May 6, 2026 · 24m · top-founders
Selling Check for $400M, Now Building a $1.5M ARR AI Startup
0:00 / 0:00
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Fintech veteran Ahikam Kaufman joins Nathan Latka to discuss building SafeBooks AI to $1.5M ARR by automating enterprise financial workflows, while sharing critical venture and equity lessons from selling his previous startup, Check, to Intuit for $400 million.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Nathan holds 27.6% of the talking time here. How this is scored →
Nathan as informed peer 5.7 Guest teaching 5.3 Guest disagreement 2.5 Nathan pushing back 3.8
speaking balance: gold is Nathan, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 2:07 Premise rejection on SMB workflow
Ahikam directly shuts down Nathan's Bill.com and QuickBooks comparison, stating that it fails to represent their enterprise target market.
Hardest push from Nathan ▶ 20:36 Challenging proprietary IP vs DeepMindNathan refuses to accept generic moat claims and challenges Ahikam on why DeepMind cannot easily replicate their graph ETL technology.
Biggest teaching moment ▶ 8:35 Correcting Check acquisition facts and historyAhikam corrects Nathan's exit numbers and year, clarifying it was a $400M acquisition in 2014 rather than $360M in 2019, and was the largest M&A deal in WSJ that quarter.
Nathan holds their own ▶ 20:36 Formulating technical counter-exampleNathan demonstrates deep technical understanding of ETL processes and complex data modeling by invoking DeepMind's AlphaFold.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Nathan as informed peer | Guest teaching | Guest disagreement | Nathan pushing back | Why |
|---|---|---|---|---|---|---|
| Introducing Ahikam Kaufman and SafeBooks AI | 5 | 6 | 4 | 4 | Nathan attempts an SMB roleplay using Bill.com and QuickBooks, but Ahikam immediately rejects the premise by clarifying that SafeBooks targets large enterprises with complex CPQ, CRM, and ERP systems. | |
| Enterprise Ideal Customer Profile and Accountant Shortages | 5 | 4 | 1 | 3 | Nathan probes enterprise revenue thresholds and ACVs, while Ahikam details how the accountant shortage and manual billing errors drive multi-hundred-thousand-dollar contract values. | |
| Graph Database Foundation and $15M Seed Funding | 5 | 7 | 3 | 2 | Ahikam corrects Nathan's timeline and valuation regarding Check's sale to Intuit ($400M in 2014, not $360M in 2019), and educates him on building proprietary banking connections before Plaid existed. | |
| Venture Funding Strategy, Dilution, and Creating Millionaires | 6 | 5 | 1 | 4 | Nathan digs into dilution management, cap table distribution, and employee outcomes, prompting Ahikam to share how they created over 10 millionaires and utilized a $25M retention pool. | |
| Scaling SafeBooks to $1.5M ARR Without Hallucinations | 6 | 4 | 3 | 3 | Nathan asks for personal take-home pay, which Ahikam politely declines to share; the conversation pivots to SafeBooks' $1.5M ARR and technical architecture for eliminating hallucinations. | |
| Defensibility and Proprietary Graph IP vs. AI Wrappers | 7 | 6 | 3 | 7 | Nathan presses on defensibility against AI wrappers by asking why tech giants like DeepMind could not replicate the ETL process, prompting Ahikam to explain the necessity of deterministic graph linking for corporate audit trails. |
Statements from this episode (17)
Kaufman: Accountants Lack Tools to Verify Cross-System Financial Data
“One of the challenges in the office of the CFO is there's a gap between what an accountant is trained to do and what he needs to do, which is basically check his data across multiple systems in real time across structured and unstructured data.”
Kaufman: SafeBooks AI Can Fully Automate Cross-System Financial Data Validation
“And we now using the benefits and power of AI, we can fully automate that work.”
Kaufman: Financial Data Verification Breaks Down at $200M to $300M Revenue
“When I'd like to think that companies that starts to exceed like two, Two, three hundred million dollars in revenues would significantly feel the pain because they would have to cope with multiple offerings, multiple products, data structure, which is very dif…”
Kaufman: Offshoring Cannot Prevent Revenue Verification Errors
“Even if you offshore this work, we see significant amount of mistakes in each company we're working with, and it's all natural in you.”
Kaufman: SafeBooks AI starter contract is around $100K to $125K
“The initial use case we start with, it's around the way we kind of sell it, it's the cost of a single resource finance resource. So it's, I would say it's around like a hundred, a 125,000 dollars for the initial starter use case.”
Kaufman: SafeBooks AI's largest customer contract is around $300K
“We just started to go to market like about less than a year ago. So, but we do have, you know, I would say our largest engagement is around 300,000 dollars right now.”
Kaufman: Falling Valuations Will Force Companies to Automate Workflows
“We are now seeing a significant reduction in stock prices, which will force companies to be more efficient and Try to remove the human in the loop as much as possible.”
Kaufman: SafeBooks AI raised a $15M seed round from six or seven funds
“We raised a fifteen million dollar seed in, like, two chunks. From like six, seven funds, early, very early stage fund, but we raised like fifty million dollars.”
Kaufman: Intuit Acquired Check for Close to $400M in Q2 2014
“Actually, it was close to four hundred million dollars.
And funny enough in Q two of 2014, when we sold the company to Intuit, we were like the largest M&A deal according to the Wall Street Journal by coincidence.”
Kaufman: Intuit Still Uses Check's Core Data Infrastructure for Bank Feeds
“We decided to create our own data foundation, which is, by the way, was adopted by Intuit, and using into that same technology is using, Intuit is using today for their bank fee.”
Kaufman: Mint's Third-Party Data Provider Caused Frequent Reliability Mistakes
“We, at that time we had a competitor called mint, which was a very popular mobile application, but they bought their data. So they relied on a third party data infrastructure again, before the plot days, which was less reliable, causing a lot of mistakes.”
Kaufman: Menlo Ventures made 3-4x return on Check in under 8 months
“This was, like, one of their highest IRR because, you know, when you return, when you do, like, three or four X, which that's what they did over the course of, like, six to eight months”
Kaufman: Check founders each owned 5% to 10% at exit
“I think we were two founders and each of us had like you know, between five to 10% of the company.”
Kaufman: Founders cannot fix poor equity distribution once acquiring talks start
“One of the things you learn when you sell a company is that at some point in time, it becomes too late. You can't reverse this decision. So you have, from the get-go, you have to decide how you're going to compensate in equity, your founding team, your staff m…”
Kaufman: Intuit added a $25M employee retention pool in Check buyout
“And we also took that retention shot, which at the time was an additional twenty-five million dollars, and we fully deployed it across the board to people”