Why I Built Steer Workforce

I built Steer Workforce because I needed it. Three businesses, four tax treatments, and a spreadsheet I couldn’t stand looking at any longer.

That’s the short version. Here’s the long one.

The Setup

I run three companies.

BlastAsia, founded in 2000, does software product engineering. Xamun Technologies builds an AI operating system for mid-market businesses. Steer Platform Pte. Ltd. — Steer Workforce — is the third.

Across these three companies, I have employees in Manila. Contractors in Singapore and London. A board that meets across four time zones. Payroll runs in PHP and SGD and GBP. Some of my team are billable on client work; others are internal. I file BIR Form 2316 in January. I file Singapore IR8As in March. I file UK PAYE returns when I have UK contractors that quarter.

I’m not unusual. Almost every Filipino professional services firm I know — consulting, law, accounting, IT services, marketing — is structured this way now. The pure-PH-employees model died sometime around 2020. We’re all running mixed teams. We’re all paying people across borders. We’re all dealing with compliance in multiple jurisdictions.

What’s unusual is how badly the available software handles this.

What I Tried

Before I built Steer Workforce, I tried everything.

Sprout. Solid PH product. The largest local SaaS for HR and payroll. They’ve been doing this since 2015 and they do it well — for employees on local PH payroll. The moment I tried to model my Singapore-based developer or my UK-based BD lead, the system fell over. Foreign contractors aren’t a use case. They’re an edge case the documentation doesn’t cover.

JustPayroll. Cheaper, simpler. Works fine for a 10-person retail business. Wasn’t built for project-driven professional services. No concept of billable hours flowing to client invoices. No way to model bill rates per project.

Salarium. I used them briefly in 2022. By August 2023, their disbursement service had collapsed. By December 2023, they suspended operations indefinitely. Customer data was deleted during the suspension. ₱15M of customer funds was frozen. The CEO vanished. I migrated out before the worst of it, but I watched friends lose months and a lot of money.

Generic global tools — Deel, Remote, Rippling. These handle foreign contractors well. They speak Filipino-as-a-foreign-country well. But they don’t understand BIR. They don’t generate 2316 in the format BIR actually accepts. They handle PH labor as an afterthought. When DOLE changes a rule, they don’t notice for six months.

So I had a choice: pick the PH tool and bolt on three other things for the foreign contractor stuff, or pick the global tool and bolt on three other things for the PH stuff. Either way, three tools and a spreadsheet.

The Structural Problem

Here’s what I figured out during the bolt-on years.

The timesheet that drives my client invoices is a different timesheet from the one that drives my payroll.

This sounds like a small thing. It isn’t. It’s the entire problem.

When my consultant Maria logs 8 hours against client A on Monday, two things should happen: her client invoice line should accrue at her bill rate, and her payroll cost should accrue at her cost rate. Same hour, two financial events.

What actually happens: Maria logs her hours in Toggl. Toggl exports to a spreadsheet at month-end. The spreadsheet feeds my invoice generator. Separately, our HR person logs the same Maria’s salary cycle in Sprout. Sprout has no idea Maria worked 8 hours on Monday for client A. Sprout knows Maria gets paid ₱85,000 per month. The two systems don’t agree, don’t talk, and don’t know they should.

Now multiply that by 50 consultants, 20 active projects, and three currencies. Now compute project margin. Your project margin is wrong. Not by a little. Structurally wrong. The inputs come from different systems that don’t know about each other.

You can’t fix this with discipline. You can’t fix it with a Zapier connector. You can’t fix it by hiring an analyst. You fix it by making the timesheet itself the source of truth — for both jobs.

That’s the architectural insight. Everything else in Steer Workforce flows from this one thing.

The Decision

I decided to build Steer Workforce in late 2023, around the time Salarium was collapsing. The convergence was useful: customers were paying attention to data sovereignty for the first time. There was clear daylight where the incumbent had been.

Three architectural commitments at the start, none of which I’ve broken.

Deterministic core, AI shell. Payroll math, statutory contributions, BIR forms — these are computed by deterministic engines. Anything else is a liability. AI explains, generates, classifies, and proposes. AI never computes the payroll itself. The first version of this rule was scribbled on a notepad. It hasn’t moved.

Country packs. Compliance is engineered country by country. PH first, because that’s where I live and where the compliance is hardest to get right. Singapore and Malaysia next. Then we’ll see. The country pack is its own architectural unit; we don’t graft features onto a one-country product. We build for many; we ship one; we add the next.

Dogfood publicly. The same payroll runs that pay my BlastAsia and Xamun teams are the same ones we ship to customers. If something is broken in the product, my team is the first to know. If something is right, I have the conviction of someone who uses it daily. It’s the cheapest, most honest QA process I know.

What I Want to Be True

There’s a thesis underneath all of this that’s bigger than one product.

Filipino software companies have spent 25 years selling manpower. We’re good at it. We built a giant outsourcing industry on the back of competent, English-speaking, hardworking developers. The compensation structure was outcome-blind: hours billed, not value created.

That model is breaking. AI is commoditizing the work that outsourced developers used to do. The customer doesn’t want hours anymore; they want outcomes. They want a working product, a closed deal, a deployed system. They want to pay for that, not for time.

This is bad news for the manpower-supplier model and good news for the product-builder model. The Philippines should be exporting software products, not just software labor. We have the technical skills. We have the English. We have the time-zone overlap with both ASEAN and the US. What we haven’t had is the cultural confidence to claim “Made in Philippines” on a global SaaS product.

Steer Workforce is one attempt. There should be a hundred. PSIA, where I’m a Director, is working on changing this. AIM, where I teach Adjunct, is producing the next generation of founders. The Speed2Seed venture studio is funding them. Founder Institute Manila is mentoring them. The pieces are coming together.

What’s Next

If you run a 20–150 person professional services firm in the Philippines and you’re tired of three tools that don’t agree on the same number, join the waitlist. No demo gate. No credit card. When early access opens, you can import your team from Sprout, JustPayroll, or a spreadsheet, and run a sandbox payroll. If it doesn’t replace three tools and reduce your monthly reconciliation to zero, walk away.

Or book 30 minutes with me directly. I do these calls personally for the first hundred customers. After that, we’ll see.

— Arup