Jan 12, 2026 · 41m · mixergy
#2293 $2 million / year AI engineering firm
0:00 / 0:00
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this podcast episode, host Andrew Warner interviews Tarun Thummala, founder of PressW, to uncover how he built and scaled a $2 million-per-year custom AI engineering agency. Thummala shares actionable insights on value-based pricing models, consultative enterprise client acquisition, and proprietary AI-driven development workflows.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Andrew holds 21.2% of the talking time here. How this is scored →
Andrew as informed peer 3.9 Guest teaching 3.7 Guest disagreement 1.8 Andrew pushing back 3.0
speaking balance: gold is Andrew, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 28:18 Inverting the dev shop premise
Thummala firmly rejects Warner's assertion that PressW is fundamentally just an ordinary dev shop that happens to use AI, deliberately reversing the host's logic.
Hardest push from Andrew ▶ 32:38 Warner challenges agency uniquenessWarner bluntly interrupts Thummala to demand what is actually unique about his setup, pressing him on whether he is just copying industry-standard agent workflows.
Biggest teaching moment ▶ 31:03 The 80/20 inversion of software engineeringThummala delivers an insightful explanation of how AI shifts software development from 80% generation to 80% review while simultaneously compressing total cycle time by 10x.
Andrew holds their own ▶ 20:23 Warner breaks down executive event psychologyWarner commands the conversation by articulating the exact strategic and psychological leverage of sharing a stage with corporate giants like Dell and Atlassian.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Andrew as informed peer | Guest teaching | Guest disagreement | Andrew pushing back | Why |
|---|---|---|---|---|---|---|
| Agency Revenue and Document Processing Case Study | 4 | 4 | 1 | 2 | Warner questions Thummala on revenue numbers and operational delivery. Warner demonstrates understanding of service agency billing pitfalls when discussing hourly rates versus value pricing. | |
| Why Start an AI Services Firm Over SaaS | 3 | 5 | 1 | 1 | Thummala educates Warner on why the historical 10-year SaaS playbook fails for AI due to heavy integration and data context requirements. | |
| Founding Team Dynamics and Securing the First Client | 3 | 3 | 2 | 4 | Warner presses Thummala on why he chose to share equity with three partners instead of hiring contractors or operating solo to maximize profit. | |
| Networking Tactics and Cultivating Investor Relationships | 3 | 4 | 2 | 3 | Thummala corrects Warner's premise regarding customer acquisition, clarifying that over 90% of business came from net new relationships rather than previous employers. | |
| Personal Anecdote on In-Person Dynamics and Austin Nights | 6 | 2 | 1 | 2 | Warner demonstrates deep expertise in networking leverage and event hosting ROI, analyzing how sharing stages with enterprise tech leaders creates an authoritative halo effect. | |
| Breaking into Private Equity and Positioning as an Engineering Firm | 3 | 3 | 1 | 2 | Warner questions Thummala's LinkedIn positioning, prompting Thummala to explain the strategic shift from consultancy framing to engineering firm. | |
| Sponsor Highlight: Zapier and Custom vs. Off-the-Shelf AI | 5 | 3 | 3 | 3 | Warner suggests that technical developers might dismiss no-code agency tooling, but Thummala pushes back against tool snobbery by focusing entirely on business outcomes. | |
| The Shift from AI Hype to Measurable Business Outcomes | 4 | 4 | 3 | 4 | Warner characterizes PressW as a standard dev shop using AI, prompting Thummala to politely reject and invert the framing. | |
| Internal AI Tooling and Specializing in Regulated Workflows | 5 | 5 | 3 | 6 | Warner directly challenges Thummala's differentiation, asking if his workflow agents are merely what everyone else is doing. Thummala defends his strategy with vertical specialization. | |
| Enterprise AI Adoption and the Expanding Blue Ocean | 3 | 4 | 1 | 3 | Thummala details his end-to-end sales-to-code operating architecture Atlas. Warner questions whether enterprise customers will eventually build these tools internally. |
Statements from this episode (14)
Thummala: PressW generated over $2 million in revenue last year
“Yeah, we did we did over two million dollars last year.”
Thummala: PressW automated 90% to 95%+ of court document processing for a client
“We were able to, like, build a little system that you know, can automate up to, like, 90, 95% plus of those types of documents.”
Traditional SaaS playbook fails for AI without professional services, says Thummala
“The typical way that SaaS-type businesses, like software-style businesses, were built, scaled, and sold in the past 10 years was not going to be the same formula that was going to work in this new age because AI requires a ton more professional services, almos…”
Thummala: PressW chose services to generate cash and learnings before SaaS
“Services make sense, we can get cash now, we can watch the world, and we can stay on the forefront of what's possible today, and ideally take those learnings, like, just maximize surface area of information and then from that, you know, start to build somethin…”
Thummala: 90% of PressW's business comes from post-launch relationships
“Actually, no, I would say only like less than 10% of this business, less than, yeah, less than 10%, let's call it, let's just make it be generous with it. Less than 10% came from like old relationships. I'd say 90% were people that I've met and interacted with…”
Thummala: Strategy is under 10% of PressW business; 80% are engineers
“I used to say consultancy, and people then thought that we were mostly doing advising work and strategy work, which is actually, like, less than 10% of our business. Like, a majority of our, almost 80% of my team are engineers.”
Clients buy business outcomes, not custom code or Zapier, says Thummala
“Who really cares about what it's made in or if it's custom or if it's Zapier, like the thing that you're, should be buying is the outcome for the business.”
Thummala: AI enterprise buyers now demand specific workflows and ROI
“In the last year, people are now sharpening up their budgets. They're sharpening their pencils, and they're like, okay, we know AI is effective. We know it's a hundred percent going to be transformative. How? Like, what are the specific workflows, outcomes, RO…”
Thummala: AI flipped coding to 80% review and cut raw time 10x
“The time spent on coding was probably 80% generation, 20% review. It's flipped that into 80% review, 20% generation. But the key is, is that the actual amount of like the raw time has dropped by like 10 X when you make that kind of switch.”
Internal AI coding workflows will eventually commoditize across all firms, says Thummala
“Eventually everyone's going to have the same sort of, like, technical architecture behind it.”
Thummala: Businesses will eventually run hundreds of AI agents internally
“Today, most businesses have zero or one agent or AI integrated into their business.
If that in the future, I'm sure that they're going to have hundreds of agents running within their business, and that unlocks an entirely new class of problems.”
Thummala: MIT report found 95% of AI pilots failed to reach production
“And that's actually why, you know, if you see, like, the MIT report, 95% of things didn't, 95% of, like, AI pilots never made it to production.”