We're a family office based in Singapore. We invest at Seed through Series B in the AI infrastructure most people don't pay attention to: model serving, the security of autonomous code, and the agentic plumbing nobody knows the final shape of yet.
We invest in the infrastructure layer of intelligent systems: the unglamorous parts that everything else compounds on. Deep technical work compounds into a real moat here, faster than most people expect.
The platform layer is being rebuilt. The winners will look unfamiliar.
Token budgets matter more than CPU cycles, and cold-start latency shapes your margin. The model gateways, vector stores, eval frameworks, and GPU orchestrators we'll all be using in three years are being written right now. Most of them by people who didn't exist in this category two years ago. That's where we want to be early.
Every AI deployment expands the attack surface.
Prompt injection is the new SQL injection. Model weights are intellectual property worth defending. Agents acting at machine speed mean defense has to move at machine speed too. We're talking to the teams throwing out the old assumptions, especially around identity for non-human actors and BPF-level runtime defense. The perimeter you used to defend doesn't exist anymore.
Copilots end at the chat window. Agents act.
Which means somebody has to build the sandboxes, the observability, and the audit trails that let a hospital or a bank turn an agent loose without losing sleep. That trust layer mostly doesn't exist yet. The founders we want to talk to are building it from scratch.
You can describe how your system actually works before you describe how big the market could be.
There's a non-obvious technical insight under the company. The kind that sounds boring until somebody tries to copy it.
You're playing the five-year game, not the five-quarter one.
There's a real reason your company couldn't have existed two years ago.
We're operators. The company we built and sold is what shapes how we read founders today, and how we plan to be useful four years in, when the contract math is harder than anyone tells you it will be.
Our due diligence is technical. We'd rather spend an afternoon at a whiteboard with you than open a deck. If your architecture has problems, it's cheaper to find them at the seed stage than at the Series B.
Our portfolio is small on purpose. We aim for meaningful ownership in each company. We're either in with conviction or we say no quickly, and you'll know which way we're leaning by the second meeting.
The intros we make tend to land somewhere. The GTM and M&A advice comes from years on the operator side, not from a playbook. We're around when there's something useful to talk about, and out of your way the rest of the time.
Family-office capital has no quarterly clock. Our horizon is the next ten or twenty years, not the next markup, so we can keep showing up through the dips other investors would punish you for.
| Stage | Seed to Series B |
|---|---|
| Check size | USD 500K to 5M (initial) |
| Sectors | AI infrastructure, cybersecurity, agentic systems |
| Geography | Global |
| Capital | Direct, family-office (no fund clock) |
| Vintage | 2024 deepq_001 |
I built a cybersecurity company before I started writing checks.
I wrote code for about a decade. Then I spent another decade selling it. Then I spent the back end of my last company navigating an acquisition into a Fortune-100 technology firm, which turned out to be its own discipline. By the end, the product was the easiest part.
DeepQ exists because of that trajectory. Shipping into security-critical markets, where trust is slow and the technical work doesn't forgive much, taught me what to look for at the seed stage. The deal-side experience taught me what kind of help a founder actually needs around year four, when the early excitement has worn off and the contract math starts getting ugly.
So my approach is plain. I read every email founders send. I look at the code when there's code to look at. I take my time saying yes. When I do say yes, expect me to be around for a long time.
Working on something in our focus areas? Lead with the technical thing nobody else sees yet. Keep it short. I read every email a founder sends.
hello@deepq.vc