AI software, built to pay for itself

We build AI software
that pays for itself, starting with a 30-day audit.

We find the highest-ROI AI opportunity in your business, build the software to capture it, and you pay for it through a recurring SaaS fee, not a consulting bill. If we can't find a path to >10% net margin improvement, your retainer comes back.

check_circle30-day audit, fixed scope

check_circleRetainer converts to build cost

check_circleEngineers from Google, Amazon, Twitter

check_circleRecurring SaaS fee, not a consulting bill

Book a 15-minute call arrow_forward See how it works

30 days · audit to readout

B2B · 20–500 employees, $5M+ ARR

SaaS pricing · no separate build invoice

MODEL How an engagement pays for itself

Illustrative

Net margin improvement targeted

>10% / year, year one

Retainer is returned if we can't find it, or converted into the build if we do. You only ever pay a SaaS fee going forward.

Step 1, 30-day audit

Map every AI opportunity, size it in dollars, rank by ROI

Day 1–30

construction

Step 2, Build the highest-ROI one

Retainer converts to development cost. No separate invoice.

Build

payments

Step 3, Pay only the SaaS fee

Recurring fee, scoped so the software pays for itself in margin

Recurring

shield

If we don't find >10% margin path

Full retainer returned. No deck-only deliverable.

Guarantee

Same mechanic on every engagement Audit → Build → SaaS

How it works

Three steps. One mechanic.
You only pay when it pays you back.

We're a software shop, not a consultancy. The audit exists to find the right thing to build, and to make sure you don't pay for any of it if there isn't something worth building.

30-day audit

We map every viable AI opportunity in your operations, size each one in dollars, and rank by ROI and effort. You get a written report and a 90-day build plan.

Retainer paid arrow_forward

We build the top opportunity

Your retainer converts to the development cost of the highest-ROI build. Same team that ran the audit ships the software, no handoff, no rebuild.

Retainer converts arrow_forward

You pay a SaaS fee

An agreed recurring fee. Scoped so the software pays for itself in margin many times over. If it ever stops doing that, you stop paying.

Ongoing

shield

If step 1 doesn't find a >10% margin path, you get the retainer back.

No partial credit, no "engagement value" carve-outs. Either we find a path worth building, or you pay nothing.

Step 1 in detail

The 30-day audit.

The first 30 days are the audit. You give us access to your team and your data; we hand back a prioritized map of AI opportunities, each with a dollar range and a time-to-value estimate. The 90-day build plan that follows is sequenced so we know what we'd ship first if you pick us to build it. Retainer pricing is scoped to the work required and aligned on your first call.

What you get

AI Opportunity MapEvery viable use case, ranked by ROI and effort Included
Process Automation AnalysisManual workflows AI can eliminate or accelerate Included
Data Asset InventoryWhat you're collecting that you're not using yet Included
ROI Estimates per OpportunityDollar ranges and time-to-value for each finding Included
90-Day Build PlanSequenced, starting with the highest-ROI opportunity Included
Executive Readout60-min session with your leadership team Included

What we need from you

A corporate POCMeets with us once a week, keeps the engagement moving ~1 hr/wk
Access to 2–3 key stakeholdersOps, product, or whoever owns the workflows ~3 hrs total
Relevant data exportsRead-only, granted after NDA and BAA are signed Read-only
60-min kickoff & 60-min readoutYou bring the decision-makers to the readout Once each

When we identify the right build

Your retainer becomes
the development cost.

No separate build invoice. You only ever pay the SaaS fee going forward.

The point of the audit isn't the audit. It's to find the right software to build, and to make sure you only pay for software that pays you back.

Book a 15-minute call arrow_forward

The 30 days

Four touchpoints. Minimal lift on your side.

You give us access to the right people and data; we do the work and come back with findings. Day 0 is the day all agreements are signed and data access is confirmed, the 30-day clock starts then.

Mon

Tue

Wed

Thu

Fri

Sat

Sun

1Kickoff

8Discovery

22Analysis

30Readout

Day 1 · Kickoff

Scope & access

60-min call to align on priorities, map the business areas in scope, and confirm data access.

Day 8 · Discovery

We map your operations

Stakeholder interviews, workflow review, and data inventory across the areas in scope.

Day 22 · Analysis

Opportunities sized

Every viable use case identified, feasibility-checked, assigned a dollar range and time-to-value.

Day 30 · Readout

Readout & build pick

Executive readout, written report, 90-day build plan. You pick what gets built; we start.

The team

Engineers who've shipped at scale,
not consultants who've advised on it.

We've built production systems at some of the most technically demanding companies in the world. That's the difference between finding opportunities that are theoretically possible and ones that will actually get built.

How engagements are staffed: every audit is led by Zach plus at least one other expert matched to your domain. The rest of the bench pitches in where their expertise applies.

Zach Chao

Co-Founder · Engineering & AI

Software engineer at Google, Amazon, and Twitter. UC Berkeley CS. Runs the technical side of every engagement, data pipeline, modeling, and the analysis that turns raw operational data into specific, actionable recommendations.

verified_userGoogle · Amazon · Twitter schoolUC Berkeley CS hubHealthcare data systems codeSearch · ads · revenue systems

Greg Lin

Co-Founder · Platform Engineering

MIT CS. Co-founded CrossInstall, $50M+/yr ad-tech, acquired by Twitter. Senior Staff Engineer at Twitter post-acquisition. Brings the systems experience to handle any data infrastructure challenge that comes up in the audit or in a subsequent build.

schoolMIT MS · CS & Engineering rocket_launchCrossInstall founder (acq. Twitter) verified_userTwitter Senior Staff Engineer dnsAd-tech · $50M+ revenue systems

Arnaud Pilpré

Engineering Leadership & Strategy

MIT MSc. Former CTO at Chartboost. Led the 65-engineer MoPub team at Twitter post-acquisition. Co-founded CrossInstall, acquired by Twitter, where he scaled engineering from the ground up. Brings executive-level engineering judgment to every audit and roadmap.

schoolMIT MSc · Media Arts & Sciences verified_userCTO · Chartboost groupsTwitter · 65-engineer org rocket_launchCrossInstall founder (acq. Twitter)

Daniel Preda

Machine Learning & Data Science

PhD CS from UC Berkeley, MIT M.Eng. Staff ML Engineer at Twitter owning real-time bidding models for ads. Principal Data Scientist at CrossInstall. 6+ years at Quantcast growing from IC to team lead across display advertising and ML systems.

schoolPhD CS · UC Berkeley schoolMIT M.Eng. · CS verified_userTwitter Staff ML Engineer bar_chartReal-time bidding · Ad ML systems

Kaitlyn Chen

Quantitative Finance & AI

PhD in Finance from UC San Diego (Rady School of Management). Scientist at Amazon designing large-scale forecasting models and agent-based tools to model traffic, revenue, and customer behavior for executive decision-making. Outstanding Graduate of Shanghai City (Top 1%), Fudan University.

schoolPhD Finance · UC San Diego (Rady) verified_userScientist · Amazon bar_chartForecasting · agent-based modeling schoolFudan University · Top 1%

Siddhartha Datta

AI Research

PhD in Computer Science from Oxford, focused on adaptive machine learning, with industry research experience at Google on multimodal models. Trading background spanning Goldman Sachs, Credit Suisse, and Société Générale, bridging rigorous ML research with real-world financial systems.

schoolPhD CS · University of Oxford

Minwoo Park

AI Research

PhD researcher in Artificial Intelligence at Yonsei University. Brings rigorous academic AI expertise to the team, from model design and evaluation to applied research that bridges cutting-edge methods with real-world deployment constraints.

schoolPhD AI · Yonsei University smart_toyArtificial intelligence research

Jason Lee

Infrastructure & Site Reliability

UBC Computer Engineering. Senior SRE at Twitter post-acquisition. 6+ years as Senior Software Engineer at CrossInstall, handling 500 billion web hits a month on AWS. Keeps the systems we build running at scale in production.

schoolUBC · Computer Engineering verified_userTwitter Senior SRE dnsCrossInstall · 500B req/mo on AWS cloudInfrastructure at scale

Bhavik Shah

Engineering Leadership & Mobile

Engineering Manager at Twitter and Principal Software Engineer at CrossInstall, the same ad-tech system handling 500B+ monthly requests. Prior: Sr. Engineering Manager and Staff Engineer at Crunchyroll. Now founder of MemReel. Brings deep mobile and full-stack leadership experience across consumer and high-scale B2B platforms.

verified_userTwitter Engineering Manager rocket_launchCrossInstall Principal Engineer groupsCrunchyroll Staff Engineer smartphoneFounder · MemReel

Why us

We ship the software, too. That changes what we look for.

Consultants who haven't built production AI systems will find different opportunities than ones who have. We know what's actually hard, what breaks in practice, and what can realistically be delivered in 90 days versus what's a two-year project. The audit's recommendations are scoped to things we'd be willing to build and stand behind on a SaaS contract, because that's exactly what happens next.

Days to findings

Scoped, delivered, and actionable inside a quarter.

Companies represented

Google, Amazon, Twitter, systems at scale, not slides.

>10%

Net margin threshold

If we can't find a path to >10% net margin improvement, your full retainer is returned.

PhD

Quant on every audit

Every ROI estimate is modeled, not guessed.

FAQ

Common questions.

You're paying for software. The audit is how we figure out what software is worth building for your business. The retainer either comes back (if there's nothing worth building) or converts into the cost of building it, at which point you're on a SaaS fee. You never get a separate consulting invoice and a separate build invoice; it's one mechanic from start to finish.

If we can't identify a path to greater than 10% net margin improvement, your full retainer is returned, no partial credit, no carve-outs. The guarantee assumes you hold up your side: a weekly POC meeting, prioritized working sessions, and data access after NDA and BAA are signed. We'll walk through those terms on the call before anything is signed.

A recurring fee scoped so the software pays for itself in margin many times over, not a percentage of savings, not a usage-priced black box. We set it during the build agreement, after the audit shows the dollar impact. If the software ever stops earning its keep, you stop paying.

B2B companies with established operations, typically 20 to 500 employees and $5M+ ARR, where there are real workflows, real data, and real decisions being made manually that don't need to be. Growth matters: headcount avoidance is one of our most defensible ROI categories, and growing companies have a much clearer path to it. Industry is less important than operational complexity and data maturity.

We sign an NDA and a BAA before any data access is granted. Access is read-only and scoped to the data needed for the audit. We don't share, resell, or use your data for anything outside the engagement, and we delete what we copied once the readout is delivered, unless we convert to a build, in which case retention is covered in the build agreement.

Minimal. The corporate POC meets with us roughly an hour a week. Two or three stakeholders sit for short interviews, about three hours total. Everything else happens on our side. The kickoff and readout are the only fixed meetings on your calendar.

No. The audit deliverable is yours regardless. The model is designed around us building it, that's how the retainer converts and how the SaaS pricing works, but if you'd rather build in-house or with another team, the report is specific enough that any competent engineering team can act on it. In that case the retainer doesn't convert.

Scoped to the work required, company size, operational complexity, and how many business areas we're covering. We align on a number in your first call before anything is signed. Because the retainer is either returned or converts to development cost, it's not a consulting fee you'll never see again, it's the down-payment on software that increases your net margins.

Agreements signed and kickoff scheduled within two weeks of your first call. Readout 30 days after that. Written findings in hand inside six weeks, start to finish, and if we're building, the first version follows on the 90-day plan we deliver at the readout.

We build AI software
that pays for itself, starting with a 30-day audit.