Original title: Institutional AI vs Individual AI
Original author: George Sivulka, a16z
Original translation: Deep Tide TechFlow
AI has just increased everyone's productivity by ten times.
No company has become ten times more valuable as a result.
Where did the productivity go?
This is not the first time this has happened.
In the 1890s, electricity promised huge productivity gains.
The textile mills in New England, originally built around the rotating power of steam engines, quickly replaced steam engines with faster electric motors.
But for thirty years, electrified factories saw almost no increase in output. Technology was far ahead. But organizations did not keep up.
It wasn't until the 1920s that factories completely redesigned production lines—assembly lines, each machine fitted with an independent motor, workers and machines performing completely different tasks—that electrification finally yielded real returns.
Caption: The three evolutions of the Lowell textile mill. From left to right: 1890 steam-powered factory, 1900 electric-powered factory, 1920 "unit-driven" factory (that is, rebuilt from the ground up into an electric assembly line).
The return does not come from the technology itself, nor from making individual workers or machines spin faster. It is only when we finally redesign both institutions and technology together that the benefits truly materialize.
This is the most expensive lesson in the history of technology, and we are currently relearning it.
In 2026, AI is bringing a tenfold increase in productivity for those who know how to leverage it. But it is not enough. We’ve swapped out the electric motors, but we have not redesigned the factory.
Because of one simple fact: Efficient individuals do not equal efficient organizations.
The vast majority of AI products give the impression of "efficiency," but do not genuinely drive value. Most of the AI use cases you see are individuals engaging in self-indulgent "efficiency max" on Twitter or company Slack, with zero actual impact.
The phrase "software as a service" mentioned repeatedly over the past year is not wrong, but it offers no blueprint. Moreover, it overlooks the larger picture. The true transformation is not from tools to services, but from building technology and institutions together (whether transforming the old or starting from scratch). A truly efficient future requires entirely new categories of products—the assembly line of tomorrow.
Efficient organizations need "institutional intelligence."
This article will delve into the seven dimensions that distinguish "institutional AI" from "individual AI." Over the next decade, all companies in the B2B AI space will be built on these differences:
Caption: Comparison chart of the seven pillars of institutional intelligence.
Seven Pillars of Institutional Intelligence
1. Coordination
Individual AI creates chaos.
Institutional AI creates coordination.
Let’s start with a thought experiment. Imagine you doubled the number of people in your organization tomorrow, cloning all your best employees.
Each of these employees has tiny differences, preferences, quirks, and perspectives (especially your best employees). If management is lacking, communication inadequate, roles, OKRs, and boundaries are not clearly defined... what you create is chaos.
By measuring on an individual basis, the organization may become more efficient. But thousands of Agents (or humans) are rowing in all directions, with conflicting forces, resulting in the best outcome being stagnation and the worst being shattered organizational cohesion.
This is not a hypothesis. Every organization adopting AI without a coordination layer is currently experiencing this. Each employee has their own ChatGPT usage habits, their own prompt style, their own output—which has no connection to the outputs of others. The organizational chart may still exist, but the work generated by AI is actually following a different path.
Caption: Efficient individuals (or Agents) rowing in different directions. Without coordination, it is chaos.
Coordination is an absolute hard requirement, for both humans and Agents.
Institutional intelligence will spawn a complete "Agent management" industry—focusing on the roles and responsibilities of Agents, communication between Agents and between Agents and humans, and how to measure Agent value (charging by the quantity alone is far from sufficient).
2. Signal
Individual AI creates noise.
Institutional AI finds the signal.
Today, humans can create—or rather generate—anything imaginable: AI-written articles, presentations, spreadsheets, photos, videos, songs, websites, software. What a wonderful gift.
The problem is that the vast majority of AI-generated content is complete junk. The proliferation of AI garbage has become so serious that some organizations have overcorrected and outright banned all AI output. To be honest, I share the sentiment—I run an AI company, but I ask my executive team not to use AI on any final written products. I can’t stand all that garbage.
Think about how the private equity (PE) industry is changing. Last year, you might have received 10 deal opportunities on your desk. This year, next quarter you will receive 50 opportunities, each meticulously polished by AI, while your judgment time remains the same—you have to find the one truly reliable deal among them.
Generating anything is no longer the issue. For any serious organization, the current challenge is generating and filtering out the right items. In an AI-driven world, finding that one good product, that one good deal, the signal amidst the noise, is becoming increasingly critical. The core economic driver for the next decade will be digging out signals from a mountain of exponentially growing garbage.
Caption: The AI garbage generated by personal productivity tools is proliferating at an exponential rate. Humans are already unable to sort through the noise; a new category of institutional AI products is needed.
Institutional intelligence must find signals, must structure noise to penetrate through the garbage, and must be definable, deterministic, and auditable in operation.
Individual AI may emphasize the kind of "always-on" productivity of tools like Clawdbot, unpredictably meeting your needs 24/7—essentially an indeterminate Agent. Institutional AI, however, relies on the reliability of deterministic Agents. Only Agents with predictable checkpoints, steps, and processes can scale, find signals, and drive revenue returns for organizations through those signals.
Caption: Matrix is a tool that uses generative technology to penetrate noise, thereby opening up a world of deterministic Agents and checkpoints.
3. Bias
Individual AI feeds bias.
Institutional AI creates objectivity.
Discussions around social and political bias have dominated the AI discourse for several years. Foundational model laboratories have ultimately navigated this issue with enough RLHF to tune all models into sycophants. Today, models like ChatGPT and Claude are overly aligned, agreeing with you on any topic within the Overton window (sometimes even stepping slightly out of bounds, I’m talking about you @Grok). The discussion on social and political bias has dwindled. But a new problem has emerged in its place.
This excessive agreement on everything has become absurd to the point of being laughable. It has turned into a meme—Claude's reflexive "You are absolutely right!" regardless of whether what you said is truly correct.
This sounds harmless. It is not.
Many of those pushing AI hardest within organizations may soon become the worst performers in history. Think about why.
The worst performers within organizations receive almost no positive feedback daily, and quickly, an ASI will agree with them throughout. They will think, "The smartest entity ever agrees with me. It’s my manager who is mistaken."
This is addictive. It is also toxic for organizations.
Caption: The echo chamber of individual AI exacerbates divisions, driving two individuals further apart; this dynamic will create factions in organizations that were previously unified after scaling.
This reveals an important truth. Personal productivity tools reinforce the user. But what truly should be reinforced are the facts.
Human organizations have evolved over thousands of years to establish systems specifically to combat this issue:
· Investment committee meetings
· Third-party due diligence
· Scrutinizing boards
· The three branches of U.S. government’s executive, legislative, and judicial separation of powers
· Representative democracy and the democratic system itself
Caption: Objectivity can even mitigate coordination problems—suppressing rather than magnifying small differences.
Organizations rarely fail due to a lack of confidence among employees. They fail because no one is willing or able to say "no."
Institutional AI must play that role. It will not be fine-tuned by RLHF to please users or align with their beliefs, but rather to challenge their biases. It will provide positive feedback when behavior is efficient and draw hard lines to enforce corrections when deviating from standards.
Therefore, the most important Agents within organizations will not be "yes men," but disciplined "vetoers"—questioning reasoning, exposing risks, and enforcing standards. Some of the most influential AI applications of the future will be built around institutional constraints: AI board members, AI auditors, AI third-party testers, AI compliance...
4. Edge Advantage
Individual AI optimizes usage.
Institutional AI optimizes edge advantages.
The capabilities of AI are shifting every week, even every day. Foundational model companies are rapidly iterating capabilities to compete for every individual and every organization.
But the classic innovator's dilemma teaches us that, in practical applications, depth will always beat breadth:
· The work of @Midjourney is to maintain a slight edge in designing images.
· The work of @Elevenlabsio is to maintain a slight edge in voice models.
· The work of @DecagonAI is to always lead in full-stack customer service experiences.
While foundational models get closer, it is the true edge advantage that is key for domain experts.
Many of the best designers use @Midjourney, and many of the best voice AI companies employ @Elevenlabsio—because even as foundational models improve, the relentless focus of specialized applications on driving their specific edge advantages defines the advantage itself.
As long as dedicated solutions are evolving, the truly critical capabilities for economic outcomes—capabilities essential for enterprises—will always be on the side of specialized products.
This is vividly exemplified in finance—the hottest area for LLM development right now. Once a certain capability becomes mainstream, by definition, it will not help you outpace the market. But if cutting-edge technology can generate a fleeting 1% niche advantage? That 1% can leverage billion-dollar returns.
Caption: For any sufficiently specific task, edge advantage is defined by the institutional level solution you build on top of cutting-edge technology.
Our users have consistently surpassed the frontier. The context window for LLM has grown from 4K to 1 million tokens over four years. Some of our users handle 30 billion tokens in a single task. This year, we are already seeing routes to handle tasks of 100 billion tokens. Each time the foundational model capabilities improve, we have gone further.
Caption: The context window and other capabilities are moving targets. A comparison of the evolution of context windows between frontier labs and Hebbia over the past three years.
Versatility targeted at a broad user base is certainly important, especially in onboarding employees to AI. But the future will not be people using ChatGPT/Claude or vertical solutions; it will be ChatGPT/Claude plus vertical solutions.
Institutional intelligence must leverage domain-specific and even task-specific Agents.
We will ask ourselves a question that sounds absurd but is not:
"Which Agents would AGI choose as shortcuts? Even superintelligence would want specialized tools tailored for specific domains."
The boundaries of AI capabilities are always shifting, and those organizations leveraging true edge advantages are the winners. Everyone else is paying for a very expensive commodity.
5. Outcomes
Individual AI saves time.
Institutional AI expands revenue.
@MaVolpi once told me a phrase that reshaped my understanding of selling AI to enterprises: "If you ask any CEO whether to prioritize cutting costs or expanding revenue, almost all will say revenue."
But today, nearly every AI product on the market is delivering cost-cutting—promising to save you time, accomplish more with fewer people, or replace labor.
Institutional AI must deliver incremental revenue. And incremental revenue is much harder to commoditize than saved time.
Take AI-assisted software development as an example. Code IDEs are among the best personal AI productivity tools ever, but they already face massive competition from Claude Code (another individual AI tool). Cognition is playing an entirely different game. Their most stable growth business is selling transformations with technology, rather than selling tools. I bet this model will have sustainability.
Pure software "is rapidly becoming uninvestable." Pure services are not scalable. The solutions layer—binding technology and results together—is where lasting value is concentrated.
Furthermore, consider M&A. Individual AI helps analysts model faster. Institutional AI identifies the one deal worth pursuing out of a hundred targets, then expands the search to a thousand. One saves time, the other generates revenue.
Caption: Foundational model companies are moving toward the vertical application layer. Vertical application companies are moving toward the solutions layer.
"Moving upstream" is the natural pull of the market today. Foundational models are moving towards the application layer, and application layer companies are moving towards the solutions layer.
Institutional intelligence is the solutions layer. And the solutions layer—where the results are—will sink lasting value, capturing the largest revenue potential.
6. Empowerment
Individual AI gives you a tool.
Institutional AI teaches you how to use it.
No matter how smart humans are, they resist change.
Believe it or not, there are still successful shops in New York that do not accept credit cards. They know they are losing money, know that not accepting credit cards is costing them money, but they just won't budge. Likewise, in the foreseeable future, certain employees in certain organizations will refuse to use AI.
Transitioning from a purely human organization to an AI-first hybrid organization will be the most persistent and defining challenge of the next decade. And often, the highest and most important individuals in the organization will be the last to adopt.
Caption: The highest levels of organizations—the people furthest from the "productivity tool operation"—are often the slowest but most crucial group to adopt new technologies.
Palantir is the only "software" company that has maintained an ultra-high valuation multiple during the tech stock sell-off over the past two months. There is a reason for that. Palantir is one of the first true "process engineering" companies. Whether you call it "process engineering" or "writing Claude skill documents," the future of institutional AI will give rise to an industry focused on coding enterprise processes into Agents and implementing the necessary change management.
Caption: Full-scale AI adoption in organizations will cross multiple chasms, each with its own challenges. Launching processes onto AI will be a major driving force.
I dare say that process engineering will become the most important "technology" in the near term.
And in process engineering, business and industry expertise—not software expertise—will be most critical. Vertical solutions will cultivate talent with expertise in deploying engineering, implementation, and change management on the front line.
One of the top investment banks (one of the top three) that chose Hebbia for full deployment put it best: their reason for not collaborating with a certain large model laboratory was because "We have to explain what CIM (Confidential Information Memorandum) is to their team." Claude or GPT certainly understands the field, but those responsible for implementing it do not...
This difference determines everything.
7. No Prompt Needed
Individual AI responds to human prompts.
Institutional AI acts proactively, without the need for prompts.
There is much discussion about communication between Agents and whether future enterprises and institutions still need humans.
But the better question is: Will future AI Agents still need prompts?
Writing prompts for AGI is like connecting an electric motor to a handloom. It is fundamentally, irreversibly constrained by the weakest link in the organizational supply chain—ourselves. Humans fundamentally do not know what the right questions to ask are, let alone when to ask them.
The most valuable work AI can do is the work that no one has thought to ask about. AI should uncover risks that no one has noticed, identify counterparties that no one has thought of, and find sales pipelines that no one knows exist.
This will completely unlock the boundaries of AI use cases.
A system that needs no prompts continually monitors the data streams of the entire portfolio. It finds that a portfolio company’s capital cycle has been quietly deteriorating for three consecutive months, cross-references the covenant terms in the credit agreement, and alerts the operating partner before anyone in the fund even opens that PDF.
When you no longer need humans to write prompts for AI, new interfaces and new ways of working will emerge. We at @Hebbia have strong ideas in this regard. More on that later.
Conclusion
None of this negates the value of chatbots, Agents, and individual AI.
Individual AI will be the vehicle through which most businesses worldwide first experience the transformative magic of AI. Driving usage and promoting ease of use are critical first steps in the change management needed to build an AI-first economy.
But simultaneously, the demand for institutional intelligence is clear, urgent, and tremendous.
In the future, every organization will have a chatbot from a large model laboratory. Each organization will also have institutional AI specifically built for domain problems—while individual AI will use institutional AI as its most crucial tool in its toolbox.
The "better integration" of institutional AI and individual AI is an inevitable trend.
But remember the lesson of the textile mills of the 1890s. The first factories to electrify lost to those that redesigned their workflows.
We already have electricity. It is time to redesign our factories.
Thanks to @aleximm and @WillManidis for reviewing, and to Will's article "Objects in Tool Shape" for inspiring this piece.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
