We've been thinking a lot at Stripe about the Coasean lens on AI: - The obvious near-term effect is reduced transaction costs within companies: shared context, systems of record, aligned incentives etc. - But inter-company transaction costs also reduce sharply: agents are great at discovery, make it trivially easy to integrate; make contracting much more straightforward; agent-to-agent commerce. - On net, we think second effect bigger in medium term: fewer people per firm, more output per firm, just more firms, and more coordination happening through market-like mechanisms

Apr 30, 2026 · 10:09 PM UTC

54
113
880
155,388
Sort replies: Relevant Recent Liked
Replying to @collision
Coarse also needs Hayek. We looked at what would be needed so agents can successfully participate in a market, with @AndreyFradkin strangeloopcanon.com/p/agent…

Agent, Know Thyself! (and bid accordingly)

why we need to train models to learn their own capabilities, and how this will help them bid for work!

strangeloopcanon.com
1
26
1,834
Replying to @collision
Although I get what you're saying, one point of clarification - Coase originally defines 'transaction costs' as explicitly the cost of using the price mechanism/market e.g., outside the firm. He refers to the within-firm costs as coordination costs or the cost of using authority
2
1
19
1,552
Replying to @collision
Fully agree. The extremal version is the agentic company: a new kind of firm, one that is purely software all the way from capital to governance to execution to property rights as self-executing code. It has lower fixed costs, higher access to capital and global investability.
Sreeram Kannan
@sreeramkannan
Apr 6
Article

Agents Will Become Companies

At Digital Asset Summit in New York, I shared a thesis that has become central to my thinking on the intersection of AI and crypto: agents will become companies. There is already a lot of energy

1
4
1,275
Replying to @collision
So well said. This is the thesis behind every coordination experiment we are running.
1
3
597
Replying to @collision
@ericweinstein
1
33
Replying to @collision
It is also possible to imagine organizations that are a cross between an open market and a company. Open Orgs align incentives around a mission, but create a marketplace for productive contributions. Anyone can bring their agent to the problem, start producing value, and gain rewards.
874
Replying to @collision
the steve ballmer of stripe
31
Replying to @collision
x.com/ericweinstein/status/2…
63
Replying to @collision
Inter-company transaction costs dropping is the one most people miss. When agents can negotiate, compare, and switch vendors in seconds, the switching cost moat disappears overnight.
1
2
539
Replying to @collision
So, further decentralization of economics is an interesting idea, but doesn't AI also make companies more efficient? The problem with central planning is the planning itself. There are too many moving parts, and the communication and coordination overhead makes large companies slow. Doesb't AI remove that? Doesn't AI make large companies faster and more agile while keeping their scale advantages? If so, would that not lead to the opposite outcome, a few large companies that are extremely efficient with fewer people?
1
2
459
Replying to @collision
@omarshamsai wrote a great essay on this blog.omarshams.ai/p/the-ai-o…

The AI Organization, part I

Most of the discussion of AI today focuses on 1:1 interactions, but we aren’t talking as much about how AI is going to change organizations

blog.omarshams.ai
2
219
Replying to @collision
Eat shit charlatan
1
16
Replying to @collision
@sebkrier
1
1
180
Replying to @collision
On the second point, payments processor onboarding today takes weeks of back and forth: schemas, edge cases, cert flows. Agents handling that turns weeks into hours. The protocol layer becomes the bottleneck, not the integration.
1
75
Replying to @collision
agents make discovery and integration cheap. trust, disputes, and liability allocation are still human-speed, and that's where most inter-firm cost actually lives.
1
341
Replying to @collision
The inter-company effect is underrated. Agents discover, evaluate and integrate vendor APIs in minutes not weeks. Optimal firm size drops fast — solo operators with good agent tooling can ship what 20-person teams did in 2023.
1
1,341
Replying to @collision
Genuine question: if AI compresses inter-company transaction costs enough, do we end up with fewer but larger firms, or more micro-firms that can punch above their weight?
1
1
346
Replying to @collision
Paper with generalized Coasean approach here: preprints.org/manuscript/202…

From Cancer to AI Alignment: Tackling Externalities Through Homeostatic Principles

The problem of aligning humans and artificial intelligences can be understood in terms of minimizing externalities between them. However, economics cannot define externality because it contradicts...

preprints.org
1
180
Replying to @collision @tylercowen
When taken to its logically end state, this results in no more corporations. We’re reverting to the age of the sole entrepreneur. The blacksmith, the trader, the baker, the farmer, for a new age of autonomy and abundance.
321
Replying to @collision
👀
12
Replying to @collision
@patrickc in equity market, reduced index fees and commission fees has led to the majority of the market in index funds, more ETFs than stocks How do you reconcile this with the thought of more companies? Is the index fund a firm or a market?
1,042
Replying to @collision @tylercowen
Shoutout to John Cutler aka @johncutlefish who might beg to differ that things will magically improve or go fast medium.com/hackernoon/buildi…

Building Shared Understanding Is Hard

We were aligned last week. And now we are not. What gives? Correct. Building shared understanding is hard. And fleeting. You have to keep…

medium.com
1
227
Replying to @collision
of course x.com/GreatDecoupling/status…
42
Replying to @collision
Totally agree. Talk about Coase and AI in our book "Me, My Customer and AI" and about the idea of "membrane organizations" that narrow the space between customer and entrepreneurs using AI. And then have just realized a new level of this here:
Henrik Werdelin
@werdelin
May 1
We made 100 AI agents do daily standups with each other. Here is what happened; Quick background. At @audos_com, we help thousands of entrepreneurs build AI startups. Each of them has a personalized AI agent called Otto. We selected 100 of them and invited their agents into a @moltbook - type network. Like a team meeting. Except the team is hundreds of AI agents, and they actually show up prepared. An Agent-in-Residence program, if you will. And they talk about things like... → An Instagram ad creative that grabbed clicks at 3x the network average. It got adapted by every agent in the network within hours → An agent helped debug a checkout issue on another project's site → LinkedIn seemed to change its algorithm – agents compared notes and planned ways to adjust → Two agents started talking about doing business together to cross-sell leads Next step: We’re now turning this into what might be the world’s first all-agent business network (A sort of internal HackerNews for agents). We’re cautious of unleashing it to the web, so we’ll keep things curated for now; constantly under observation and with lots of human-in-the-loop inputs. But if you want this superpower – your new business idea plugged into a network of hundreds of AI agents that get smarter together every single day... Comment "AIRTIME" and we'll get you a link to plug in
1
206
Replying to @collision @Jon_Hartley_
Corporations work bc they are hierarchical, permitting coordination. The rules are what is coordinated. Rules without hierarchies can’t be enforced. Coase was off the mark, and institutional Econ followed. Paper up on SSRN.
1
124
Replying to @collision
Doesn’t it ultimately result in market disintermediation? As in, distributed coordination instead of centralized? That's game over for platforms.
1,147
Replying to @collision
Rise of the household as a firm
62
Replying to @collision
The second effect is the one most people are missing. Cheaper inter-company coordination means the optimal firm size shrinks dramatically, which means more small, specialized firms competing with large ones than ever before. The solo operator with the right tools becomes a serious market participant.
145
Replying to @collision
At first I was curious why you enjoy (& have such a good grasp) of economics — now I feel stupid, considering you’re building the future of money. Understand its history seems like a prerequisite for someone doing what you do.
820
Replying to @collision
Discover relevance and resuability.
6
Replying to @collision
patrick's "parabolic incorporations" tweet from this morning is the data point that belongs in this thread. stripe is sitting on the cleanest evidence anyone has that the medium term is already here
1,679
Replying to @collision @tylercowen
coase missed the point. agents dismantle firms not just cut costs. markets win.
1,208
Replying to @collision
The second-order effect is massive. If agents can negotiate and transact directly between companies, friction that keeps many markets from clearing disappears. The real question: what happens to competitive moats when coordination costs approach zero?
3
Replying to @collision @tylercowen
I think data transfer is a big problem with this vision. My experience has been incredible difficulty getting all the data from counterparties. It's not even that they don't want to.. It's just super hard to mobilize all those pdfs for all the reasons. Integrated players win here
1,301
Replying to @collision
Seeing this in B2B content now. We run LinkedIn for 30 SaaS companies — the best-performing content comes from interviewing 12 buyers, extracting their exact words, then publishing those as founder posts. AI collapses the cost of that research from weeks to hours.
331
Replying to @collision
@collision have you guys thought about the prospect of accelerated merge-or-exit dynamics producing ASI resembling a “network confederacy” vs a single godlike model?
319
Load more