Michael opens on a milestone he didn't expect to feel so strange: he's up to nine people, four of them in the last month, and for the first time his job is almost entirely infrastructure and people management. The individual contributor work is almost gone — except sales calls, which he's starting to find numbingly repetitive and knows need to be handed off. Matt's feeling the same fatigue on the other end of a late, no-show customer who turned a routine call into one of his worst sales performances in a year. Both are building toward something that doesn't require them in every seat — and starting to feel the gap between where they are and where that actually is.

On the marketing and sales side, Michael is deep in email marketing infrastructure — building a Klaviyo MCP server so Claude can write and push campaigns without him touching the interface — and is watching Meta ad fatigue eat into ROAS in real time while he's too buried in production to film new creative. Matt's EA Trish is now handling invoicing so well that month-end close happened the day after month end for the first time in Develop history, and he's down to under two hours a week in accounting. Matt also debuts AirDAM — an AI-powered digital asset management vault his team built so agencies and contractors always pull the right approved content, not the same four photos.

Engineering covers CoolantClear supply chain hitting a wall so hard that Michael is spinning up in-house production capacity on the idle spindles, a new Spindle Storm first article in production with triple the airflow, and a greenfield facility buildout proposal moving fast at Develop with multiple robots and parts showing up daily for Projects 2902 and 2903. The episode closes with a conversation both hosts keep circling back to: what happens when AI compute stops being subsidized by venture money — and whether the entire infrastructure Michael is building quietly becomes three people's salaries overnight.

Matt comes back from Automate brain dead but buzzing — the LED wall was a hit, the custom trade show software finally worked instead of fighting them, and podcast listeners stopped him in the hallway of the Airbnb. Michael returns from Maui with four new hires in the last month and a company that's somehow handling its busiest month ever with fewer things breaking than usual. Both are starting to feel something unfamiliar: the systems are actually working.

On the sales and marketing side, Matt officially announces the hire he teased last episode — Andrew, the account manager from his agency who was leaving, is coming on full-time July 13th to run top and middle of funnel. It's the commercial hire Matt has been circling for over a year. Michael, meanwhile, is watching ad creative fatigue eat into Meta ROAS in real time — campaigns that should be returning 4–10x are returning 2x — and knows exactly why: he hasn't filmed new content in weeks because sales volume has him buried. Both hosts go long on the middle of funnel problem — over a hundred stalled deals at Develop, a growing backlog at Gimbel — and what it actually takes to build reporting and follow-up systems that don't require the founder.

Engineering covers Spindle Storm getting a full product redo — triple the airflow, investment cast instead of billet, with free replacements going to every existing customer — and a client 1701 case study video dropping soon that Matt clearly can't stop smiling about. The episode closes with Gim Command's vendor portal vision, Develop's push toward NIST and ITAR compliance, SolidWorks PDM that Michael knows he needs and keeps not doing, and Matt's ten-year anniversary with Amanda.

Both hosts open on the same diagnosis: the sine wave problem. Michael keeps bouncing between B2B proposal mode and Shopify e-commerce focus, and whichever one he's in, the other declines. Matt's seeing the same pattern smooth out slightly as style guides and brand kits he built over three months start coming back to him from the team — nearly finished, almost no edits needed. When the tools work, the founder stops being the bottleneck. Both are trying to get there faster.

On the sales and marketing side, Matt teases a major hire he can't fully announce yet — a dedicated marketing and sales role that's been in the works quietly and is close to being inked. It's the move he's been circling for over a year, and it changes the commercial structure of Develop significantly. Michael, meanwhile, moves email marketing entirely from HubSpot to Klaviyo to break the contact list ceiling, debates whether to build email flows there or inside Gim Command before September, and gets talked off a bad IMTS booth partnership by Matt in real time. The two also go long on what Gimbel Automation actually is — CNC accessories ecosystem or something else — in one of the most strategic conversations the podcast has had.

Engineering covers CoolantClear 3 already in prototype — with a production prototype that solves every install issue from versions one and two — and a machine at a customer site approaching one million units processed in four months. Matt pulls data on what 98.5% versus 99.4% reliability actually does to customer perception, and both hosts dig into why customers obsess over cycle time while ignoring in-process probing that prevents crashes. The episode closes with Develop building toward running without Matt by end of 2027, Michael flying to Maui for a corporate retreat, and a frank admission that the number to sell either company would have to be so absurd it's basically not worth thinking about.

Matt opens with a gut punch: his agency account manager — the person carrying the entire team for two years — is leaving, and the contract runs through November. He's weighing a settlement negotiation, bringing marketing in-house, or both, and Michael doesn't hesitate with his position: for niche B2B manufacturing, agencies that can't tell a vise from a clamp aren't worth the training cost. It's the second person that day to tell Matt the same thing, and he's starting to listen.

Michael is running on fumes. The proposal generation tool is done — three to four hours down to ten minutes — and Gim Command is humming, but CoolantClear demand has completely outrun supply, and the mental load of managing a product growing that fast while hiring two assembly techs, finishing the configurator rollout, and trying not to burn out is taking a visible toll. The configurator suite is now complete across TumbleBlast, CoolantClear, and grippers — all with email capture built in and targeted drip campaigns being built behind each one.

Engineering covers Develop's largest group orientation in company history — electrical and mechanical engineers starting the same week — and an onboarding process that was compressed from three weeks to three days. Matt and Michael go long on vibe coding vs. real technical debt, whether B2B SaaS is actually going to collapse or just slow down, and how manufacturing is one of the last industries where Claude still surprises people on the shop floor. The episode closes with the best accounting meeting Matt has ever had in company history, Michael already prototyping CoolantClear 3 for fun, and both hosts landing on the same honest admission: entrepreneurs love the first 90% of every project and can't stand the last ten.

Matt opens the week buried in content review — 18 pages for Vac-Fixture, eight for Develop, two press releases, and a 12-email drip campaign audit, all pushed through in one week to hit the Automate deadline. It's good work, but he's sick of reading about his own company. Michael is dealing with a softer May on e-commerce after a record April and trying to figure out whether it's a content problem, a sales cadence problem, or just a lumpy business reality — then a second CoolantClear reel hits 1.1 million views and 30K in sales in 14 hours, which doesn't fully answer the question but does clarify the recipe.

On the marketing and sales side, the TumbleBlast 3D configurator is live and already eliminating four calls a day about cabinet compatibility. A CoolantClear compatibility checker and a full G-code kinematic simulator for 500 CNC machine models are both in progress — though Matt pushes back on priority order, and Michael concedes mid-conversation that he's doing the exciting thing again instead of the highest-leverage one. The HubSpot vs. Gim Command debate resurfaces with an actual bet: Michael thinks Matt drops HubSpot within 18 months, Matt takes the other side, and they both agree to let the business decide.

Engineering covers Develop's new CNC machinist starting his first real parts on day four — promising but emotions deliberately regulated — and a new peer review process being tested where junior engineers review senior work first for exposure before a senior signs off. Michael is getting pulled back into CoolantClear 2 electrical details he shouldn't be touching, and names the real bottleneck: his lead engineer is remote, and he needs a local one whether he wants to spend the money or not. The episode closes with a packaging foam machine that ROIs in a month, Matt's 90-day review process being handed to team leads for the first time, and a shared reflection on what it feels like when the business finally starts running without you in every seat.

Matt opens the week buried in content review — 18 pages for Vac-Fixture, eight for Develop, two press releases, and a 12-email drip campaign audit, all pushed through in one week to hit the Automate deadline. It's good work, but he's sick of reading about his own company. Michael is dealing with a softer May on e-commerce after a record April and trying to figure out whether it's a content problem, a sales cadence problem, or just a lumpy business reality — then a second CoolantClear reel hits 1.1 million views and 30K in sales in 14 hours, which doesn't fully answer the question but does clarify the recipe.

On the marketing and sales side, the TumbleBlast 3D configurator is live and already eliminating four calls a day about cabinet compatibility. A CoolantClear compatibility checker and a full G-code kinematic simulator for 500 CNC machine models are both in progress — though Matt pushes back on priority order, and Michael concedes mid-conversation that he's doing the exciting thing again instead of the highest-leverage one. The HubSpot vs. Gim Command debate resurfaces with an actual bet: Michael thinks Matt drops HubSpot within 18 months, Matt takes the other side, and they both agree to let the business decide.

Engineering covers Develop's new CNC machinist starting his first real parts on day four — promising but emotions deliberately regulated — and a new peer review process being tested where junior engineers review senior work first for exposure before a senior signs off. Michael is getting pulled back into CoolantClear 2 electrical details he shouldn't be touching, and names the real bottleneck: his lead engineer is remote, and he needs a local one whether he wants to spend the money or not. The episode closes with a packaging foam machine that ROIs in a month, Matt's 90-day review process being handed to team leads for the first time, and a shared reflection on what it feels like when the business finally starts running without you in every seat.

Both hosts open the week running on fumes but with cautious optimism — Matt's new EA is three weeks in and already taking real weight off his calendar, while Michael is back from his last planned in-person install and feeling the full weight of how much travel has been quietly killing his momentum and marketing output. The shift to Turnkey Direct is officially underway, IntraLoad is fully wound down, and Michael is now focused on building an ecosystem play where every product on the floor feeds recurring hardware orders for years.

On the sales and marketing side, Matt heads out on a full-day site visit to a large packaging company with half a dozen automation projects worth one to several million dollars each — and comes back focused on how Develop can become their fractional automation department rather than a one-time vendor. Michael confronts a stretch where deals just weren't closing, then three Turnkey Direct orders land within two days of each other, and he lands on the same diagnosis every time: his follow-up cadence and marketing output both fall apart when he's on the road. The HubSpot vs. Gim Command debate resurfaces — Matt's all in at $9,600 a year, Michael is counting down to cancellation.

Engineering covers Develop's largest single component spend in company history clearing in one month for its confidential flagship project, two more engineers starting June 1st to give the team some buffer back, and a candid admission that the customer's product is still slightly evolving mid-design. Michael's CoolantClear 2 saga continues — a motor change cascaded into a driver change cascaded into a new electrical enclosure — while demand on CoolantClear 1 keeps outrunning supply. The episode closes with a long, honest conversation about CI prioritization, P&L forecasting discipline, and the strange disorientation of becoming a manager who's now worse at the individual work that built the company in the first place.

Michael opens on the road — again — and makes a decision he's been circling for a while: IntraLoad is done, full-service in-person turnkeys are done, and the business is going all-in on product revenue and remote Turnkey Direct installs. The goal is scalability without him being the critical path. Matt's feeling a different kind of drain — not hours, but context switching. Negotiating contracts, building comp structures, prepping for Automate, and trying to find time for a deal pipeline audit that hasn't happened in seven weeks. Both hosts land on the same diagnosis: the grind doesn't get lighter, the weight just shifts.

On the marketing and sales side, Michael launches a 3D configurator with 300 machine models that lets customers visualize and add-to-cart a full automation package in one click — built on Replit, deployed via iframe, and now wrestling with an SEO tradeoff Matt pushes back on. They go deep on keyword strategy for machine-specific CNC tending pages and whether low-difficulty brand keywords are actually worth chasing. Matt, meanwhile, is seven weeks behind on a deal pipeline audit and actively designing his first sales hire — with a margin-based comp structure his roundtable helped him shape that ties commission to delivery, not just close.

Engineering covers Develop completing its full hiring wave — CNC machinist, two new engineers on June 1st — and a site visit to a multi-robot machine that's running beyond KPIs with operators onboarding in under a day. Michael's Coolant Clear 2 production prototype is wrapped, supply is being reshored, and a viral Instagram reel hit a million views in two minutes of filming. The episode closes on cash strategy — Matt's three-tier war chest structure and a frank conversation about what happens when you pour every dollar of margin back into R&D.

Michael opens with something he hasn't been able to say in a while — things are calming down. Passive revenue is growing, repeat customers are showing up, and he's starting to feel less like the linchpin of every transaction. Matt, meanwhile, is deep in financial planning for a line of credit renewal that has to account for two of the largest projects in Develop's history landing back to back in 90 days, with multi-million dollar material procurement timelines to match.

On the sales side, Develop closes one of its biggest deals to date — booked for Q3 to fit existing capacity — while Michael sells through his entire CoolantClear inventory and launches Rev 2 with domestic machining, an integrated solenoid, and a redesign that expands compatibility to Brother machines. The Spindle Storm billet chip fan is also weeks from launch at $129, going head to head with a $400 competitor. Matt credits Claude and connected HubSpot and QuickBooks data for building his most thorough bank model yet.

Engineering covers Develop's full mechanical team locked on the complex case packing machine through Q2, a FANUC training week for team members, and a growing facility question — add 12,000 square feet onto the current building, or move entirely? Michael wraps with a candid take on AI-driven ad optimization: out of roughly 100 SKUs, Google Shopping ads were only profitable on about a dozen. Turning the rest off nearly doubled ROAS overnight.

Matt opens the week with a closing runway feeling — Automate is on the horizon and there's a lot to do. He walks through Develop's most ambitious trade show setup yet: a 10x30 booth with a 10x20 LED wall, live HMI demos where attendees can actually run a machine, and a complete website overhaul cutting everything that isn't hyper-focused on automation. The website nearly tripled in traffic over the last six weeks — partly explained by an A3 and FANUC SI certification press release going out around the same time.

Michael's week is defined by one thing: Gim Command going live internally. The custom AI-built ERP has already replaced multiple SaaS subscriptions, moved all task and Kanban management out of Notion, and has Claude embedded directly into the system for user-level queries. He's 100-plus hours in, spending $3–4K a month on AI credits, and describes the experience as "transformative" — while Matt remains skeptical and reminds him that every software project he's ever seen runs long. Meanwhile, CoolantClear goes viral again with a million-view Instagram reel — and breaks the supply chain. The AI-caused Tumble Blast product ID deletion from last week wiped out Google Shopping ad learning worth tens of thousands in spend. Michael's reflection: AI is like automation — you can make a lot of parts fast, or you can make a lot of scrap fast.

The episode closes with a substantive exchange on CI discipline — Matt's WIP-cap system, budgeted CI pools, and why letting small improvements die on the vine is often the right call — and a fractional CFO update: two weeks in, already identifying changes, and told to stop asking for permission.