Source: /Users/relic/kreilkamp-audio/kreilkamp.txt (whisper-medium, no diarization).
Timestamps from kreilkamp.vtt. Total runtime approx 2:21.
Caveat: speakers attributed by content + turn-taking patterns + named call-outs. Where attribution is uncertain, marked unattributed or hedged. Whisper-medium garbles trucking jargon ("Sam Sarah" = Samsara, "Motiv" = Motive, "10th Street" = Tenstreet, "MV CSA" / "FM CSA" = FMCSA, "MVR" preserved, "Carl Camp"/"Girl Camp" = Kreilkamp). Transcript clearly degrades around 01:11 (line 1564 in .txt) — punctuation collapses, speaker-break >> markers stop, and from there speakers must be inferred almost entirely from content.
Luke leads with the Fathom anecdote (his AI agent caught his boss saying "no" without him hearing it). Lands well. Tim opens with the values frame: "we have in common… we are not looking to eliminate dispatchers or people." Luke matches with the "three friends, a few months into this, big AI companies will out-tech us, but relationships are untouchable" speech. Tone: warm, value-aligned. (Driven by both Tim and Luke equally.)
Tim walks Luke through their Power BI dashboards (revenue pay detail, odometer miles, daily dispatch, "Tim's report," rate per mile). Tim's at +30 cents over national average. Tim explains his use of Copilot ("she goes, can you put this through AI?" — he summarized a long document in 3 minutes). Luke seeds: confabulation warning + custom-agent-vs-chatbot distinction + the "scary downstream of agentic AI" line. Matt Hauncey (the Power BI consultant) introduced — Tim meets with him weekly, has a 4pm today. (Driven by Tim showing, Luke teaching/seeding.)
This is the section containing the iRecruit blow-up. Luke asks "from random fella on the street to in the seat making money." Jeff: leads come from Facebook, Indeed, Job Trucking. They PAY 3-4 lead vendors (CDL Life, Trucker's Report, Drive My Way, multi-carrier leads). Two of four vendors do prescreening. Then "they end up in our ATS." Luke says "Which is iRecruit?" — Jeff: "That is one of them." TIM (immediately): "If we're talking about drivers, we don't use iRecruit for drivers" (00:12:09). Jeff: "It's Tenstreet" (transcript: "10th Street"). Luke: "A bit of information that I got wrong." Tim: "I may have given that to you wrong." Smooth recovery. Jeff demos Tenstreet's lead-list view, "Pulse" (in-app messaging on driver tablets — NOT SMS). Numbers: 1,200-1,400 leads/mo; 1,924 outreach attempts in April; 192 phone interviews scheduled; ~1% lead→hire conversion. (Driven by Jeff.)
Jeff describes the 30-min phone screen content. 200 phone screens in April → 75 moved forward. After phone screen: MVR + Clearinghouse + PSP (pre-employment screening program) → Safety dept gates yes/no → offer → background check + drug test (72-hour window, no re-do) → orientation (Wisconsin, mostly fly or drive in) → 2-day orientation. Experienced drivers (3+ months) test out and can be earning revenue Friday after orientation; inexperienced go to 4-6 week finishing program. Cost difference "bigger than this building." Tenstreet already does AI fit-scoring (zero accidents, zero violations, etc.). Boomerang drivers come back faster. Phone-screen format would lose too much human-read if fully automated. Tim slips in: "I'd say it has to be done by a person, because I get broker phone calls and AI brokers — I hang up on them" — important Tim-stance data. (Driven by Jeff with Tim color commentary.)
Tim's question "if you had 40 qualified interested drivers, you could hire them tomorrow." Confirmed. Then the numbers expand: 285 trucks, ~275 available, 140 currently producing revenue, 180 is the benchmark, 40 empty trucks today. Turnover ~70% (Jeff/Josh note it's actually a retention number, but industry avg is 100%). 150-200 drivers/year is the hire target to keep up with attrition (00:34:34). Top of funnel ~18,000/year. They could pay for more leads but the recruiters can't reach them — "if we pay for more leads, we can't get to all of them." Luke crystallizes: the automation opportunity is shrinking the pre-screen interview + capturing missed calls because reach-rate is the bottleneck, not lead volume. Jeff mentions Paradox at his old job did exactly this (text-first pre-screen + scheduled callback). Critical Tim line: "I think when they call you from a truck stop at 9:30 at night and you don't answer, they call the very next name on their list. So 30-40% of missed phone calls of people who are interested, they go to somebody else" (and confirms this is Tenstreet's data). (Driven by Luke pulling, Jeff giving, Tim sharpening.)
Luke pitches: "as soon as the voicemail hangs up, a text goes out automatically with a mobile-friendly web form." Tim immediately pushes "why only after hours? we miss them in business hours too." Luke: agreed, every missed call. Discussion about the 1-800 number painted on trucks (general line) vs. the recruiting number (1-800-999-7112) — nobody on the team knows how many calls hit which number; "IT could tell us". Jeff's bounded-rule design feedback: less than 5 questions, less than 50 chars each, or experienced drivers self-select out (esp. on "years of experience"). The two-demographic insight: experienced drivers don't trust AI + don't like texting; younger drivers love it. Tim's "I'm a millennial, I hate texting" beat. Then Tim's $150,000 derail: when Luke wonders if anyone is shoring up driver-pain, Tim says "you could probably pay $150,000 a year and you would still have a [turnover problem]" — context is driver pay, NOT consulting pricing. Tim then catalogs the emotional drivers of turnover (dispatcher tone, mechanic yelled at him). (Driven by Luke design-pitching, all three pushing back on edges.)
Jeff: two ad agencies have call-center "matchmakers" doing follow-up texts on no-response leads (Jeff doesn't want OBB to outsource similarly — "I don't think we're going to use that"). 25% / 50% / 25% lead-vendor spend mix. Calendly used for self-schedule. "Tim, Josh were both saying, like, having that experienced driver gets them out the door so much quicker" — Josh confirmed in room. New (third) recruiter being added. (Driven by Jeff.)
Tim asks: can an agent predict which driver might have an at-fault accident based on Samsara data + log + history? Long discussion. Jeff explains Samsara overrides data after ~year. Discussion of Motive trial — "Motiv" in transcript — that's currently competing for Kreilkamp's whole fleet (will likely replace Samsara entirely). Motive has AI coaching, in-cab voice alerts. Both Tim and Jeff lean toward Motive ("which my guess is we're going to go with bottom" — likely "going to go with Motive"). Luke explains the actuarial framing ("not predictive, math-ing"). Tim loves "actuarial." CRITICAL counter-objection raised: "the other thing we have to be careful of collecting too much information because that's discoverable… unless we figure out how to legally store that" (01:12:00ish, line 1588). Luke backs off: "knife without a handle, maybe we won't touch that for now." (Driven by Tim's "wake up at 2am" obsession; Luke handled well by deferring.)
Tim mentions his 4pm with Matt-the-PowerBI-guy at some point but does NOT actually leave the meeting. Luke uses this stretch to nail success criteria: "metric for success — anything higher than zero for any calls after 5 p.m., mine was really easy." Jeff: also wants his team to "not have to go through one out of a hundred freaking versions." Brief on lead-vendor ad-spend optimization (more dollars to better-converting vendor). Then Luke asks the killer question: "is the decision democratic, one person one vote?" Tim: "we usually talk through it, we don't split, I don't think we've ever had a split vote." (Driven by Luke locking criteria; Tim closing.)
Luke asks: "outcome-based, retainer, hourly?" Jill (the "queen of money"): "we don't really have anything that's paid based on result" — has lawyer (hourly), insurance (premium). Luke surprised. Tim takes the floor for the trucking-market macro (3.5-year downturn since covid+1; illegal-driver supply surge crashed rates; supply rebalancing now; market just turned up "in the last couple months"). Luke: "wait till white-collar people lose their jobs to AI and find out they can make $150K in a truck." Tim: "I wish we could pay them $150K, economics don't allow that." Then the ERC tangent — Luke discloses he consulted for an ERC firm (reputable, processing appeals now); Tim emphatically declined to take any ERC ("not one government penny… we can hang our heads very high"). Around 01:41 Jeff asks the durable-storage question (data needs to live in Tenstreet ideally). Luke clarifies: this is automation, no AI presence on Kreilkamp servers. Luke names the stack: Opus 4.7. Jeff is technical too — runs Claude Code off a Mac mini at his house from old Ethereum mining rigs; has an agent that answers his voicemails and emails. "Why aren't you doing this? Why am I here?" — Jeff jokingly. Around 01:46 JEFF leaves (Tim stays). The exchange "good poker face / Chinese arithmetic / great to meet you / looking forward to swapping emails this week" is Jeff and Luke doing the polite handoff at Jeff's departure. (Driven by Luke locking commercial terms; Tim driving macro + ERC; Jeff closing personally.)
This is the highest-leverage section for Luke's prep. With Jeff gone, the room is now Luke, Tim, and Jill — meaning the principal AND the dollars-gate are both present for the entire pricing conversation. Every signal here carries more political weight than it would have with Jeff in the room.
Luke's gut: "doable, in some ways quite a bit more simple than what I scoped out." Then Luke's explicit pricing walk-down — said directly to Tim and Jill: "I was thinking $25K to deliver 10 qualified warm leads, then $20K for the next 10… I'm thinking proving it out in the pilot gives access to all the other things in your ecosystem." Then immediately self-correcting: "it may not — there may not be enough there for a fifty-thousand-dollar pilot. It's still worth doing. But I don't think it should cost you fifty thousand dollars." Luke pitches outcome-based pricing. Reading the room in real time with the principal AND CFO watching — a strong move, not a weak one.
Tim asks about NDA / "are you going to take this to a competitor" — Tim's protective instinct. Luke commits to NDA. Luke names the relationship anchors: "you're the home team, you're in our backyard, you're a friend of Dan Weber's, you're connected to a feed mill" (i.e. the FGF connection in Tim's portfolio).
Then Jill delivers the killer pricing-down feedback via the 401k advisor analogy: "we have somebody that manages our 401k — investors usually charge 1%, he's charging significantly less, my mom is now investing with him, my neighbor is, he's got this huge book of business, he's making way more than he would have at 1% because he only would have had a few people." Tim is present for this. Jill is openly steering Luke on price WITH TIM'S TACIT CONSENT. Translation: come in lower, build the relationship, expand through trust. Luke heard it: "I would so much rather get the pricing on the durable down the road every month, tied to the actual deliverable… the test needs to be properly priced so that it wins and it wins at a price that everyone's excited about."
Luke makes the explicit ask to Tim + Jill to set the number: "the AI agent is going to give me a number that is too high… I would be helped if you know the number that makes sure everybody is happy at the win benchmark." Neither Tim nor Jill named a dollar figure. They deflected. Open loop — Luke owns coming back with it.
Discussion moves into fleet-of-businesses scope: warehouse inventory automation (license plate + location scanning), 401k census automation (HR retiring after 35 years), income statement automation. Luke clarifies model: "I want all my clients to fit on this hand… I'd much rather do five or six clients deep than expand." Tim: "I want this to be ours, not to go to competitors." Luke: agreed.
Other businesses confirmed: two trucking companies (Kreilkamp + Brett Redman in California), two brokerages, plus a thing-they-thought-they-acquired-in-Iowa-that-was-actually-a-brokerage-but-they-took-the-drivers-and-customers. Greenfield mentioned (likely another entity). HR/admin staffing has very low turnover (in contrast to drivers). Florida mentioned (Tim's wife has a place, Tim "between 0 and 5 years" until succession).
Then Luke's "is this 4 years too soon" question to Tim about succession (02:15:52). Tim discloses fourth-gen, fifth-gen-online, succession-planning underway with younger managers (John Shiz, controller, Tim's son at the feed mill, Monica). Tim explicitly affirms: "I represent fourth generation… fifth is online… probably between zero and five years." Luke surfaces the temperamental risk of next-gen: "let's play to not lose rather than let's play to win… for me that would be the end." Tim agrees the chutzpah is in him, not guaranteed in the next gen.
Closes with the meeting wrapping — Tim: "What's your agent's name?" Luke: "Oh, God." [meeting ends]
(Driven by Luke (pricing reset directly to principal + CFO), then Jill (key analogy with Tim's consent), then Luke (partnership frame), then Tim (succession disclosure).)
| # | Timestamp | Speaker | Quote / near-quote | Why it matters |
|---|---|---|---|---|
| 1 | 00:12:09 | Tim | "If we're talking about drivers, we don't use iRecruit for drivers." Jeff: "It's Tenstreet." | Structural correction of the entire dossier integration plan. The pilot's "iRecruit ATS write-back" assumption is wrong. Tenstreet is the actual driver ATS. Tenstreet has its own API and its own AI fit-scoring already. |
| 2 | 00:17:28 | Jeff | "We end up converting about 1% of our leads into hires at the end of the day." | Confirms top-of-funnel math: 1,200-1,400/mo → ~12-14 hires/mo natural rate. To hit 150-200 hires/yr, the leak isn't volume — it's reach (no-response rate at 592/1924 in April = ~31%). |
| 3 | 00:30:12 | Jeff/Josh | "It's around 70% as we currently sit. ... your turnover is 50%, but your retention changes." | Their reported "turnover" is a misnamed retention metric. Real industry turnover is ~100%. Either way: huge churn against ~285-truck fleet. |
| 4 | 00:30:34 / 00:32:00 | Josh | "We have 800 trailers and about 285 trucks. ... right now I have about 10 in the shops of 275 trucks would be the available right now of those about 40 of them are empty." | Hard fleet numbers. The "40 empty trucks" framing is real and current. |
| 5 | 00:34:34 | Luke (synthesizing) → Tim confirms | "150 to 200 drivers hired a year is where you have to be." | The "40 drivers" framing in the dossier understates the scope by 4-5x. Pilot needs to be reframed around sustaining attrition replacement, not closing a one-time 40-driver gap. |
| 6 | ~00:40:00 | Luke (citing Tenstreet's data) | "30 to 40% of missed phone calls of people who are interested go to somebody else." | The headline ROI claim — and it's Tenstreet's number, not OBB's, which makes it defensible. |
| 7 | 00:46:00 | Tim | "Why are you only after hours?" | Tim spontaneously broadened the scope from after-hours-only to every missed call. This is a Tim-led scope expansion — good signal. |
| 8 | 00:51:58 | Tim | "You could probably pay $150,000 a year and you would still have a [turnover/journey]." | NOT a pricing anchor. Context is driver-pay. Tim's saying pay doesn't solve turnover. The "$150K" appears again at 01:36:32 in the same driver-pay context ("I wish we could pay them 150, economics don't allow it"). The dossier should not treat this as a price ceiling. |
| 9 | 01:00:23 | Jeff | "Tim, Josh were both saying, like, having that experienced driver gets them out the door so much quicker." | Confirms Josh is in the room and aligned with Tim on the experienced-driver priority. The fit-scoring layer (Tenstreet's, or whatever replaces it) should weight experience. |
| 10 | 01:11:30ish | Tim | "The other thing we have to be careful of collecting too much information because that's discoverable… unless we figure out how to legally store that." | Legal/discovery objection raised on the predictive-accident agent idea. Luke deferred ("knife without a handle"). This shape of objection will likely return on any data-aggregating tool — Luke should pre-build a "discoverability posture" answer for any Phase-2 expansion. |
| 11 | 01:32:38 | Jill (likely) | "We don't really have anything that's paid based on result. ... lawyer paid per hour, insurance premium." | Vendor culture is hourly/flat-fee. Luke's outcome-based pitch is novel for them. He needs to translate it into hourly-equivalents they can compare. |
| 12 | 01:38:02 | Tim | "I did not do it, nothing, no, not one government penny. I couldn't sign on the bottom line." | Tim's ERC abstention. Critical character signal. He decided money on the table was better than risk exposure. This is the same temperament he will apply to your pilot. Conservative on commitment, won't sign anything he can't defend. |
| 13 | 01:49:18 | Luke | "It may not — there may not be enough there for a fifty thousand dollar pilot. It's still worth doing. But I don't think it should cost you fifty thousand dollars." | Luke walked his own pricing down — said directly to Tim and Jill, both in the room. This was the read-the-room moment with the principal AND the dollars-gate watching. He chose the durable-relationship play over the one-time win. |
| 14 | 02:04:19 | Jill | "We have somebody that manages our 401k. Investors usually charge 1%. He's charging significantly less. My mom is now investing with him. He's got this huge book of business, making way more than he would have at 1%." | The clearest pricing-down signal of the entire meeting — delivered with Tim present. Jill is openly steering Luke on price WITH TIM'S TACIT CONSENT. Their decision-shape ("we usually talk through it, never had a split vote") means Tim's silence is endorsement. Low entry → big book → more money. |
| 15 | 02:09:54 | Luke | "I would be helped if you know the number that makes sure everybody is happy at the win benchmark." | Luke asked Tim + Jill to set the price. Vulnerable, partnership-shaped move. Neither named a number. They deflected. Open loop Luke owns. |
| 16 | 02:10:51 | Tim | "Are you going to then take that to let's say a competitor? Make me sign a non-disclosure." | Tim's protective instinct. He has been burned before — "we beta things and develop things for our software and then we pay for it and then they go and take it and they make it a part of their master." Luke immediately committed to NDA. Small contract action-item, not a problem. |
| 17 | 02:12:06 | Tim | "You're the home team, a friend of Dan Weber's, we love Dan Weber, he's our friend, you're connected to a feed mill." | Dan Weber referral validated explicitly. "Feed mill" = the FGF connection in Tim's portfolio — Tim is mapping OBB's relationship anchors into his own business network. (Earlier synthesis read "female" — that was a whisper mishear of "feed mill"; Luke corrected.) |
| 18 | 02:16:23 | Tim | "I represent fourth generation… fifth is online… probably between zero and five years [until succession]." | The single biggest strategic signal. Tim is potentially exiting within 0-5 years. The "chutzpah" required to do this work may not survive the transition. Luke flagged it. Tim agreed. Luke's exposure: long-tail recurring revenue evaporates if succession brings in a risk-averse next gen. |
| 19 | 02:00:24 | Tim/Jeff | "You don't own one trucking company, you own two — two okay, and two brokerages." | Confirmed fleet: Kreilkamp + Brett Redman (CA) + 2 brokerages + an Iowa thing that wasn't actually an acquisition. Plus the warehouse + the HR/401k operation + a feed mill (Tim's son works there). This is at least 5-7 distinct businesses inside the holdco. |
| 20 | 02:01:00 | Luke (admission) | "There are many times where I feel really confident that I know a thing because my AI told me… [the Iowa] was an instance of my AI agent having a rosier view." | Luke voluntarily named an AI-research error to Tim's face. Charming + disarming, but worth flagging: the dossier had at least one wrong factual claim about an Iowa acquisition that Tim corrected in person. |
Confirmed: structural pilot-scope question. Driver ATS is Tenstreet (transcript reads "10th Street" throughout). iRecruit is used for SOMETHING at Kreilkamp Holdings — Jeff says "That is one of them" when Luke asks "Which is iRecruit?" (00:12:00) — implying iRecruit is used elsewhere in the org (the other companies — recruiting for non-truck-driver roles at Brett Redman, brokerages, warehouse, feed mill, etc.). For the driver pilot, iRecruit is irrelevant.
00:11:00 (lead-gen vendors):
Jeff: "CDL Life, Trucker's Report, Drive My Way. … Two out of the four do like a form of prescreening. So they'll text that person, go over some of the things that we need them to know about us in order to move forward. Then they end up in our ATS."
00:12:09 (the correction):
Luke: "Which is iRecruit?" Jeff: "That is one of them, yeah." Luke: "That is one of them, okay." Tim: "If we're talking about drivers, we don't use iRecruit for drivers." Luke: "Okay, that's—" Jeff: "It's Tenstreet." Luke: "Tenstreet, okay. That—okay, great, yeah, helpful, thank you. A bit of information that I got wrong." Jeff: "I may have given that to you wrong." Luke: "All good. I'm not pointing fingers."
00:13:00 — Tenstreet UI walkthrough begins:
Jeff demos Tenstreet — shows leads coming in with name, contact, class A status, experience tags. Mentions "Pulse" feature.
00:14:22 — Pulse:
Jeff: "This is called Pulse, which is like an instant messaging feature. And then we have templates set up that we can — we do drip campaigns to convert these people, and then the recruiter sends one-off e-mails to schedule phone interviews, plus give phone calls." Luke: "And Pulse, the vast majority of drivers in the field, have Pulse now, because most people do required." [Pulse is NOT SMS — Luke confirms: "It's just an app, like an in-app messaging?" — Jeff: "Yes."]
00:15:00 — Full application flow:
Jeff: "Now we have all of their previous work history, their license information, and then going into the questions about the FMCSA stuff." [Confirms FMCSA fields ARE in the full application — but NOT in the lead-stage data.]
00:16:00 — Funnel numbers (within Tenstreet):
1,200-1,400 leads/mo → 1,924 outreach attempts in April → 192 phone interviews scheduled (~10% of attempts) → 200 phone screens executed → 75 moved forward → ~12-14 hires (1% of leads).
00:28:54 — Tenstreet AI scoring:
Jeff: "So like Tenstreet does some AI scoring already. So like the people here with like these little green checks. … their fit score is, you know, so they're telling us they had zero accidents, they've had zero violations, and their experience in this case is zero."
00:58:57 — Tenstreet does NOT have SMS:
Jeff: "We don't have it through there [Tenstreet]."
00:59:21 — Calendly is used for self-scheduling:
Jeff: "We have other things we'll use, like Calendly, that sends text messages out, e-mails. So that's how we get them to schedule on our calendars." [So the current text-touch in the funnel is via Calendly, not via Tenstreet directly.]
00:59:53 — Tenstreet data-flow:
Jeff: "Updates into Tenstreet, and then we can call them then, and they'll send them reminder texts that, hey, you know, Jeff is calling you in an hour." [Calendly bookings push back into Tenstreet contact record.]
01:41:54 — End-state data residency (after Tim leaves):
Jeff: "Where does all the data end up living at the end of the day because in your stuff you're all set Power BI but recruiters don't use that." Luke: "Yeah, no, I — where do you want it to live?" Jeff: "Tenstreet ideally, that's the only place." Luke: "That's exactly where it should — I will do some research into how that data gets there but I know the data that we produce can get there… and does it have two-way [API]? Like I applied, I'm excited to work at Kreilkamp, your technology picks up — does that filter back onto that person's profile? I don't know but I will know."
Summary of what Jeff actually uses for drivers today: 1. Tenstreet — primary ATS for drivers. Receives leads from paid vendors + direct applications. Hosts full application. Hosts FMCSA data on full apps. Does basic AI fit-scoring (experience, accidents, violations). Has Pulse (in-app messaging — driver tablets) but NOT SMS. 2. Calendly — used to self-schedule phone interviews. Sends reminder texts. Writes back into Tenstreet. 3. Lead vendors (Tenstreet-feeding): CDL Life, Trucker's Report, Drive My Way + multi-carrier leads. 25/50/25 spend mix. Two of four do prescreening text drips on Jeff's behalf. Two of four have call-center "matchmakers" for no-response re-engagement. 4. VoIP recruiting number: 1-800-999-7112. Painted on trucks? No — that's the general 1-800 line. The recruiting number is separate and "everybody" doesn't know how many calls hit it. No call-data instrumentation today. 5. iRecruit: USED FOR OTHER KREILKAMP COMPANIES (non-driver roles). Not part of the driver pilot.
Pilot integration target = Tenstreet API (write-back of SMS/voice intake → contact record), plus VoIP carrier API for missed-call detection / text-rescue trigger. NOT iRecruit. Luke needs to research Tenstreet's developer surface this week.
The meeting went well — better than well. The room is bought in conceptually, Tim's intuition pushed scope wider (every missed call, not just after-hours), and Jeff is genuinely excited and technically literate enough to be an ally rather than a gatekeeper. The two automation pieces (text-rescue on missed calls + pre-screen compression with calendared callback) are correctly scoped and Jeff already validated the design intent (less than 5 questions, no "years of experience" as a gate).
But three things shifted under your feet:
The relationship is stronger after this meeting, not weaker. Tim's NDA-ask, the explicit "you're the home team, friend of Dan Weber's, connected to a feed mill" relationship-anchor speech, and the "fleet of businesses" disclosure (Brett Redman + 2 brokerages + Iowa-thing + warehouse + 401k + income statement automation potential) are gifts. The "0-5 years to succession" disclosure is the structural risk Luke needs to weigh in pilot pricing — front-load durable value because the long tail might not survive a generational handoff.