Why we're here.
TLDR
The story of civilizational growth is one of increasing collaboration powered by trust at increasing scale. The ability to connect & trust at scale has enabled our networks to grow from local tribes to the global Internet.
Ideas, goods, and services flowing across the Internet depend on high trust interactions. Yet, AI has made manufacturing false identities, narratives, messages, software, goods, and services trivial. Amidst this war on trust, we are on the cusp of welcoming trillions of agents online. The highest impact agents must route through trust to accomplish goals that require co-ordination with third parties, procuring services, purchasing goods, and processing verified information.
At Outtake we have spent 3 years building products that defend the integrity of the Internet with the core insight that trust is an economics problem. A high trust network makes falsehood expensive to maintain & truth universally accessible. This way bad actors pay high costs & good actors interact fluidly within the network.
To accomplish this mission, we mapped the public Internet in depth - websites, authors, social content, software packages, and agents. On the behalf of companies like NVDIA, Anthropic, OpenAI and governments like the United States - we went to war with nation-state attackers, bot farms & malicious agentic swarms to ensure falsehood becomes expensive for bad actors. In doing so, Outtake uncovered the underlying trust graph of the Internet – blooming clusters full of impersonations, scams, fraud, and malware sitting in stark contrast to high integrity nodes of verified good actors & their infrastructure. The rapidly growing volume of agentic traffic now seeks accessible truth as it routes for those high integrity nodes which constitute the best counterparties, services, goods, and information.
Outtake’s mission is to build a high trust internet because we believe trusted agent & human collaboration will unlock the next chapter of civilizational growth. We accomplish this by building products which continuously make truth more accessible & falsehood more expensive on digital networks.
Yesterday
The story of civilizational growth is one of increasing collaboration powered by trust at increasing scale.
Through history, progress has been marked by our ability to co-ordinate in ever larger groups. We have moved from Families to Tribes. Tribes to States. States to Nations. Every leap is based on the idea that we can connect with, organize around, transact across, and ultimately trust ever greater sets of humans.
The original joy of the public Internet was to rip right past all borders, creeds, and religions to create a global network where the digital community transcended physical constraint.
Of course one cannot simply have nice things. Since the beginning of the internet we have also had – the bad guys.
A small set of people who decided that you connecting with them online was an opportunity to exploit you. Malware, scam websites, phishing emails, and sophisticated network attacks multiplied.
From petty criminals to nation-state spies, this set of bad guys & their tricks brought a chill to the utopian idea of the internet.
Cybersecurity emerged as an industry to restore order. Lock down the network, monitor the machines, secure the inboxes.
Security engineers around the world made sure that while everything globally became digitized over the decades – flights, hospitals, banks, and government – our lives kept moving smoothly.
Of the ~$125 Trillion global economy, roughly $85 Trillion depends directly on digital networks. Cybersecurity is the unsung hero of the modern world, and thus - it's achilles heel.
Today
In the last decade - slowly & then very quickly - AI models have gone from great writers of text, to great writers of code, to great writers of secure code. These capability jumps have led to superhuman cyber attack capabilities because it turns out writing secure code is quite close to understanding vulnerable code.
As a person planning for the future, you must ask - which kinds of attacks will dominate over the next decade? Where will the bad guys focus?
We expect two divergent trends - one short term (software battle) & one long term (trust war). AI will resolve the former & exacerbate the latter.
Short-term (the next 2 years) there will be an extremely turbulent period of securing software as frontier AI models discover & then assist in patching legacy tools. On the other side of this transition, we should see the majority of AI written code to be secure by default & production systems be aggressively red-teamed by trusted AI models for any chained exploits. Soon the number of software based exploits will asymptotically approach.
As a specific example - the famous OpenAI swarm hack into HuggingFace is made possible in part by the agents finding HuggingFace credentials that a Hugging Face developer accidentally committed to GitHub. An AI code review would catch this today & a superhuman AI code writer would never make this mistake.
Long-term if software is secure by default, it is very unlikely you will be able to hack your way inside via vulnerabilities. Instead you will need to persuade someone to open the door.
Deceiving whoever holds the keys – human or agent – to unlock the door will be the canonical digital attack.
In 2011, if you received a text from an unknown person you would assume it was a well intentioned message from an acquaintance whose number hadn’t yet been saved. In 2026 every unknown text is assumed to be a scam.
Now extrapolate this distrust of digital communication to every business interaction - emails from potential business partners to ads from new retail brands - it’s all suspect. We are surrounded by an ever increasing number of trust exploit attempts. If you are on the internet - human or agent - you will contend with this.
Our new stance of “default I don’t trust anything online” is corrosive to the fluid exchange of ideas, goods, and services in the digital world. Given civilization prospers based on our ability to trust & coordinate at ever greater scales, the incoming Trust War is a real limiting factor.
Compounding the problem, bad actors are benefitting from AI too. An exciting vibecoding platform builds impersonating login portals. An excellent sales outbound tool writes great phishing copy. A first class image generator creates perfect clone ads. An empathetic voice agent makes a phone call for that extra nudge on a target.
Bad actors & their AI powered tools will ensure that there are infinite public surface areas to entrap you. Which websites, software packages, social profiles, and agents you chose to interact with is an ever more serious decision. As the software exploit window closes, trust exploits will be the strongest possible adversarial attack so we expect absurd amounts of sophistication in this domain over the next decade.
The internet is becoming impossible to navigate without a machine readable understanding of authenticity. The solution will require simple access to digital trust.
It follows that trust will be an evermore scarce & valuable commodity in digital interactions. But can we concretely define trust? It turns out the solution is economic.
A high trust network consists of two features – falsehood is expensive (maintaining fake entities & information costs infinite money) & truth is accessible (discovering real entities & information is frictionless).
Making falsehood expensive describes the origins of Outtake’s first products which focused on the proactive discovery & automated takedown of impersonations, scams, and fraud. In 2023 we recognized that critical institutions must proactively ensure they are perceived as high trust.
It’s not sufficient to be the blue chip J.P Morgan Bank. Now you must actively ensure there is no one lurking around the corner also claiming to be J.P Morgan scamming your customers over text messages & wooing your employees to leak their internal credentials via fake web page. To solve this, you must go on the offensive - proactively discover impersonating infrastructure, tear it down at machine speed, and ultimately make it very expensive for a bad actor to impersonate you.
J.P Morgan’s ability to defend their public digital presence dictates whether customers, employees, business partners, and external AI agents can continue to engage quickly & safely.
At Outtake we entered the Trust War on the side of companies like NVDIA, Anthropic, OpenAI and governments like the United States. We take on nation-state attackers, bot farms & malicious agentic swarms to ensure falsehood becomes expensive for bad actors.
Institutions are now in an escalating arms race with bad actors to secure their own public integrity online which means constantly guarding against impersonating websites, software packages, social profiles, employees and agents.
In 2025, Outtake ran ~20 million autonomous cyber investigations to assist with the proactive discovery of impersonations, scams, and fraud. In 2026, we expect to run 10x that number at ~200 million.
Every investigation has meant Outtake uncovers the underlying trust graph of the Internet – blooming clusters full of impersonations, scams, fraud, and malware sitting in stark contrast to high integrity nodes of verified good actors & their infrastructure.
We believe future internet traffic increasingly depends on accessible truth as it routes for those high integrity nodes which constitute the best counterparties, services, goods, and information.
In parallel to defending their public presence – organizations will need to find high trust information & third parties to conduct business with. The increasingly artificial digital environment makes this incredibly hard.
Is this job applicant’s resume real? Is this trending product complaint legitimate or artificial promotion via bots? Should I install this software package my coding agent suggested? Which public AI agent can I interact with to transact with this company? Every single one of these is a question an Outtake customer has asked us.
The scope of workflows which depend on public digital information is enormous & easy to take for granted by those of us who grew up in a world where public information was trusted by default. The future will not resemble the past here.
These questions will best be answered by the verification of digital identities & behavior across many public surfaces. This is where Outtake’s trust graph approach shines because trust isn’t a static label & depends on the ongoing behavior of various nodes whose actions require constant update.
The ability to make this trust graph quickly & cheaply accessible is going to be incredibly important in a world where we expect internet traffic to 1000x as agents come online. These agents must make decisions grounded in reality & quickly route to trusted external agents for organizations to collaborate.
Tomorrow
2026 was the first year where humans became a minority on the internet. Agents multiply with every task, workflow, and datacenter – humans are bound to old fashioned procreation. Soon a human population in the billions will be accompanied by an agent population in the trillions.
Large swaths of important (it influences a human stakeholder), valuable (money is changing hands), or informative (it writes a fact about the world) actions will be made by agents across the internet.
For agents to make intelligent decisions in the real world they will need clarity on what information, entities, services, and goods they can trust online. We expect tremendous trust pressure here precisely because agents will increasingly seek out the third party agents who will be the digital front door to most goods, services, and information.
Outtake’s trust graph just so happens to be the first serious mapping of this set of agents, goods, services, and information. Future agentic traffic may navigate accordingly.
We are witnessing the next stage of civilizational growth as the potential GDP of the digital economy explodes with the arrival of an agentic internet. However, the AI-powered utopias of the future are all but guaranteed dystopias if we don’t solve the basic question of “can I trust this” in every digital interaction. For humans, agents, and organizations.
Outtake tackled the problem of trust with a simple premise - high trust networks make truth highly accessible & falsehood expensive to maintain. Every Outtake product built has re-enforced this idea while mapping the network itself.
We believe this product foundation enables low friction & high trust digital interactions to open the next chapter of the Internet.