Train computer vision on 250M+ images to fingerprint a vehicle independent of its license plate

Outcome: Fingerprint the whole vehicle from a large proprietary image dataset rather than relying on the license plate alone.

Context: Metropolis needed near-perfect, real-time accuracy (e.g. granting access to secured buildings) that 70%-accurate OCR can't deliver. It captured 250M+ images and trained heuristics on dents, scratches, stickers, and color, making the plate just one variable in a robust vehicle fingerprint — a data moat that compounds with scale.

“How do we create a fingerprint of your vehicle? How do we actually capture your vehicle independent of whether or not your license plate is covered in mud or snow... I think it's over 250 million images. So how do we think about leveraging those images to train our dataset to identify a vehicle independent of the license plate where the license plate just becomes one variable, not the most important singular variable.”

Spin up a Chief AI Officer to drive both top-down and bottom-up internal AI adoption

Outcome: Create a Chief AI Officer role to balance top-down and bottom-up AI adoption across the whole workforce.

Context: Metropolis stood up a new AI office to build internal tools and drive adoption across 23,000 employees. The hire was an internal, highly-trusted leader combining organizational-transformation experience with a technologist's ability to manage engineers — chosen for existing trust with department heads as much as technical depth.

“We're actually spinning up right now a chief AI officer, and one of the reasons we're doing that is we actually wanna drive even more innovation and more adoption of tour. So we're creating a new office, we've hired that team member, placing them in that role, and their job is going to streamline and build tools for our organization internally, directly tied to artificial intelligence.”

Stop pitching VPs and analysts; route the raise to a senior partner on IC

Outcome: For a non-legible company, take the pitch straight to a senior IC partner.

Context: Metropolis had almost no fundraising luck with VPs and principals. Only a partner on IC with deep experience could grasp and sponsor the vision, so the company shifted to talking only to senior decision-makers — a posture that changed once a $2.1B-revenue track record made the pitch legible on its own.

“If I get on the phone with a VP or a principal and I pitch them historically, almost no luck. It took a partner on IC with a significant level of investment experience to understand the vision of Metropolis, to understand where we're going. So we shifted, we stopped talking to junior team members”

Retain member data in-house and license it to no third parties

Outcome: Keep sensitive member data in-house and refuse third-party licensing to protect the trust the platform runs on.

Context: Because Metropolis handles license plates, credit cards, and biometrics, privacy is discussed at board and executive level. The explicit posture — best-practice security, no third-party data licensing, member co-ownership — is what sustains the fair exchange of value on which the recognition economy depends.

“we use best practices and we also don't license that data to any third parties. So that data is retained by Metropolis and the member itself. So you have your data and we have a license to your data”

After validating the model, raise a mega-round to acquire the market leader

Outcome: Use the proven first deal to raise a mega-round and acquire the category leader.

Context: Premier's success (2x gross profit) let Metropolis return to the venture community, which pushed to do it again. It raised a $1.6B Series C to acquire SP Plus — a public company with ~20,000 employees and ~3,600 locations — making Metropolis the largest parking company globally, roughly two years before the interview.

“we went out and we raised $1.6 billion as our series C to go acquire the largest operator in this space, which was a company called SP Plus, which was a publicly traded company that had, you know, just north of 20,000 employees, just north of 3,600 locations.”

Onboard members with a one-time QR scan, then make every subsequent visit seamless

Outcome: Capture identity once via a QR signup so every later visit across the network is frictionless.

Context: First-time users scan a QR code and enter credit card, license plate, and phone number; from then on they drive in and out of any Metropolis-enabled facility with automatic recognition and text-message charging. The one-time signup has crossed 24M+ Americans and produces a preference to choose Metropolis lots.

“you scan a QR code, you enter your credit card, your license plate, and your phone number. And from that moment on, you're a member on the Metropolis platform, which means you have access to the entire Metropolis network”

Blend private equity, private credit, and venture investors on one cap table

Outcome: Assemble a cap table of PE, credit, and venture investors by offering each its native risk band.

Context: Metropolis's rounds mix Silver Lake, Eldridge, BDT/MSD, Dragoneer, Temasek, and Vista — names rarely on one cap table. The idiosyncratic downside-plus-upside profile lets it recruit cross-box capital, though it required brilliant investors willing to operate outside their usual niche.

“if you look at our cap table, you have this blend of traditional private equity investors, traditional credit investors and traditional venture investors in every round. You don't see them, you don't see those players playing together in one company.”

Acquire an incumbent operator to buy go-to-market and scale the tech overnight

Outcome: Acquire a mid-sized incumbent operator to convert organic go-to-market into an overnight installed base.

Context: Metropolis had unit economics and PMF but a broken go-to-market against risk-averse real estate owners. It acquired Premier Parking (~$120M, on $10-12M EBITDA), jumping from ~50 to ~450 locations and ~200 to ~2,200 employees overnight, and used it as the proof point it could not sell its way to.

“We bought our first parking operator. So we bought a company based in Nashville called Premier Parking, probably the 10th or 11th largest parking operator in the United States. And we tested our mob, we used it to scale our technology to 400 location overnight.”

Use the acquired member base as a jumping-off point to cross-sell into new verticals

Outcome: Reuse the parking-acquired member base to enter new verticals at near-zero incremental acquisition cost.

Context: Parking was chosen as the first vertical precisely because it is everywhere and could build a flywheel: members and license plates captured there become the launch base for mobility (gas, car wash, tolling) and non-mobility (office, doctor's office) experiences. Owners who see one app cover parking, fast food, and building access reinforce the expansion.

“we could use it as a jumping off point to scale into so many other verticals where we had these license plates, where we had these members on platform, and then we could expose that same member value to so many additional verticals, both in and outside of the vehicle.”