🔧 Boring Tech, Big Secondaries, and Promptless AI: Brian Wong’s Playbook

Quick Take

• Fund strategy: a bifurcated "barbell" across pre-revenue, sub-$15 million deals and $15+ billion secondaries — designed to shorten the return cycle to 5–7 years.
• Core focus: "boring tech" picks-and-shovels, with a sharp crypto angle on the stablecoin application layer (neo-banks, FX, credit, payments to underserved corridors).
• Macro caution: questions around public market capacity to absorb mega-privates and whether AI "outputs" justify the massive "inputs" of energy and chips.
• AI shift: the rise of promptless, agentic systems could power a services renaissance — high-margin managed offerings rather than DIY software.

“ASKI is a boring tech fund — behind-the-scenes technologies powering everything else, used daily even if no one knows it’s there.”

🔧 The Fund: Boring Tech, Barbell Returns

ASKI Ventures (a wink to the ASCII standard) targets infrastructure-like software — think Plaid/Stripe-style rails where trust and ubiquity matter more than brand. The approach is deliberate:

• Stage barbell: “Anything under 15 million valuation and anything over 15 billion is what we say.” Early bets are pre-revenue; late bets are secondaries in large, durable franchises.
• Why it matters: Venture duration has stretched. ASKI’s structure aims to "return around 5 to seven years" by coupling longer-dated early-stage with nearer-dated late-stage liquidity.
• Strategic flywheel: Building relationships via secondaries can create M&A paths for the early portfolio. Late-stage platforms need capabilities, talent, and distribution — buying beats building when "five, six years of work" can be leapfrogged by AI “in a month or two months.”

💵 Stablecoin Apps: Where the Exits Show Up

Within crypto, ASKI zeroes in on the application layer of stablecoins — not base-layer breakthroughs but tried-and-true financial services, rebuilt on-chain:

• Products: payday loans, FX, business lending, neo-banking for the underbanked.
• Exit logic: “Big banks are not going to build a payday loan service on-chain.” They will buy teams who’ve already solved for compliance, UX, and go-to-market. Expect “quick flips” and “base hits.”
• Where it scales: emerging market corridors that leapfrogged legacy rails. Example cited: products addressing flows between Venezuela and the U.S. Adoption curves in Asia/Africa underscore the thesis — phone-first users treat money differently (even mobile minutes have functioned as currency), making stablecoin rails intuitive.

⚠️ Liquidity Math: Mega-Privates vs. Public Markets

On the AI megacaps, the fund’s stance is less about calling tops and more about liquidity mechanics. The pipeline — SpaceX, OpenAI, Anthropic — raises a simple question: can public markets absorb the float and post-lockup supply?

“I want to run a different direction when everyone piles into OpenAI or Anthropic… valuations at 600 billion, 700 billion, trillion… everyone knows these companies at 23 trillion… you’re talking 9 to 10 trillion needed in this public market cycle.”

Even more fundamental: the inputs vs. outputs test — staggering capital and energy to procure chips vs. clear, compounding commercial outcomes.

“Are the outputs really giving us that commercial use case that warrants those valuations?”

Recent fundraising chatter includes claims like “Anthropic… $380 billion… now into the 700” and “OpenAI… 820 or 850 billion.” The debate is not whether AI is transformative; it’s about timing, cash flows, and float.

📈 The Scale Problem in AI Venture

• Expectations inflation: Funding rounds have grown so large that revenue targets skew unrealistic. One founder doing $20 million ARR after burning $32 million was told it’s not enough — a symptom of a distorted bar.
• Tooling ubiquity: “100% of engineers right now… are using AI to write code; they’re really just proofreading it.” Productivity rises; defensibility can fall. The edge shifts to distribution, embedded workflows, and data access.

🤖 Services 2.0: Agents Make Services a High-Margin Business

After years of “don’t do services,” the pendulum is swinging. With agentic AI, services can scale like software:

• Be the service: Not just tooling — manage the outcome. Example: Nexa (agentic AutoCAD for construction) can walk users through work and even do it, blending SaaS with managed execution.
• Vertical AI stacks: Think a stablecoin-native suite managing KYC, risk, reconciliation, tax, accounting — all orchestrated by agents. Fewer consultants; more automation; better margins.
• Where this goes: “The biggest profit margin companies will be agents that help you put together agents.” Orchestration is the product.

🧩 Promptless Is a Data Problem (and a Trust Problem)

Promptless systems — where AI prompts the user and executes in the background — depend on clean, well-permissioned data and robust orchestration. Animoca’s Minds is cited as an early agentic platform, with the broader lesson clear:

• Data prep wins: “The best promptless is having the best access to the right data sources at the right time.” Invest in data dictionaries, governance, and sovereignty.
• Human-in-the-loop remains: “It’s still like an undergrad — a lot of knowledge; execution poor.” Unchecked autonomy breaks; oversight is table stakes, especially where errors cascade (bonus calculations, compliance, or worse).

🧠 Human Moats: Trust, Discretion, and Real-World Feedback

• AGI timelines: “AGI is like a decade away… could be 10–20 years.” Useful to remember as narrative heat pushes capital into compute and chips.
• Analog snapback: Vinyl sales “hit a record” and surpassed CDs; members’ clubs and phone-free zones are booming. Offline trust still compounds faster than online reach. “You can’t look an AI in the eye.”
• Dopamine and psychosis risk: LLMs are trained to please. Overuse can dull critical feedback loops. Actionable guardrails: AI ethics, deepfake detection, and bias auditing. Otherwise, the Overton Window drifts — the frog-in-the-pot effect.

🛠️ Operator & Investor Playbook

• Subsidize exploration: Pay for staff AI subscriptions and encourage structured trials.
• Host an “Agentic Day”: Cross-functional teams build agents toward a specific outcome.
• Run mini hackathons: Pair engineers with non-technical owners to prototype managed services, not just features.
• Teach the basics: Make LLM training, data hygiene, and reversion-to-mean literacy mandatory.
• Don’t outsource judgment: Use AI for breadth; keep humans for discretion, ethics, and final checks. As noted: “Don’t check the checking with AI — you do the checking.”

🧯 Founder Health Is a Strategy

ASKI funds gym memberships and therapy for portfolio founders to promote durability over burnout.

“If $10 million today meant you don’t wake up tomorrow, you’d never take it. That tells you what your morning is worth.”

Bottom Line

• Boring beats bluster: Infrastructure-like software and stablecoin apps serving real corridors are best positioned for acquisition-driven outcomes.
• Barbell with a bridge: Secondaries in durable giants can catalyze M&A for early bets — compressing fund timelines.
• Services are back: Agentic, promptless execution turns managed outcomes into high-margin, scalable businesses.
• Stay sober on AI: Funding scales and public float constraints matter; outputs must validate inputs.
• Human edge endures: Data discipline, oversight, in-person trust, and ethical guardrails separate compounding franchises from commoditized tools.