Cryns says the trait sought isn't cultural or geographic and can't be named objectively. A past front-desk job and its end explain wariness about asking staff for that dedication (00:25:49 to 00:27:16). - Kurt von Ahnen proposes a screening probe: sign into an interview five minutes late and see whether the candidate waits, messages or leaves (00:28:00 to 00:28:42). - Toby Cryns objects that one moment can have any cause, so the probe is noisy. Neither host claims a validated hiring method (00:28:42). - Scaling pain: three people was fine, five was fine, seven was a "nightmare." A project manager promoted to general manager is floated (00:29:25). #### ⏱️ [00:30:07](https://www.youtube.com/watch?v=E4EoTMExhaw&t=1807) Accountability metrics and when to cut - Toby Cryns names the missing ingredient and what is being built about it. > [!quote] [00:30:07] > What the Mighty Mo I think needs more of is clear accountability. - Per-role weekly yes/no metrics: every client email answered within half a business day (project manager); every task has an estimated delivery date and hours (developer) (00:30:51). - Missing these targets over one to two months is treated as a signal that a person is not the right fit (00:31:32). - Kurt von Ahnen raises the cut-loose timing problem, with a layoff story from Suzuki (00:31:32 to 00:32:57). Toby Cryns doesn't know the right answer, and says a wrong fit adds the wrong kind of headaches (00:33:40). - Mixed client and staff calls double the stress, especially over scope (00:34:23). #### ⏱️ [00:34:23](https://www.youtube.com/watch?v=E4EoTMExhaw&t=2063) How to define growth success - Toby Cryns asks how growth counts as success or failure. - Kurt von Ahnen recounts a recent offer from the automotive industry and the household's answer: do we need that? Success is defined as financial comfort without dependence on others to cover the house (00:35:05 to 00:35:49). - Goal: add apprentices and revenue "at a marked rate" without returning to a 60-hour week (00:35:49 to 00:36:32). #### ⏱️ [00:37:15](https://www.youtube.com/watch?v=E4EoTMExhaw&t=2235) Adaptive versus technical, and sign-off - Toby Cryns asks whether training four people over three to four months at 60 hours a week pays off, or whether to pull the plug. - Kurt von Ahnen replies that only the technical challenge should end; if the business stays in adaptive mode, process is missing (00:37:58 to 00:38:42). - Toby Cryns questions whether business operating systems like EOS deliver ever-easier operations, or whether goalposts always move (00:38:42). - Closing (00:39:25 to 00:40:09): training lessons go stale, but updating one is a bounded task. Contact paths named: LinkedIn, and the WP Minute Slack, which the speakers think is still free. The spoken Mighty Mo address is garbled as "themo.com". ## 📚 References The transcript itself contains no links; the following come from the video description or are primary sources for named entities. - [AI Is Changing WordPress Agencies Faster Than We Think](https://www.youtube.com/watch?v=E4EoTMExhaw): the source video, 40:49, uploaded 2026-09-29, channel WP Minute Podcasts, show Agency Action. - [WP Engine](https://wpengine.com/): the only link on the show-notes page, under "Important Links." The transcript never mentions it. Whether it is a sponsor, an affiliate or an unrelated link is not established by this source. - [The WP Minute](https://thewpminute.com/): the publisher whose Slack channel is named at 00:40:09. The transcript gives no URL for it. - [Claude](https://claude.ai/): the AI assistant used throughout. Anthropic's product, described in the transcript as connected to WordPress to edit a site. - [WordPress](https://wordpress.org/): the CMS under discussion, including the Gutenberg block editor (00:04:17). - [Elementor](https://elementor.com/): the page builder one host stopped renewing (00:00:00). The transcript says "no shade." - [Astra theme](https://wpastra.com/): the likely intended theme at 00:03:35 ("Astra"). The 00:00:42 caption says "Astro," which is a different project (a JavaScript web framework); verify the host's intent before citing either. - [Yoast SEO](https://yoast.com/): the plugin whose database entries were cleaned (00:02:09). - [Gravity Forms](https://www.gravityforms.com/): the form plugin in the invoice-ID example (00:15:08). - [LifterLMS](https://lifterlms.com/): the LMS plugin Kurt von Ahnen's agency builds on (00:17:15). The caption writes "Lifter LMS." - [FluentCRM](https://fluentcrm.com/): the CRM plugin in the failed-install example and the agency's "CRM of choice" (00:24:24, 00:36:32). - [Lovable](https://lovable.dev/): the AI app builder named at 00:17:58 and 00:24:24. - [SiteGround](https://www.siteground.com/) and [GoDaddy](https://www.godaddy.com/): the hosts in the forgotten-subscription anecdote (00:14:23). The caption writes "Site Ground." - [EOS Worldwide](https://www.eosworldwide.com/): the business operating system named at 00:38:42 ("Let's say EOS is one"). - [Model Context Protocol](https://modelcontextprotocol.io/): the "MCP stuff" at 00:04:17. The transcript doesn't name the protocol; this link is an inference from the abbreviation in an AI-and-WordPress context. - Unlinked: the Mighty Mo (Toby Cryns's agency), "mñana" (Kurt von Ahnen's business, per the caption "mñana nom site"), Perplexity, OpenAI, Grok. The transcript doesn't give a URL for any of them and none of them is analyzed here. ## 🗣️ Paraphrase Two agency owners compare notes on what AI does to their work. One used Claude to strip a page builder out of a neglected site and noticed the tool also removed roughly 400,000 junk database entries nobody had asked it to look for. That small win sets up the bigger worry: if a developer can finish a former 10-hour job in one prompt, and clients can build their own demos, what's left to sell? The answer they circle is expertise. A master mechanic bills book time for a job that takes minutes, so an agency should price the result and the risk, not the clock. They admit they don't always do that, because on a quick call it's easier to give the work away than to quote it. Clients drowning in AI-generated ideas pay for someone to say which ones to drop, though one host isn't sure that's a business they want to run. The second half turns inward. One agency owner says growth made things harder, not easier: three open roles, higher totals, lower profit, staff turnover and communication gaps. The recurring finding is that you can hire for tasks but not for judgment, so the fix being tried is a short list of yes/no weekly metrics per role and a recorded library of procedures. Success, for the other host, means staying at a comfortable level while adding apprentices without working 60-hour weeks. Neither host claims to have solved it, and both call the situation an ongoing, adaptive problem. ## 📰 Newsletter **Hours are shrinking. Judgment isn't.** Agency Action's latest episode, from the WP Minute podcast feed, gets honest about what AI does to a small WordPress agency's books. One host asked Claude to turn a neglected Elementor site into a lean build, and the AI cleaned out roughly 400,000 junk Yoast entries on the way. The other describes client marketing teams showing up with AI-built demos, turning 20-to-40-hour projects into 3-to-5-hour ones. The useful part for agency owners is the pricing and hiring candor. The brake-job analogy (a master tech bills two hours for a 12-minute job) makes the case for value-based pricing in one sentence, and the hosts own up to giving quick wins away for free. Then comes the unusually frank bit: one agency says growth has meant lower profit for a year, and the fix being tried is per-person weekly yes/no metrics. Plan on a 40-minute listen, with a lot of hiring talk after the 00:19:22 mark. Caveat: all numbers here are the hosts' own anecdotes, not survey data. Links: [the episode](https://www.youtube.com/watch?v=E4EoTMExhaw), [The WP Minute](https://thewpminute.com/), [WordPress](https://wordpress.org/). ## ✅ Action items #### DevRel - Developer advocacy: write one talk or post on "price the outcome, not the hours" using the brake-job analogy, with a worked example showing both prices for a 3-hour build. - Education: run a workshop on reviewing AI-generated sites before client delivery, using the transcript's examples (an unconfigured CRM and a form that goes nowhere). - Developer experience: add a pre-delivery checklist to any AI-and-WordPress quickstart covering sender address, SMTP, form destinations and a staging step. - Developer marketing: don't cite the 20-to-40 to 3-to-5 hour figure as a trend. It's one client's anecdote at 00:09:15. - Community management: collect agency owners' own project-hour before/after numbers in a single thread so claims like this have a sample size greater than one. - Internal/external/cross-project: ask both hosts what the Elementor-to-Gutenberg conversion cost in tokens and review time, since only the outcome is reported. #### OSPO - Policy and governance: no direct action from this source. Nearest relevance: write down who approves an AI tool's write access to a client site, since each step in the Claude session needed human permission (00:03:35). - License compliance: no direct action. Nearest relevance: the tool-sprawl story (00:13:38) is about forgotten subscriptions; inventory paid licenses an agency inherits from AI-recommended tools. - Security management: Claude connected to a live WordPress site can delete 400,000 rows. Define scoped credentials and a backup requirement before any such connection, as both hosts say they did. - Software inventory: no direct action from this source. Nearest relevance: each AI-recommended plugin or SaaS tool is an inventory entry with an owner. - Developer support: no direct action. Nearest relevance: document how to review AI-built plugins (the Gravity Forms invoice-ID example) before they go live. - Community engagement: no direct action from this source. - Strategic alignment: no direct action. Nearest relevance: the per-role weekly metrics at 00:30:51 are an example of objective, yes/no KPIs. #### Community - Maintainers: no direct action from this source. Nearest relevance: maintainers get AI-built pull requests from clients too, so review criteria matter. - Contributors: no direct action. Nearest relevance: the recorded-SOP-as-lesson model (00:25:06) maps to contributor onboarding videos. - Sponsors: no direct action from this source. - Ecosystem: no direct action from this source. #### WordPress - Core: no direct action from this source. The "AI is coming to WordPress" claim at 00:04:17 is a forecast, not a Core proposal. - Core Performance: the planned page-speed comparison after removing Elementor is a measurement worth reproducing. The result isn't in this episode. - Core AI: the "MCP stuff is already built in" claim at 00:04:17 needs a precise statement of what exists today. The transcript doesn't specify the implementation. - **Core Program Team:** no direct action from this source. - Mobile: no direct action from this source. - Playground: no direct action. Nearest relevance: a throwaway Playground site is a safe place to test an AI conversion before touching a live site. - Community: no direct action. Nearest relevance: meetup talks on agency pricing under AI would fit this discussion. - Training: no direct action. Nearest relevance: a lesson on migrating from a page builder to blocks, with a verification checklist. - Marketing: no direct action from this source. - Support: no direct action from this source. - Design: no direct action. Nearest relevance: the claim that marketers accept "generic designs" (00:17:15) is a design-quality question worth user research. - Documentation: no direct action from this source. - Photos: no direct action from this source. - TV: no direct action. Nearest relevance: an agency-owner panel on AI and pricing. - Polyglots: no direct action from this source. - Accessibility: no direct action. Nearest relevance: AI-generated "generic designs" accepted as good enough should still pass an accessibility check. - Meta: no direct action from this source. - CLI: no direct action from this source. - Hosting: no direct action. Nearest relevance: bundled licensing through hosting (00:12:54) is offered as an antidote to subscription sprawl. - Tide: no direct action from this source. - Openverse: no direct action from this source. - Themes: no direct action. Nearest relevance: the Astra theme was the rebuild target, per the caption at 00:03:35. - Plugins: no direct action. Nearest relevance: the Yoast bloat came from an automation plugin duplicating entries. #### The WPOCC - Programs: no direct action from this source. Nearest relevance: agency-owner peer groups could trade hiring and accountability practices. - Projects: no direct action from this source. - Funding experiments: no direct action from this source. ### FAIR - Technical independence: no direct action from this source. - Policies and covenants: no direct action from this source. - Productized governance artifacts: no direct action from this source. ## 🧪 Hypotheses 1. Agencies that move from hourly to fixed-outcome pricing for sub-10-hour AI-assisted tasks retain at least as much revenue per project as agencies that keep hourly billing. 2. When a client's AI-built demo needs a database, an email API and version control, the agency's completion work still lands at 20 hours or more, so the "3 to 5 hours" figure is limited to demos without backend needs. 3. Per-person weekly yes/no metrics (reply within half a business day, tasks with estimates) cut missed-communication incidents by half within one quarter. 4. Marketing-team demos hosted on a vendor URL are never promoted to production in at least a third of cases, as one client at 00:08:32 chose. 5. Agencies that bundle licenses into hosting report fewer forgotten-subscription incidents than agencies where clients buy tools directly. 6. Scheduled-interview "lateness probes" predict on-the-job initiative no better than chance. ## 📊 Methodology - **H1.** Survey 40 or more small WordPress agencies for per-project hours, price and billing model on matched project types over 12 months. Compare revenue per project and per hour across billing models. - **H2.** Collect 30 AI-assisted client builds, tag each as with or without backend requirements, record total agency hours from demo to launch, and report medians by group. - **H3.** Pilot the metrics at two agencies over 12 weeks, logging missed client emails and unestimated tasks weekly, with a pre-period baseline. - **H4.** Track demo hand-offs at several agencies and record whether each moves to production, stays on the vendor URL or is dropped. - **H5.** Compare subscription audits (count of paid, unused tools per client) between agencies that bundle licenses and those that don't. - **H6.** For a sample of hires, record the lateness-probe result and rate on-the-job initiative with a blind manager score at 90 days; test the correlation. - **Controls.** Pre-register thresholds, separate self-reported hours from invoiced hours, and note each agency's client mix, since the hosts themselves say their clients sit "on totally different spectrums" (00:09:15). ## ❓ Questions - What did the Elementor-to-no-builder conversion actually do to page speed once measured the next day? - Which MCP implementation does the claim "the MCP stuff is already built in" refer to? - What price did the invoice-ID plugin, built in one prompt, actually deserve under value-based pricing? - How many of the AI-built demos that marketing teams create reach production without agency help? - What does a client's willingness to pay for "noise filtering" look like in dollars per month? - What did the hire who was likely holding about 10 other jobs reveal about contractor vetting, and what changed in the process? - Does the weekly yes/no metric approach hold after the first quarter, or does it become box-checking? - If profits fell for a year while totals rose, which cost line moved most: salaries, tooling or rework? ## 🔮 Future research - **Pricing under compression.** No dataset yet compares value-based and hourly billing for AI-compressed WordPress work. A shared benchmark of hours-before and hours-after for common project types would replace anecdotes. - **Demo-to-production gap.** How often do AI-built client demos fail on database, email and deployment needs? The episode names the gap at 00:11:29 but doesn't measure it. - **Hiring signals for judgment.** The hosts say judgment is subjective and can't be screened directly. Validating any proxy (trial tasks, weekly metrics, probes) is open. - **Page-builder exits.** Quantified outcomes of AI-assisted migrations away from page builders (performance, cost, breakage) are absent from this source. - **Tool-sprawl cost.** The cost of forgotten subscriptions and abandoned AI-recommended tools per small business has no figure here. ## 💡 Implications - **Practice.** Agencies can test a pricing change by quoting one small AI-assisted job by outcome and comparing it to the hourly equivalent. The brake-job analogy gives the argument for the quote. - **Policy.** Human-approval gates (each Claude step needed permission, 00:03:35) are a useful default for any AI connection to a client site. Writing the approval rule down makes it auditable. - **Security.** AI with write access to a live site can remove hundreds of thousands of records. Backups and staging, which the host says are standard, are a precondition. - **Education.** Training material that treats hiring for judgment as a skill (observing questions asked, initiative shown) belongs in agency-owner education. - **Funding.** Not covered by this source. Nearest relevance: the host's recurring question "how do you make money when the whole world changes?" (00:04:17) is a business-model question funders of small agencies would face too. ## 🧠 Learning styles assessment - **Reading level estimate.** Flesch-Kincaid Grade Level about 5.8 and Flesch Reading Ease about 82 for the transcript text (8,879 words, 557 sentence units), computed with a syllable-count heuristic. Auto-caption punctuation makes sentence boundaries unreliable, so read both numbers as plain conversational speech, not a precise score. #### 👀 Visual learners - Present: none confirmable from the transcript. The video is a two-host conversation and the transcript mentions no screen share, chart or demo. - Improve: show the Elementor-before and Astra-after pages with page-speed numbers, and a one-slide pricing comparison (hourly versus outcome) for the brake example. #### 🔊 Aural learners - Present: the format is audio-first conversation with natural turn-taking and repeated restatement of key ideas. - Improve: add a 2-minute spoken recap at the end. The current close at 00:39:25 moves to contact details without summarizing. #### 🗣️ Verbal learners - Present: strong analogies (brake jobs, coal and oil, the master technician) and concrete stories (the SiteGround account, the unconfigured CRM). - Improve: define vibe coding, MCP, adaptive and technical challenge, and EOS on first use. #### 🧩 Kinesthetic learners - Present: one process is described step by step, the Claude-driven site conversion at 00:02:09 to 00:03:35. - Improve: link the prompts used and the staging and backup checklist so a viewer could repeat the conversion. #### 🧠 Logical learners - Present: clear cause-and-effect chains (AI shortens projects, so price on value; judgment can't be hired, so measure behavior) and a named objective-metric example at 00:30:51. - Improve: put the hours and prices in a table. Several numbers (400,000, 47, 20 to 40, 3 to 5, $300, 10 hours) appear in speech only. #### 🧑‍🤝‍🧑 Social learners - Present: two voices that disagree on points (consulting as refuge, whether the lateness probe works) and an invitation to comment or join the WP Minute Slack. - Improve: pose one discussion question at the start of each topic, so listeners can reply in comments. #### 🧘‍♀️ Solitary learners - Present: the episode asks implicit self-reflection questions ("What's your time worth to you?" at 00:19:22). - Improve: add reflection prompts in the description, such as "What would you charge for a 12-minute job?" #### 🌿 Naturalistic learners - Present: coal and oil are used as an analogy for industries that move instead of vanish (00:18:41). - Improve: no environmental or sustainability content appears. The energy cost of AI-assisted work isn't discussed. ## 📊 Accessibility assessment Judged as published: a YouTube video with auto-generated captions, no chapter list and a show-notes page. #### Screen reader semantics - Present: YouTube supplies the player and page landmarks. The show-notes page exists as text. - Improve: the video description carries no chapter list, which blocks quick navigation. WCAG 2.2 AA 2.4.5 (Multiple Ways) and 2.4.6 (Headings and Labels) benefit from chapters and descriptive headings. #### Keyboard navigation - Present: the YouTube player is keyboard operable with visible focus by platform default. - Improve: nothing in the content adds a barrier. 2.1.1 (Keyboard) and 2.4.7 (Focus Visible) are met by the platform. #### Contrast and readability - Present: not assessable from a transcript. A talking-head or audio-only presentation carries little text. - Improve: if on-screen text or lower-thirds exist, confirm 1.4.3 (Contrast Minimum) for them. #### Text resizing and reflow - Present: caption size is adjustable in the player. - Improve: any burned-in text won't resize, and 1.4.4 (Resize Text) and 1.4.10 (Reflow) fail for it. None is confirmed in this source. #### Media and captions - Present: captions exist, and a transcript was available to produce this analysis. - Improve: the auto-captions garble both hosts' names ("Toby Crimes," "Kurt Vonan"), "Cryns" (spelled out as "C R Y NS"), the business "mñana," the theme "Astro" versus "Astra," "LifterLMS" as "Lifter LMS" and "the Mighty Mo" address as "themo.com." That is a 1.2.2 (Captions, Prerecorded) gap in substance: captions exist but misspell every proper noun. There are no speaker labels, which makes turn-taking hard to follow in a two-host conversation. Human-corrected captions with speaker labels are the fix. #### Forms and interactions - Present: the episode asks viewers to comment or join Slack, both handled by platform forms. - Improve: the spoken contact details are garbled ("themo.com"). Put the correct URLs in the description as selectable text, which supports 3.3.2 (Labels or Instructions) on the destination. #### Standards reference - Primary gaps against WCAG 2.2 AA: **1.2.2** (inaccurate captions), **2.4.5** and **2.4.6** (no chapters, no descriptive headings in the media), and possibly **1.2.5** (audio description) if any on-screen content exists. Caption accuracy is the highest-impact fix. ## 🌍 DEIB assessment #### Cultural breadth - Present: examples span automotive, hosting, marketing and agency work, and both hosts mention prior careers outside WordPress. - Improve: all examples are small, English-language, US-style agencies. Add a non-US agency and a non-agency WordPress user to test whether the claims hold. #### Inclusive language - Present: no exclusionary terms in the transcript. Speech is conversational and second-person. - Improve: "nightmare" for team growth, "scammed" for a contractor hire and "Hey, ... fellas" at 00:10:00 are casual choices a written version would revise. The "scammed" claim (00:21:31) is stated as near-certainty about a contractor ("almost certainly") with no evidence given. #### Neurodiversity support - Present: topics are chunked: Claude experiment, AI in core, client demos, pricing, growth, hiring, accountability, success, close. Hosts restate key points. - Improve: no chapter markers, no on-screen outline, and long unstructured stretches (00:27:16 to 00:30:07 on hiring). Chapters and a short written agenda would help. #### Representation - Present: two voices, both owners of small agencies. Employees, contractors, clients and the marketing teams are described, never heard. - Improve: no one who is hired, managed or affected speaks. Add a staff or contractor perspective on the accountability metrics and the "hire for judgment" framing. #### Psychological safety - Present: hosts disagree politely, admit failures ("I'm not beating you up... I do it too") and acknowledge that their hiring process is unvalidated. - Improve: the lateness-probe idea and "cut them loose" discussion treat candidates and staff as variables. A viewer in a hiring role gets no guidance on fairness, bias or documentation. ## ✅ Logical validity #### Strong reasoning - **Analogical, 00:15:50.** The brake-job analogy (a master technician bills the same labor time for a 12-minute job) maps cleanly onto value-based pricing. It works because both settings have a market-standard book time. To clarify further: state whether web agencies have a comparable standard. - **Causal, 00:09:15.** If the client can build the demo, the agency's role shrinks from a 20-to-40-hour build to a 3-to-5-hour finish. The mechanism is specific. It rests on one client, noted as "50/50." - **Self-limiting honesty, 00:11:29 to 00:12:54.** Toby Cryns says outright that it "hasn't flushed out" whether agencies will end up porting work, and offers that "maybe my question prodding" gave the client the answer. Admitting the possible counter-reading raises credibility. - **Direct evidence, 00:20:06.** The report that profit is down while totals rose comes from the operator's own books, not a forecast. It is first-hand data, though unquantified. - **Specific measures, 00:30:51.** The yes/no metrics (reply within half a business day; estimate on every task) are testable and unambiguous, which makes the accountability argument checkable. #### Fallacies - **Prediction without evidence, 00:04:58.** "all those jobs are going to face trouble being real paying jobs." The forecast rests on the premise that AI will reach 80 percent of Claude's current capability inside WordPress, which is asserted, not shown. Fix: state the capability threshold and a date range, or label it a bet. - **Hasty generalization, 00:17:15.** "Most people who are marketers... are 100% fine with the generic designs that AI pumps out." The support is one co-working acquaintance and a Lovable story. The speaker does flag it as possibly untrue. Fix: cite client feedback counts. - **Anecdotal evidence, 00:08:32.** Two client calls in one week become a pattern ("how does that change the MightyMo if all our clients...") without a sample. Fix: track the share of clients with AI-built demos for a quarter. - **Survivorship framing, 00:18:41.** Coal and oil are "still pretty robust" is used to suggest industries adapt. Some industries disappeared too, as the same speaker says first. Fix: name the conditions under which an industry shifts versus ends. - **Appeal to personal experience in hiring, 00:28:00.** The lateness probe is justified by what it would show, but the speaker's co-host immediately notes alternative explanations. The transcript doesn't resolve this. Fix: validate any probe against later performance before using it. - **Undisclosed interest, throughout.** Both hosts sell agency services and explain why expertise is the durable value. The self-interest is visible but never named. Fix: state it at the start. #### Tone of voice - Casual, candid and peer-to-peer, with contractions throughout and frequent self-correction ("sorry no you go you go," 00:12:12). Both hosts mix business talk with personal stories (a doctor's-office phone call, a layoff, a front-desk job). The register is vulnerable at 00:20:06 ("it's been a struggle, man") and reflective at 00:35:49. Superlatives are rare. Predictions are often hedged with "I think" and "who knows" (00:04:17, 00:18:41). ## 🛡️ DARVO analysis Not applicable to this source. DARVO (Deny, Attack, Reverse Victim and Offender) describes a response pattern to accusation or criticism. This is a peer conversation with no accusation or accuser. Covered for completeness: **Deny**: nothing is denied. The hosts volunteer mistakes, including working for free, hiring a likely multi-job contractor and slipping client communication. **Attack**; the aside at 00:00:00 ("no shade to Elementor") heads off a possible attack on a vendor, and no critic is attacked. **Reverse victim/offender**; no such move appears. Toby Cryns's claim that staff left when expectations changed (00:21:31) is stated as fact without blame. ## 🎯 Conclusion The strongest content is practical and first-hand: the brake-job argument for pricing outcomes, the observation that hiring for judgment is harder than hiring for tasks, and a frank admission that growth made one agency harder to run and less profitable for a year. Those points come with specifics (weekly yes/no metrics, recorded SOP lessons, a 12-minute job billed at two hours) and the hosts flag their own uncertainty. The weak content is the forecasting. "AI is coming to WordPress" and the loss of design and plugin jobs (00:04:17 to 00:04:58) is opinion with no evidence, and the project-hour compression rests on two anecdotes. No result from the page-speed test is available because it hadn't happened yet. The WP Engine link on the show-notes page doesn't appear in the audio, so no sponsor relationship can be shown from this source. **Next steps for readers:** quote one small AI-assisted job by outcome this week and record hours, price and client reaction. Write down which tools an AI assistant has write access to on client sites, and add a staging and backup requirement. Pick two objective yes/no metrics per role and track them for a month before judging any hire. Treat the correct spellings (Toby Cryns, Kurt von Ahnen, Astra, LifterLMS) as the page and product sites give them, because the auto-captions garble them. ## 🔄 Refactor Appendix — AI Is Changing WordPress Agencies Faster Than We Think > Density: 2–5 Literature atoms (H3 pad), 0–3 Permanent candidates (H4 pad). > Emitted: 4 Literature, 2 Permanent. Resonance filter applied — the AI-in-WordPress forecast and the > coal-and-oil analogy were dropped as unevidenced or off-claim. > > Konik claim-header rules: present tense, declarative, ≤10 words, searchable, unique. > > Atomicity: one declarative sentence per atom. "and"/"also" → split. > > Tag pool: #bmf · #addendum · #xref · #articleseed · #storystem · #share · #aphorisms ### 📚 Literature Split Pad #### [H1] Experts bill for outcomes, not minutes > [!quote] [00:15:50] > he can knock out those brakes in 12 minutes. He's still going to get paid two hours to put those brakes on. **Context & use**: - 00:15:50, Kurt von Ahnen's analogy offered while Toby Cryns describes building a plugin for free in one prompt. - The analogy carries the pricing argument for the whole episode. Its limit: auto repair has a published labor-time standard and agency work usually doesn't. **Tags**: #xref #articleseed <!-- atomicity: 1 idea ✓ --> #### [H2] Hiring buys task skill, not judgment > [!quote] [00:23:40] > you can hire for the task, but you can't really hire the judgment. **Context & use**: - 00:23:40, said while describing validating every change on a development URL. - Follow-up evidence: an unconfigured CRM and a form with no destination (00:24:24). Pairs with the weekly-metrics approach at 00:30:07. **Tags**: #xref #aphorisms <!-- atomicity: 1 idea ✓ --> #### [H3] AI-built demos compress agency builds > [!quote] [00:09:15] > But at that point, we're taking it from a 20 to 40 hour project to like a 3 to 5 hour project. **Context & use**: - 00:09:15, after two client calls in one week. One client kept the vendor URL; the other was "50/50." - One agency's anecdote, not a trend. The backend needs raised at 00:11:29 (database, email API, GitHub) are the counterweight. **Tags**: #xref #share <!-- atomicity: 1 idea ✓ --> #### [H4] AI assistants multiply tools and subscriptions > [!quote] [00:12:54] > You end up with extra steps or extra tools or extra subscriptions to things **Context & use**: - 00:12:54, said while explaining why bundled licensing through hosting appeals to customers. - Evidence is anecdotal (the forgotten SiteGround account at 00:14:23). The show notes call this tech debt; the transcript's own words are "baggage" and "extra subscriptions." **Tags**: #xref #articleseed <!-- atomicity: 1 idea ✓ --> ### 🧠 Atomic Split Pad #### [P1] Price the outcome, not the hours **Claim** (one sentence, Courtney's voice): When tooling cuts the time a task takes, the price should follow what the result is worth to the client, not the hours spent. **Evidence pointer**: [[#H1]] · [[#H3]] **Scope & conditions**: Holds when the client can state what the result is worth and a comparable alternative exists. Breaks when the client can build the result themselves, or when scope is unclear enough that a fixed price causes disputes. Untested here: no host reports a completed outcome-priced project. **Links**: - [[Value-based pricing for agencies]] — because extends - [[Time-and-materials billing]] — because contrasts on what the client is buying - [[WordPress agency services]] — because applies-in small-agency pricing <!-- atomicity: 1 idea ✓ --> #### [P2] Judgment shows in behavior, so measure behavior **Claim** (one sentence, Courtney's voice): Because judgment can't be hired directly, a short list of objective weekly behaviors is the practical way to find out whether someone has it. **Evidence pointer**: [[#H2]] **Scope & conditions**: Holds when behaviors map to goals and are yes/no checkable, as with the half-business-day email reply and the task-estimate metrics. Breaks when metrics become the goal, or when the role needs initiative no metric captures. The hosts say they aren't sure the quality is objective. **Links**: - [[Hiring for judgment]] — because extends - [[Probationary trial tasks]] — because contrasts on when the signal arrives - [[Contractor onboarding SOPs]] — because applies-in agency training <!-- atomicity: 1 idea ✓ --> ### ✏️ Fleeting Queue - [F1] Confirm the correct spellings (Toby Cryns, Kurt von Ahnen) against the show-notes page before citing the episode; auto-captions give "Toby Crimes" and "Kurt Vonan" | https://www.youtube.com/watch?v=E4EoTMExhaw #fleeting #inbox - [F2] Check whether the 00:03:35 theme name is Astra, not Astro, by asking the host; the caption says "Astro" at 00:00:42 #fleeting #inbox - [F3] Look for the follow-up page-speed comparison the host planned for the day after recording (00:03:35) #fleeting #inbox - [F4] Open question: how many AI-built client demos reach production without agency help? The episode names the gap at 00:11:29 and doesn't measure it #fleeting #inbox - [F5] Article seed: "You can hire for the task, not the judgment" and per-role yes/no weekly metrics as a screening substitute #fleeting #inbox #articleseed ### → Things 3 (max 2 actions) - [ ] Verify the host names, theme name (Astra) and any WP Engine relationship from the show-notes page before citing this episode — research, 10 min #things - [ ] Draft one outcome-priced quote for a small AI-assisted job and compare it with the hourly equivalent — agency pricing test #things <!-- deep-read:sections end --> 📄 **Full text:** [[Full Document Contents/Articles/AI Is Changing WordPress Agencies Faster Than We Think-2.md|AI Is Changing WordPress Agencies Faster Than We Think by Toby Cryns and Kurt von Ahnen (The WP Minute, Agency Action)]] ## Highlights - [00:08:32] taking orders but kind of creating the story and the path and the workflow. Along these lines though, I had two calls this week from existing clients whose marketing teams built AI apps using cloud or whatever tooling they're using. And these are marketing people, not techies, but they built full, you know, fully functioning demos of the apps they want. And I had two, this is literally two calls this week, different clients. And the first one, I just got an email cuz I was like, I can take this across the finish line. It's sitting up at like claw.ai/app whatever. It's on some wonky URL. And I was like, I can take this across the finish line, integrate it with your website. Here's the price of that. The first one responded with, uh, we'll just keep it on this wonky URL and link to it from the site. Second one, I'm there's [00:09:15] 50/50 at this point. I just talked to them before this call. I'm like, they could do the same thing. And like at that point, they don't need me except maybe a few hours here and there to like maybe to tidy it up, you know, like to fix some API issue or or just to like do work that they don't have time to do. But at that point, we're taking it from a 20 to 40 hour project to like a 3 to 5 hour project. And I go, how does that change the MightyMo if all our clients are bringing in 3 to 5 hour projects versus 20 to 40hour projects? Yeah. There there's still that consultative expertise that's that's needs to be sold. So I have clients and and it's interesting because you and I have clients on like totally different spectrums. So, I have clients that are heavily involved in AI, but from a ([View Highlight](https://read.readwise.io/read/01m3w9jq21mm659ewhp6easr8d)) - Note: [H3] AI-built demos compress agency builds - Tags: atom/literature/h3 - [00:12:54] internally, but you know, if you're a marketing team at a, you know, $10 million company and you have a important intake form that is now know maybe maybe they figured out how to make it do because maybe my question proddding maybe I should have charged them to ask those questions, you know, like cuz maybe my questions gener maybe they took the notes from that and said, "Hey, AI, do what Toby said." It could have. It could have. I I found that it's that's that fine-tuning stuff that's still really really messy. You end up with extra steps or extra tools or extra subscriptions to things and you're like, why why do I have all this? Why do I have this baggage? It's like with the hosting that we provide. We provide a lot of licensing to different tools through the hosting. And so our customers aren't paying for the hosting [00:13:38] and six subscriptions, right? They're just they're just paying us. And I think teams that work with AI find themselves in these weird spots because before I started connecting AI directly to the WordPress website and working that way, I was trying to work in the AI chat window right in the browser. And it didn't matter if it was OpenAI, if it was Claude, if it was Perplexity, if it was Grock, it would always say, "Well, try this tool. Go to this website, download this thing. You probably need the pro version. Do this." And by the and you're like, "Okay." And you follow that and you go, "Well, that didn't do exactly what I wanted." But you've already given somebody else your credit card and signed up for something and now you've got a tool you forgot about because you're still building something. And I see a lot of people going down this pathway. I had a meeting with a client just today who didn't remember a ([View Highlight](https://read.readwise.io/read/01m3w9jsce44224gjspnxy45wy)) - Note: [H4] AI assistants multiply tools and subscriptions - Tags: atom/literature/h3 - [00:15:50] back to the car business. If I want to put brakes on the front end of a Ford Mustang, it's 2 hours in labor plus the cost of the brake pads. Boom. it is what it is. And then let's say that a new technician works on that car and it takes them 3 hours to put the brake pads on and resurface the rotors. It's still going to cost the customer 2 hours. But when they go to an experienced technician, right, when they go to that master tech that's got the giant toolbox and all the air tools and all the stuff, he can knock out those brakes in 12 minutes. He's still going to get paid two hours to put those brakes on. And that's where the hell the whole AI question in the agency space kind of cycles around that, right? It it's it's it's like, yeah, you whipped up something, you know, took a phone call, whipped something up, but and you did it [00:16:32] for free. And, you know, my heart goes that way, too. I tend to be real generous and help people on really quick things, but somewhere we have to recognize like there's a value to that. And we we have to go back to that value based pricing so that we can say, hey, you know, it might take someone else, I don't know, two days, three days a week to knock something like this out. I could probably get you a result by the end of the day, but it's going to be 300 bucks. Is is it $300 important to you? And then you have something to work with. I I And I'm not beating you up. Don't because I do it too, Toby. I I get on a call with somebody and sometimes it's easier just to do it than to sell it. But at the end of the day, you know, we still have to have the value for what we know. And that's where so like I'm in the Lifter LMS space. I can make a ([View Highlight](https://read.readwise.io/read/01m3w9jjdnh7v87qe08hjhx852)) - Note: [H1] Experts bill for outcomes, not minutes - Tags: atom/literature/h1 - [00:22:57] and this gets back to what we were talking about earlier. There are great developers out there who just want to be told what to do and they can deliver amazing stuff, but that's not exactly what we need. We need people who can, you know, I'm curious too about about your your hires. You know, h how how do you look at hiring when when you're even if it's contractors, how are do you generally hand them a to-do list or or do you kind of expect them to come with questions? >> So, it's it's been very very interesting, Toby. I'm in a different position just because of structure. And so I tend to give out tasks. I tend to be the guy that runs the agency, the guy that's the project manager, and then I tend to delve divvy out tasks on different projects. And then to your [00:23:40] point, it's like I think it's actually in one of your notes. It's like you you can hire for the task, but you can't really hire the judgment. You're trying to be able you're trying to develop people and pass the ability to use sound judgment in making decisions, but sometimes you just don't see it. And so what happens is we work on a development URL and then I got to validate and verify everything. And if everything meets my standard, then I say, "Okay, now we can push that to production, right?" It adds a lot of steps. And in some cases, like we talked about with with you making that tool for somebody, it's like sometimes it's easier just to do it myself and get it done than it is to divvy it out and then check on it later. But if I don't do the divvy it out and check on it later, I end up with weird things. I end up with like someone ([View Highlight](https://read.readwise.io/read/01m3w9jmrneyzq3xss4z5cvs1v)) - Note: [H2] Hiring buys task skill, not judgment - Tags: atom/literature/h2