💫 Ramplify: March Momentum

Code Red: The Org Chart?, Anthropic vs. The Pentagon, Brain Cells Playing Doom

Mar 02, 2026

(Rampersand's Claude Coffee Melbourne session — founders, engineers, product, design and marketing leads sharing AI workflow best practices over coffee. hosted by Andrew Poesaste with AWS Startups.)

👋 Code Red: Your Team Is Too Big - by Andrew Poesaste

Last week, Jack Dorsey laid off 4,000 people at Block. The stock jumped 20%.

Whether this was truly AI-driven efficiency or just a disguised restructuring doesn’t really matter. What matters is the narrative: public market CEOs now have permission—and pressure—to radically rethink their org charts. The question isn’t “Will this happen?” It’s “How fast will it spread?”

If you’re a founder and you’re not treating this as a code red, your competitors already are.

What Changed Between December and Now

Every December at Rampersand, we run a hackathon. We write down every workflow, every process we wish we could automate, and spend a week trying to build them.

In 2024, we had a list of our top 10 priorities. Six of them were impossible—the ROI math didn’t work, the payback period was too long, or the engineering lift was too complex.

We revisited that same list in January 2025 with Claude Code and rebuilt our infrastructure. We can now do all ten. The ones that would’ve required custom APIs, multiple engineering sprints, and months of back-and-forth? Built in hours.

The payback period went from “probably not worth it” to “obviously worth it.” We’re saving roughly 20 hours a week, but the real shift isn’t time—it’s possibility. The hackathon is no longer a discrete week in December. It’s now a permanent Slack channel where we’re auditing every part of the business.

The Real Implications

The obvious story is efficiency. But the deeper shift is alignment.

Previously, getting customer feedback into product required a chain: customer success captures the call, writes it up, passes it to product, PM writes a PRD, hands it to engineering, engineer builds it, cycles back for feedback. 3-4 weeks, minimum.

Now? One PM can take the customer call transcript, use AI to extract bugs and features, spin up a working prototype with a real GUI, and push it back to the customer for feedback—all in an hour. Maybe engineering tweaks it before production, maybe they don’t.

When one person can do what used to take a cross-functional team, alignment friction disappears. You move faster not just because tasks are automated, but because the handoffs are gone.

We’re seeing this across our portfolio:

Teams replacing $10-15K SEO audits with Claude skills for under $100/month—and getting 90-95% of the value
Finance dashboards that would’ve been 6-12 month projects, costing $30-50K, rebuilt in days to work exactly how the team wanted
Customer enrichment and outreach that would’ve required Apollo or Clay subscriptions, now handled in-house

This isn’t about tools. It’s about rebuilding from first principles and asking: “What would we do if headcount wasn’t the constraint?”

The Revenue Per Employee Explosion

The old benchmark was $100K revenue per employee. Great firms trended toward $300K.

Gamma, the AI presentation tool, hit $2M per employee at their recent round. Claude Code is trending over $10M per employee.

These aren’t outliers anymore. They’re the new reference class.

What Founders Need to Do Now

If you’re early-stage: You can do far more with far less. The question isn’t “Should we hire for this function?” It’s “Can we build the capability first, then hire only when we’ve maxed out what’s automatable?”

We’re seeing founders delay hiring their first growth person, extra engineer, or PR lead not because they can’t afford it, but because they’re building internal infrastructure that gives them 80-90% of the capability without the headcount.

If you’re later-stage with existing teams: This is harder, but more urgent. You need to put a mirror on the company and audit every process. Reward the people pushing into new tools. Be honest about whether experience in “the old way” is now a liability.

The uncomfortable truth: you may have to choose between people who know the experienced way of doing things and people willing to go on the journey with you into AI-native operations.

For everyone: Stop hiring for outputs. Start hiring for infrastructure thinking. The best early hires now are multi-hyphenate generalists—engineers who want to talk to customers, business people who write code, high-agency operators who default to “Can I automate this?” before “Who should I hire for this?”

As Jenny Wen, Head of Design at Claude, puts it: hire three archetypes that I like: "strong generalists, deep specialists, and cracked new grads."

The Timeline

This isn’t theoretical. Your startup counterparts are already doing this. They’re compounding efficiency gains while you’re still thinking about it.

If you’re running a bigger team, the window is even shorter. Public markets now reward this narrative. Every CEO will face the question: “What are you doing with AI to become more efficient?”

This is a code red because the gap between teams moving on this and teams waiting is widening by the week.

If you’re building an AI-native company, I’d love to chat.

🗞️ Tech News:

Trump Blacklists Anthropic After Pentagon Standoff

What happened: President Trump ordered all federal agencies to “immediately cease” using Anthropic’s technology after the company refused the Pentagon’s demand to remove safeguards around mass surveillance and autonomous weapons. Defence Secretary Pete Hegseth designated Anthropic a “supply chain risk to national security.”

Anthropic CEO Dario Amodei held firm:

“These threats do not change our position: we cannot in good conscience accede to their request. We do not believe that today’s frontier AI models are reliable enough to be used in fully autonomous weapons.”

Hours later, OpenAI announced it had struck a deal with the Pentagon for classified networks. CEO Sam Altman said he shares Anthropic’s “red lines” on surveillance and autonomous weapons — but his agreement includes safeguards the Pentagon accepted.

Takeaway: The first major test of whether AI companies will hold ethical lines when billions are on the table. Anthropic chose principle over contract; the market will decide if that’s sustainable.

Related: Anthropic lets you transfer your ChatGPT memories into Claude in under a minute 🔄

Living Human Brain Cells Are Now Playing Doom

What happened: Australian biotech company Cortical Labs demonstrated its CL1 biological computer playing Doom using 200,000 lab-grown human neurons on a microchip. The neurons receive game data as electrical signals and respond with firing patterns that control the character’s movement and shooting.

The CL1 costs $35,000 per unit (or $20,000 each in 30-unit server racks) and draws only 850–1,000 watts. Neurons stay alive for up to six months in an internal life-support system.

“Is it an esports champion? Absolutely not. Right now, the cells play a lot like a beginner who’s never seen a computer. And in all fairness, they haven’t. But they show evidence that they can seek out enemies, they can shoot, they can spin.” — Dr Brett Kagan, Cortical Labs

Experts believe this moves biological computers closer to practical applications like controlling robotic arms. Notably, the chips learned faster than conventional silicon-based machine learning systems.

Takeaway: The “Can it run Doom?” meme meets synthetic biology. An Australian company is building actual biological intelligence — not artificial — that learns and adapts.

🛠️ How to Startup:

What VCs Actually Mean When They Say Your Startup Is “Too Early”

A decent proportion of founders have heard this phrase while pitching VCs: “You’re too early.” But according to Rampersand Co-founder and Managing Partner Paul Naphtali, copping this particular brand of rejection is rarely just about timing.

“’Too early’ is a product of something’s missing. It’s also a code for ‘we don’t necessarily fit… and I don’t want to have a long conversation’.”

Paul says founders deserve an investor who believes in the business, but it can be “very hard to say to people who put their heart and soul into a business, ‘it’s not for me’.”

What to do about it:

Ask yourself: “What is in the stack that I’m not demonstrating now? How do I demonstrate it?”
Push for clarity — but “don’t be offended if you get an answer that’s maybe not that helpful”
If you’re on a watch list, you’ll hear from the VC proactively with leads, observations, or questions

Takeaway: “Too early” often means “I can’t get confidence that you’ve thought about X.” Figure out what X is.

(L-R: Rachel Yang (Giant Leap), Paul Naphtali (Rampersand), Dan Krasnostein (Square Peg), and Leila Oliveira (Antler) at the SmartCompany Growth Summit, Melbourne.)

Proprietary Data Is the Only Real Moat Now

Larry Ellison made the point bluntly: as AI models converge on the same public training data, the only sustainable competitive edge is proprietary datasets.

This isn’t abstract. Every foundation model — Claude, GPT, Gemini — is trained on roughly the same corpus of internet text, books, and code. The models are getting better, but they’re getting better together. Differentiation at the model layer is compressing.

What this means for founders:

If your product can be replicated by anyone with API access to the same model, you don’t have a moat
The companies pulling ahead are the ones generating unique data through their product usage — every customer interaction, every workflow, every correction becomes training signal
First-mover advantage now compounds: more users → more proprietary data → better model performance → more users

The startups winning in 2026 aren’t chasing frontier models. They’re building exclusive datasets that no one else can access.

Takeaway: Stop asking “which model should we use?” Start asking “what data can we generate that no one else has?”

Your Fintech Pitch Sounds Exactly Like Everyone Else’s

Alan Jones breaks down a live pitch from Phinly founder James Horan and identifies the gap that makes most fintech pitches forgettable: they’re logical when they need to be surprising.

The core insight: The “did you know that…?” test. After your pitch, investors head home and their partner asks, “Was there anyone who stood out today?” Will they mention something you taught them?

“If the answer is no, your pitch is forgettable. Not bad. Just forgettable. And in a crowded room, forgettable is fatal.”

How to fix it:

Lead with a surprising insight — Find the counterintuitive number that makes an investor go quiet and think, “Wait, that’s me.”
Don’t read your slides — Your deck exists to support what you’re saying, not replace it. When you read slides, you’re presenting, not persuading.
Own a truth, not a category — The best companies don’t just solve a problem — they expose a truth that makes you feel slightly seen and slightly uncomfortable.

Takeaway: The pitch isn’t “we built an AI agent for X.” It’s “Here’s something true and weird and slightly embarrassing about how you do X. And we built the thing that fixes it.”

👕 VC in Australia/New Zealand:

How AI is Changing Venture Capital in Australia

KPMG’s Amanda Price moderated a panel with Rampersand Partner Andrew Poesaste, Justine Carzino (OneVentures), and Sarah Josey (KPMG Australia) on how AI is reshaping VC decision-making.

Key shifts the panel identified:

AI is table stakes — “If we rebuilt your product AI-first, would it look different? If not, you may have an ‘AI-washed’ feature, not a defensible business.”
Defensibility lives in workflows and data — Bolt-on chatbots won’t cut it. VCs want deep workflow integration or proprietary data moats.
Speed has broken old heuristics — Companies hitting $50–100M revenue in record time globally. The “$1M ARR then raise” playbook is increasingly irrelevant.
Market > Margins (early on) — For AI-native businesses, category dominance and retention trump near-term efficiency.

Takeaway: If you’re not AI-native, you need a clear answer to “why won’t you be disrupted?” Fundraising is now fluid and opportunistic — momentum matters more than calendar timing.

69% of Australian Organisations Now Using Autonomous AI Agents

Deloitte’s 2026 State of AI report shows Australian enterprises have moved past experimentation. More than two-thirds of organisations surveyed are now deploying autonomous AI agents — systems that take action, not just generate insights.

The shift from “AI for analysis” to “AI for execution” mirrors what we’re seeing in our portfolio: teams aren’t asking Claude to summarise documents anymore. They’re asking it to draft the response, send the email, update the CRM.

Takeaway: If 69% of Aussie enterprises are already using autonomous agents, the question for founders isn’t “will enterprises adopt?” — it’s “are you building for the way they’re already working?”

Australia’s $7 Billion AI Infrastructure Push

The federal government’s National AI Capability Plan commits $7 billion to AI infrastructure — including sovereign compute capacity — alongside $17 million for responsible AI adoption programs and $30 million to establish the AI Safety Institute.

This isn’t just policy signalling. It’s capex. The Firmus $10B data centre deal we covered last month, Neara’s unicorn raise, and this federal commitment all point to the same thesis: Australia is betting that AI infrastructure is strategic infrastructure.

Takeaway: Sovereign compute is no longer a policy talking point. It’s a funded priority.

Neara Becomes Australia’s Newest Unicorn with $90M Series D

What happened: Sydney-based digital twin company Neara raised $90 million, pushing its valuation to $1.1 billion. The round was led by TCV (backers of Netflix, Spotify, and Xero), with Square Peg, Skip Capital, EQT, and Partners Group participating.

Neara builds physics-enabled 3D models of power grids that help utilities identify underutilised capacity, accelerate renewable connections, and improve resilience. They work with nearly 90% of Australian network utilities and have expanded to Southern California Edison, CenterPoint Energy, and utilities across Europe.

“Power grids all over the world are reaching their limits under the combined pressure of AI, electrification, and rapidly rising demand.” — Jack Curtis, Neara Co-founder

This is Australia’s second unicorn of 2026 (after Gilmour Space) — already matching 2025’s full-year total.

Takeaway: As AI data centres strain power grids globally, Neara’s timing couldn’t be better. Infrastructure digital twins are suddenly a strategic priority.

🤩 Portfolio News

Parachute Raises $1.8M and Launches SME Legal Community

Big week for Parachute. The legal AI startup raised $1.8 million in a pre-seed round led by our Rampersand Partner Andrew Poesaste, with backing from Maxine Minter (Co Ventures), Cheryl Mack (Aussie Angels), and other strategic investors.

Before founding Parachute, Ryan Zahrai was running his own legal practice where he realised two things: he didn’t want to pay “stupid amounts” to Harvey, and enterprise legal AI vendors wouldn’t give him the attention he wanted as an SME operator.

“There’s a huge gap in the market for law firms like mine.” — Ryan Zahrai, Parachute co-founder

Parachute’s platform acts as a single workspace where firms can draft and review documents, collaborate with clients, and white-label AI-assisted legal tools — with lawyers remaining “human in the loop.”

Takeaway: Harvey and Legora captured enterprise. Parachute is going after the 80% of Australian law firms they’re ignoring.

Then they threw a party. Despite “stupid amounts of rain,” 90% of their 70 RSVPs showed up for Parachute’s first community event — bringing together sole principals and small-to-mid-sized commercial, corporate, family, and migration law firms.

“There was definitely a ‘yes, this is us’ sentiment shared. This won’t just be an event. We’re creating a community to cascade our collective pain points into what drives delivering a solution that helps us all do less of the stuff we hate doing.”

(This won't just be an event — we're building a community." Ryan Zahrai, Parachute)

💼 Portfolio Job Board

Product Manager at Cuttable

Location & type: Melbourne (Hybrid), full-time
What they do: Cuttable is a digital platform that simplifies the creation and management of ads. It streamlines post-production workflows by making versioning, approvals, and collaboration faster and more transparent.
What you’ll do: Own Cuttable’s AI-powered creative production platform end-to-end to improve how performance marketing teams create, publish, and optimise high-performing Meta ads. You’ll drive cross-functional delivery and continuous iteration.

Founding Head of Marketing - Voice AI at Cor

Location & type: Melbourne, full-time
What they do: Cor builds AI-driven voice and multi-modal agents like Obi that elevate customer onboarding and product adoption by guiding users through training and activation in real time. Its intelligent conversational agents replicate a human-led onboarding experience inside apps, helping teams scale onboarding, increase adoption, and reduce support burden.
What you’ll do: You’ll own Cor’s build-in-public and social presence (LinkedIn, X, niche B2B communities), partnering with creators to make Cor synonymous with “AI for product activation and adoption,” while shipping high-velocity content that ranks and resonates

Senior Software Engineer at PredictHQ

Location & type: Auckland, New Zealand, full-time
What they do: PredictHQ is a demand intelligence company that provides AI-powered event data to help businesses forecast demand fluctuations caused by real-world events.
What you’ll do: You’ll build, improve and maintain scalable APIs, data pipelines, and services using Python, FastAPI, Kafka, Elasticsearch and others. Comfortable with distributed systems and event-driven architecture (Kafka).

Do you have a job you’d like us to promote? Add it to Hatch and share the link with us.

See all jobs across the Rampersand Portfolio.

🔧 Tool of the Month

SkipUp — AI Scheduling Without the Link

Yes, it’s a portfolio company. We’re biased. We also genuinely use it.

SkipUp is an AI scheduling assistant that handles the back-and-forth without requiring calendar links. Simply CC skip@skipup.co on any email and it runs the conversation for you — collecting availability across time zones, sending reminders, negotiating a time, and booking automatically.

Why we love it:

No behaviour change required — you’re already CC’ing people on emails, now you CC Skip
Cross-timezone magic — we use it for Sydney ↔ SF scheduling constantly
Works 24/7 — follows up when you forget to

Unlike Calendly (passive, one-to-one links), SkipUp actively coordinates groups of 2–50 people asynchronously. It reasons through each situation, remembers preferences, and improves with use.

Congrats to founders Sasha Krecinic and Dheer Gupta on the launch 🚀

🔗 Sign up here

🌏 Rampersand Travel Diary:

Andrew is in Sydney next week (March 11–13) for VCs at Sea. Please reach out.

Taryn will be in Brisbane 9 - 13 March and Adelaide 17 - 20th March. Please reach out.

🎉 Upcoming Events

📍 ADELAIDE - Mar 17–19, 2026

SouthStart 2026

Australia’s premier startup and innovation conference returns to Adelaide. Chanie Hyde, Taryn Pieterse, and Rod Hamilton will be there — and we’re bringing GWI for the first time since 2023.

🔗 Get tickets

📍 SYDNEY HARBOUR - Mar 2026

Retreat26: VC at Sea

A unique gathering bringing together founders and investors aboard Sydney Harbour. Andrew Poesaste will be attending.

🔗 Learn more

📍 SYDNEY - Apr 30, 2026

Sunrise 2026

Sydney’s flagship startup event. Andrew Poesaste will be attending.

🔗 Learn more

Thanks for being part of the Rampersand Community! You can stay updated with even more news and info on our LinkedIn and X/Twitter pages.

Discussion about this post

Ready for more?