Your accounting software, your CRM, your project management tool — all sold to you as efficiency gains. What most of them actually did was move the same administrative burden into a different interface. The paperwork didn’t disappear. It just got a new window. A startup called Lassie, backed by $35 million from Andreessen Horowitz in a round that closed in June, is betting the next wave of AI doesn’t rearrange the work — it eliminates it.
Lassie started in medical and dental practices, where the back-office admin problem is unusually severe: insurance reimbursements, payment reconciliation, appeals, and records scattered across portals that don’t talk to each other. It’s a narrow starting point. The bet behind the funding is that the same approach — an AI agent that runs a specific back-office function autonomously, not just assists a human doing it — generalises well beyond healthcare.
Why a Healthcare Starting Point Is Worth Watching, Not Ignoring
It’s tempting to read “healthcare AI startup” and assume it’s irrelevant to a logistics company or a retail SME in Singapore. That’s usually the wrong read. Healthcare gets picked as a proving ground precisely because the admin burden is extreme and the workflows are well-documented — it’s a hard test case, not a niche one. When these tools work in healthcare, the same underlying pattern — an agent handling a fully-scoped, repetitive back-office function end to end — tends to spread into any industry with a similar shape: high transaction volume, fragmented systems, and a human currently doing reconciliation work that a well-built agent can do without supervision.
Accounts receivable follow-up, supplier invoice matching, and claims-style processes in insurance or logistics all fit that shape. If your business has a role whose entire job is “check system A against system B and fix the mismatches,” that’s the kind of task this category of AI is being built to absorb.
What “AI Runs the Back Office” Actually Means in Practice
The distinction worth understanding is between AI that assists — drafts an email, suggests a category, flags an anomaly for a human to review — and AI that runs a process end to end with a human only handling exceptions. Most of the AI tools SMEs have adopted so far are the first kind. What Lassie and similar startups are betting real money on is the second kind becoming viable at SME scale, not just for enterprises with dedicated engineering teams.
That’s a meaningful shift if it holds up, because it changes what “automating the back office” costs to implement. Assistive AI still needs someone to review every output. Autonomous AI, done properly, needs someone to review the exceptions — a much smaller job.
The Caution Worth Holding Onto
A $35 million raise is a signal that sophisticated investors believe this is where AI is heading. It isn’t proof that any specific tool in this category is ready for your business today, and it definitely isn’t a reason to hand an unproven agent full autonomy over your accounts receivable on day one. The sensible approach is the same one that applies to every new AI category: identify the single back-office function costing you the most staff hours, test a tool against it with a human still reviewing the output, and only remove that supervision once the tool has earned it over a real stretch of time.
Why This Matters More in Singapore Than the Headline Suggests
Singapore’s own numbers back the shift up too: IMDA’s 2025 Singapore Digital Economy Report shows SME AI adoption jumping from 4.2% to 14.5% in a single year (2023 to 2024), with the digital economy overall now worth 18.6% of GDP. The back-office functions this category of AI is targeting — reconciliation, admin follow-up, cross-system data matching — are exactly the tasks that workflow automation already addresses at a smaller scale for most SMEs here. The direction of travel is the same; what’s changing is how much of that work a tool can eventually take on without a human in the loop for every step.
Enterprise Singapore’s Productivity Solutions Grant still co-funds automation tooling in this category, but only through pre-approved vendors listed on the Business Grants Portal, capped at 50% of cost or S$30,000. It’s also due to stop taking new applications on 29 September 2026, ahead of the incoming EDGE Grant — worth checking what the PSG grant covers now, while it’s still open.
About the Author
Keith Kwai is an independent consultant helping Singapore SMEs with digital transformation and AI adoption. He has 25 years of marketing and digital experience across global MNCs including Motorola, Singtel, Creative Technology, Epson, and Scholastic International. He builds agentic AI systems — not just advises on them. Connect at keithkwai.com or LinkedIn.
Sources: Andreessen Horowitz, investment announcement; IMDA Singapore Digital Economy Report 2025; Enterprise Singapore Productivity Solutions Grant (PSG) guidelines.