Steve Hu

Leaders still need to connect the evidence and make decisions; AI should help them move faster without replacing judgment. Hu also stresses the need to deliver useful results quickly when joining a company, because early execution helps build trust.

Steve Hu describes joining a company where the ERP was QuickBooks, the financial model was in a Google Sheet, and there was no monthly close. They used an agent to parse existing contracts into structured data and set up the same process for new contracts, then built a revenue-recognition layer alongside the accounting team’s GAAP process.

Read original

Finance teams need to understand what is owed and what a payment means, not just record money moving through a bank account.

• Hu sees integrated finance platforms as a challenge to standalone billing and AR tools, while reading customer demand for AR as evidence that the problem remains unsolved.
• They frame Ramp’s launch as both pressure on separate tools and validation for Stuut’s work, and say Stuut will keep building autonomous AR.

Hu recalls that even with 50–100 customers, invoicing, reconciliation, and revenue recognition burdened a lean finance team. They argue that bank data alone cannot explain whether a deposit is a refund, a fundraise, or a customer payment—or when that payment was earned as revenue.

Read original

They also point to the data held by accounting firms as part of the strategic appeal. Their post presents firm acquisition and operational change as a way to reshape the economics of the industry, not merely automate individual tasks.

Discussing OpenAI’s stake in Thrive Holdings and the reported plan to acquire and modernize accounting firms, Hu focuses on how AI agents trained for tax workflows could change how those firms work. They contrast that approach with simply selling firms software or relying on ChatGPT or Codex alone.

Read original

Start here to see the operating problems behind Steve Hu’s later comments on accounts receivable and accounting.

Steve Hu opens with a finance-leadership scenario: a new CFO asking Claude, “How do I run this business correctly?”

• The warning is against a circular process where teams feed AI the same questions, pass the output to another team member, and never add leadership judgment.
• They give a concrete example from joining a company: the ERP was QuickBooks, the model lived in a Google Sheet, and there was no monthly close.
• They pointed an agent at every contract in the company’s history, had it parse each contract into JSON, and set a routine to do the same for new files landing in Drive.
• They also built a revenue-recognition layer that runs its own GAAP process alongside the accounting team’s.
• The post is equally clear about what they say they are not doing: building 50-page slop documents, asking ChatGPT what to do to make an impact, or using Grok to run 1x1s.
• They recommend two approaches when joining a company: move faster and more efficiently, but also step back and ask when leaders become so reliant on AI that they cannot function without it.
• They cite Tarek’s point that if results take 30, 60, or 90 days, trust is being built too slowly.

For brands planning a campaign

For a fintech or finance-software campaign, this shows Steve Hu can discuss AI adoption in finance operations with specific process detail, not just general enthusiasm. It supports a co-authored piece on avoiding AI theater in AR and accounting operations, especially for an audience of finance leaders in their first 90 days.

Read it to see how Steve Hu’s Ramp experience connects to current work at Stuut on autonomous AR.

Steve Hu writes that Ramp, described as a $60B fintech giant, is entering the accounts-receivable space.

• As an ex-employee, they say this was expected and makes them more bullish on Stuut.
• They helped build early requirements for Ramp’s own team.
• When Ramp launched a SaaS offering, many customers wanted to be invoiced directly instead of using auto debit.
• Even with 50 to 100 customers, invoicing, reconciliation, and revenue recognition were already creating a mess for a lean finance team.
• Ramp helped customers manage money going out, but there was still a gap in understanding what was coming in.
• Bank data could only tell so much: a deposit could be a refund, a fundraise, or a customer payment, and even a customer payment does not show when the revenue was earned.
• They say Ramp’s launch starts connecting contracts to invoice drafts, collections follow-ups, payment matching, and revenue recognition, which they see as a logical extension of Ramp’s office-of-the-CFO ambition.
• Their read is that this pressures standalone billing and AR software because bundling invoice creation, reconciliation, and revenue recognition into a broader finance platform makes a separate tool harder to justify.
• For Stuut, they see validation: a $60B AI-first company with many customers says AR was a top request, and customers are still asking for help getting paid.

For brands planning a campaign

This is strong material for a co-authored piece on AI-driven AR and accounting operations.

• Steve Hu can speak to product requirements, buyer objections, incumbent bundling, and the difference between managing money out and understanding money in.
• Disclose the ex-employer relationship and current Stuut role in any campaign use.

Read it for the market structure behind the AR and accounting products Steve Hu writes about elsewhere.

Steve Hu notes that at the end of 2025 OpenAI took a stake in Thrive Holdings, which Forbes reported was spending $1B to acquire and modernize accounting firms with AI.

• They call “optimizing accounting firms” the least sexy thing to do with AI, but argue the space is huge.
• They spent months studying the industry before joining Stuut.
• Modeling, tax, advisory, and audit all involve many fees, and optimizing practices means better margins.
• The problem, according to them, is that selling software to these firms does not fundamentally change how they operate.
• ChatGPT is not enough, and Codex is not enough; relying only on those tools means clawing at a tiny bit of margin.
• Changing how a firm’s team operates with AI, by contrast, can take apart the industry’s margin structure.
• They describe Thrive and OpenAI’s angle as buying an accounting firm, upending the infrastructure and operating layer, and enabling the firm to operate differently.
• They note that OpenAI provided a dedicated team that trained AI agents specifically on tax workflows.
• They also highlight the data advantage: labs are buying this data, so OpenAI can sell into accounting firms and gain value from the data those firms have.
• They say this explains why OpenAI spun up a new company to deploy AI in the enterprise.
• The post closes with the line “All this to make the audit sexy again.”

For brands planning a campaign

For a fintech or finance-software campaign, this post shows Steve Hu can explain why AI in finance should address operating-model change and data ownership, not just software features. It supports a co-authored piece connecting accounting-firm economics, AI agents, and competitive positioning for AR and accounting products.

They describe pointing an agent at every contract in a company's history, parsing each into JSON, and setting a routine to handle anything new arriving in Drive, then building a revenue recognition layer that runs its own GAAP process alongside the accounting team's. On the AR side, they trace the operational problem from the ground up: at roughly 50 to 100 invoicing customers, reconciliation and revenue recognition were already straining a lean finance team, and bank data alone could not distinguish a refund, a fundraise, or a customer payment, or show when revenue was earned.

Walks through the build in operational order — source documents, structured parsing, recurring automation, then a parallel recon and rev rec layer — and names what they deliberately avoid building, such as long generated documents to hand to colleagues.

For brands planning a campaign

A finance-software brand can use this as the technical spine of a co-authored piece: contract-to-JSON pipelines, invoice drafts, payment matching, and rev rec as a connected workflow rather than separate features.

• Their specificity gives the collaboration real detail.
• The brand should supply the comparative product landscape they do not cover.

Examples: Source 1 · Source 2

They describe the trap directly: a founder asks the CFO for the next 6 to 12 months, and the CFO immediately pastes the same question into Claude and forwards the output.

• They call this a circular cycle of teams feeding AI the same questions and passing the same output along, and argue that a leader has to assemble the pieces beyond what a model returns.
• Their alternative is to move faster and more efficiently while asking when leaders become so reliant on AI that they cannot function without it.

They concede the habit is widespread and that the line between using AI for speed and outsourcing judgment is genuinely fine, softening an otherwise firm stance.

For brands planning a campaign

This is the most quotable argument for a brand that wants a co-authored piece with a point of view rather than a product walkthrough. A piece that reads as an uncritical AI endorsement would sit badly with their stated position, so the collaboration should address where AI falls short in finance workflows.

Examples: Source 1

Their reasoning is that modeling, tax, advisory, and audit carry large fees, so improving how firms operate changes their margins.

• Selling software to these firms does not fundamentally change how they work — ChatGPT or Codex alone, in their phrasing, leaves firms clawing at a small amount of margin.
• Changing how the team operates is what lets the industry's margin structure be taken apart.
• They also flag the data angle: labs buying firm data, and OpenAI effectively selling into the firms while gaining value from their data, which they say explains the enterprise-deployment company OpenAI spun up.

Reads a news event through industry cost structure and incentive analysis rather than reacting to the announcement itself, and names the data asset as the second, less-discussed payoff.

For brands planning a campaign

Useful for a brand targeting finance leaders weighing whether AI changes firm economics or only tooling.

• It gives a co-authored piece a structural argument and a rollup angle.
• Because this is their commentary on a reported deal, the brand should pair it with primary or reporting-level material before making claims about the transaction.

Examples: Source 1

They read Ramp's move into AR as exactly that extension of the office-of-the-CFO ambition.

• Part of this is firsthand: they helped build the early requirements for their own team's SaaS invoicing path, and describe Ramp as strong on money going out, with a remaining gap in understanding what was coming in.
• They also treat the launch as validation for their current company's focus on autonomous AR, answering the recurring question of whether AR is saturated by pointing to a large, well-resourced company naming AR as a top customer request.

They say they remain a Ramp fan and an ex-employee while making a competitive read, combining loyalty with a market argument rather than offering a clean neutral assessment.

For brands planning a campaign

Gives a brand a ready-made buying-consolidation thesis to co-author against, with the pull quote that customers still ask for help getting paid despite abundant software. Since the read comes from inside two competing companies, frame the piece around the workflow and buyer decision rather than one vendor's win or loss.

Examples: Source 1

They ground claims in specifics — contract histories, JSON parsing, a lean team handling 50 to 100 invoicing customers, months spent studying the accounting industry before joining Stuut — rather than market-size rhetoric. Their posts also consistently give the counter-position: what not to do, which tools are not enough, and where a hot announcement is less significant than it first appears.

For brands planning a campaign

For a co-authored piece, they are best suited to a build-and-tradeoffs format — a practitioner account of how AR and accounting operations actually change under AI — rather than a broad trends essay. The brand should provide benchmark comparisons, buyer segmentation, and product specifics.

Examples: Source 1 · Source 2 · Source 3

When Ramp launched its AR offering, the post says a large share of customers wanted direct invoicing rather than auto debit, which created invoicing, reconciliation, and revenue recognition work for a lean finance team.

View example

They treat Ramp's entry into AR as validation for Stuut's market rather than as a partnership or sponsorship. Ask them to confirm their title and ownership stake.

View example

One post traces a 15-year run at Connex Ecommerce automating workflows and eliminating manual data entry; a TEDx talk argues that automation creates room for human work. A second post asks a question Hu raises from the other direction: what happens when integration with accounting, ERP, and fulfillment systems stalls an otherwise good deal.

Anderson's frame is broad workforce automation and the personal history behind it, written partly for salespeople facing integration questions. Hu focuses on finance operations, accounts receivable, revenue recognition, and accounting-firm economics, reading industry moves analytically rather than advocating for a mission.

Read original · Original 2

Both argue that how a firm runs week to week matters more than occasional brilliance. Their posts draw on firsthand advisory or client meetings and end with a concrete practice, such as Shmilovich's 17-touchpoint client communication calendar.

Shmilovich's subject is leadership communication and client trust: stating the vision, maintaining response times, and mapping touchpoints across a year.

• Hu focuses on the records and systems underneath the work—contracts, invoices, revenue recognition, and where AI changes them.
• Shmilovich teaches through a remembered story and a list; Hu, through a build narrative and market analysis.

Read original · Original 2

Like Hu, Saperia argues that measurement changes behavior only when someone accountable is asking. Both compare their fields to a metrics-rich world: SaaS ARR for Saperia and software-only AI for Hu.

Saperia focuses on management cadence and metrics discipline in services firms; neither post covers an AI or automation build.

• Hu focuses specifically on AI agents and accounting and AR workflows.
• Saperia writes toward a PE-style operating rhythm; Hu, toward what the tooling can and cannot do.

Read original · Original 2

Anderson's posts focus on that friction: a deal stalls on “can it work with the rest of our systems,” and Connex Ecommerce maps the workflow and decides whether the integration is supportable.

• That makes Anderson a better fit than Steve Hu for a co-authored piece aimed at sales engineers, solutions consultants, or account executives who need to keep integration complexity from slowing a deal.
• Anderson also brings a long-running automation story: more than 15 years eliminating manual data entry, and a TEDx talk on automation freeing people for higher-value work.
• That can frame why finance and accounting workflow automation matters beyond the software demo.

The tradeoff is depth in AR, revenue recognition, and accounting operations.

• Anderson focuses on workflow automation and integration support in sales cycles, not contract parsing, GAAP revenue recognition, invoicing and reconciliation pain, or buyer decisions around bundled versus standalone AR tools.
• For a finance-software piece on those topics, ask the brand to supply the accounting-specific substance and proof of Anderson's credibility with CFO-level AR buyers.

Examples: Source 1 · Source 2

Shmilovich writes directly to owners, CEOs, partners, and practice leaders at accounting firms about unspoken vision, trust built through predictable communication, and a 17-touchpoint client-communication calendar across a year.

• For a finance-software company selling AI into accounting practices, that is the adoption layer: the piece can address why firms resist changing long-standing client rhythms, how to communicate a new AI-assisted process to staff and clients, and where consistency matters more than a clever one-off campaign.
• Shmilovich is also useful when the brand wants to reach firms from $500,000 to $150 million in revenue and speak to practice management, not just the technology stack.

The tradeoff is technical and product specificity.

• Shmilovich's posts do not cover AR automation, invoice creation, payment matching, revenue recognition, or AI agents in accounting operations.
• They come from a marketing-agency and advisory perspective, not from inside a finance-software product.
• Ask the brand to supply the AI, AR, and revenue-recognition substance; Shmilovich can make the change-management and client-engagement case.

Examples: Source 1 · Source 2

Saperia's posts contrast SaaS metrics with professional-services reality: demand is human work, utilization measures last month's capacity, and booked work is the closest thing a services firm has to ARR.

• For a finance-software company targeting finance leaders adopting AI in accounting and professional-services firms, Saperia can build the operating case for cleaner demand records, early visibility into next quarter, and why firms want private-equity-style discipline without a private-equity backer.
• That is a better fit than Steve Hu when the campaign's value proposition is financial visibility and operating cadence rather than AR collections or revenue recognition.

The tradeoff is that Saperia's published posts focus on professional-services finance metrics and private-equity-style operating discipline, not AI-driven AR or accounting workflow automation.

• They do not discuss invoice creation, collections follow-up, payment matching, contract parsing, or revenue recognition.
• For a piece on AI-driven AR, ask the brand to supply the AR product mechanics and customer proof; Saperia can frame why those mechanics matter to margin and enterprise value.

Examples: Source 1 · Source 2