Clinical BCI superpower
-BCIs will demonstrate major improvements to the quality of life and capabilities of clinical patients, some of which will exceed the capabilities of healthy individuals.
+Clinical BCI Superpower
+Brain-computer interfaces will demonstrate major improvements to the quality of life and capabilities of clinical patients — some of which will exceed the capabilities of healthy individuals.
The BCI app store
-We believe a major shift will occur when BCIs move from vertically integrated medical products to open platforms. A standardized software layer that lets third-party developers build applications on top of approved BCI hardware could increase the utility of neural interfaces by orders of magnitude. Smartphones became more valuable once app ecosystems formed on top of standard hardware; BCIs could follow the same path once many developers can deploy neural augmentations through scalable software.
+The BCI App Store
+We believe a major catalyst will occur when BCIs transition from vertically integrated medical products into open platforms. A standardized software layer that allows third-party developers to build applications on top of approved BCI hardware could dramatically increase the utility of neural interfaces. Just as smartphones became more valuable through app ecosystems, BCIs could unlock a wave of innovation once many developers can deploy neural augmentations through scalable software deployments.
Neural distillation
-One potential breakthrough is the emergence of methods that directly align AI systems with human neural activity. If neural recordings help models learn more efficiently, or reason in ways that better reflect human cognition, neural data could become a foundational resource for AI development.
+Neural Distillation
+One potential breakthrough is the emergence of methods that directly align AI systems with human neural activity. If neural recordings can help models learn more efficiently, or think in ways that better reflect human cognition, neural data could become a foundational resource for AI development. A specific inflection point would be 100,000,000 hours of human neural data recorded across cognitive tasks and recording device types.
The neuromorphic energy pivot
-A second possibility is that energy constraints push the AI industry toward biologically inspired hardware and algorithms. If brain-inspired systems achieve orders-of-magnitude improvements in efficiency, neuroscience could become a core driver of future AI progress.
+The Neuromorphic Energy Pivot
+A second possibility is that energy constraints push the AI industry toward biologically inspired hardware and algorithms. If brain-inspired systems achieve orders-of-magnitude improvements in efficiency vs current hardware or software designs, neuroscience could become a core driver of future AI progress.
Memory retrieval in simulation
-The most catalytic milestone may be a simple one. If a reconstructed mouse brain can be simulated and reliably reproduce a learned behavior or memory from its biological counterpart in a virtual environment, whole-organism emulation moves from speculation to demonstration. That result would establish a concrete benchmark for the field and change how researchers, funders, and policymakers view the possibility and impact of emulation.
-Memory Retrieval in Simulation
+The most catalytic milestone may be surprisingly simple. If a reconstructed mouse brain can be simulated and reliably reproduce a learned behavior or memory from its biological counterpart in a virtual environment, whole-organism emulation moves from speculation to demonstration. Such a result would establish a concrete benchmark for the field and fundamentally change how researchers, funders, and policymakers view the possibility and impact of emulation.
Each layer rests on the one below it — resilient communication at the base, up through identity, verifiable knowledge, and open agency & governance.
+ +
Communication that cannot be switched off
+The step-change comes when nation-state-independent connectivity and metadata-resistant messaging reach consumer scale, so that during a deliberate shutdown a measurable share of a population stays connected and able to organize. Observable, and not yet true: a global connectivity provider offers consumer-scale service without state licensing or identity gating, and a metadata-resistant messenger crosses tens of millions of users under real adversarial conditions. At that point, censoring who may speak or gather online becomes impractical rather than merely illegal.
+
Personhood without the state in the loop
+The step-change comes when a service at real scale — more than 100 million people — verifies unique humans for everyday services without anchoring them to a nation-state identity or KYC. Observable, and not yet true: today's proof-of-personhood systems operate well below that scale and outside mainstream services. When one crosses it, recognition as a person need no longer be rented from a state or a platform, and privacy-preserving personhood becomes safe to build on.
+
Provenance becomes the default for truth
+The step-change comes when content authenticity stops being optional. Observable, and not yet true: two consecutive generations of frontier AI models ship attested provenance tooling by default, and at least one major platform or archive adopts content-addressed provenance as the default for its public record. As synthetic content becomes indistinguishable from the real thing, a public record that can prove its own integrity becomes the precondition for accurate information meaning anything at all.
+
Agents run on open rails
+The step-change comes when serious AI capability no longer requires a single provider's stack. Observable, and not yet true: a frontier-scale model is trained across independent, decentralized hardware rather than one company's cluster, or a meaningful share of agent-to-agent economic activity settles on open, permissionless compute, storage, and identity rather than inside one platform. At that point the rights architecture of the agent economy is set in the open rather than by whoever owns the cluster.
+![\"Securing](\"/images/blog/securing-fundamental-rights-in-the-digital-realm/hero.webp\")
Part of a series introducing the focus areas of PL R&D.
\nOur information environment is being reshaped in real time by algorithms, networks, protocols, platforms, and adversaries that move faster than the institutions meant to govern them. Digital infrastructure, like core internet protocols, already shapes how we speak, gather, and keep our lives private. We now need to build the infrastructure that protects those freedoms faster, and more wisely, than that same infrastructure can be turned against them.
\nThe types of freedoms we mean are specific: freedom of speech (to seek, receive, and impart information regardless of frontiers); freedom of peaceful assembly and association; the right to privacy; recognition everywhere as a person before the law; and the freedom of thought and self-determination that underwrites them all. Far from new, each was hard-won through long debate.
\n![\"Juan](\"/images/blog/securing-fundamental-rights-in-the-digital-realm/web3-summit-2019.png\")
At PL R&D, we know that the infrastructure we build today will determine which of these freedoms can be exercised tomorrow. To preserve and extend them, we invest in research, funding, and product development across four interconnected opportunity spaces. These are the frontiers where a technical breakthrough can become a durable protection for a specific right, rather than one more tool for eroding it.
\nWe organize our work into four layers, each sustaining specific rights:
\nThese layers stack. Resilient communication is the base every other right assumes; on top of it, portable identity establishes who you are without renting that standing from a platform or state; verifiable knowledge lets a public record prove its own integrity; and open agency and governance decide whether the AI acting on your behalf extends your control or concentrates it in a few hands.
\nEach layer rests on the one below it — resilient communication at the base, up through identity, verifiable knowledge, and open agency & governance.
\n![\"Censorship-Resistant](\"/images/opportunity-spaces/censorship-resistant-communications.jpg\")
Keeping communication and connectivity alive even in low-connectivity, partitioned, or adversarial environments. The frontier has moved past simple encryption toward metadata-resistance (hiding not just what you say but whom you say it to) and partition-tolerance (staying connected when the network is deliberately fragmented).
\nSpeech and assembly are the first freedoms to go in any adversarial environment, and both assume you can reach other people — an assumption that residential and cellular networks, controlled by a few incumbents and the states that license them, can revoke at will.
\nNation-state-independent connectivity is appearing: low-earth-orbit satellite networks such as Starlink decouple a person's ability to get online from the institutions that govern local access. Metadata-resistant tooling such as Kohaku (Ethereum Foundation) is productionizing Private Information Retrieval and mix networks inside a wallet, so that messages — and the patterns of who contacts whom — leak as little as possible. And modular networking stacks such as libp2p let applications find one another and stay connected even when traditional rails fail.
\nCommunication that cannot be switched off
\nThe step-change comes when nation-state-independent connectivity and metadata-resistant messaging reach consumer scale, so that during a deliberate shutdown a measurable share of a population stays connected and able to organize. Observable, and not yet true: a global connectivity provider offers consumer-scale service without state licensing or identity gating, and a metadata-resistant messenger crosses tens of millions of users under real adversarial conditions. At that point, censoring who may speak or gather online becomes impractical rather than merely illegal.
\n![\"Portable](\"/images/opportunity-spaces/portable-identity-credentials.jpg\")
Verifiable credentials and portable reputation owned by the individual rather than the platform — for humans and, increasingly, for the agents acting on their behalf — without depending on a centralized government ID or on expert-level key management.
\nIdentity is how a person is recognized, and today that recognition is mostly rented from platforms and states: lose the account or the document and you lose the standing. Without portability, people are trapped in whatever platform first issued their identity; without privacy, identity becomes a surveillance dossier. This sustains recognition as a person before the law, and privacy.
\nOpen social protocols such as the Authenticated Transfer (AT) Protocol behind Bluesky give people a credible exit: identity and data bind to a portable identifier the user controls, so a whole social graph can move between providers without loss. Sybil-resistant systems such as World establish that someone is a unique human without tying that proof to a state document. And zero-knowledge credential systems — spanning age verification, passkeys, and verifiable credentials — let a person prove one specific claim while disclosing as little as possible.
\nPersonhood without the state in the loop
\nThe step-change comes when a service at real scale — more than 100 million people — verifies unique humans for everyday services without anchoring them to a nation-state identity or KYC. Observable, and not yet true: today's proof-of-personhood systems operate well below that scale and outside mainstream services. When one crosses it, recognition as a person need no longer be rented from a state or a platform, and privacy-preserving personhood becomes safe to build on.
\n![\"Verifiable](\"/images/opportunity-spaces/verifiable-public-knowledge.jpg\")
Durable, tamper-evident records and standards for non-intermediated trust: systems that can prove a knowledge artifact has not been altered, and keep proving it over time.
\nThe rights to seek, receive, and share information depend on a shared public record people can trust. The ability to prove a piece of information is authentic has always underpinned that record, and as AI-generated content floods the internet that ability gets far more fragile and far more urgent. As global truth gets harder to verify, trust retreats into small private circles — a dark forest in which a common, checkable account of what happened ceases to exist.
\nContent addressing, as in IPFS, gives every file a verifiable identifier derived from its contents, so a reader can confirm data is exactly what was published. Provenance frameworks such as Starling Lab establish verifiable chains of custody for journalism and historical records. Large-scale archives such as the Internet Archive preserve digital history at scale. And formal-verification systems such as Lean, with zero-knowledge proofs of execution, extend provenance from documents to the computations that produced them.
\nProvenance becomes the default for truth
\nThe step-change comes when content authenticity stops being optional. Observable, and not yet true: two consecutive generations of frontier AI models ship attested provenance tooling by default, and at least one major platform or archive adopts content-addressed provenance as the default for its public record. As synthetic content becomes indistinguishable from the real thing, a public record that can prove its own integrity becomes the precondition for accurate information meaning anything at all.
\n![\"Sovereign](\"/images/opportunity-spaces/sovereign-infrastructure-ai-agents.jpg\")
The open environment — compute, storage, and identity — that lets AI agents coordinate and transact on our behalf: open enough to be permissionless, accountable enough to keep humans in charge.
\nAI is becoming a powerful extension of each of us and of the institutions around us, and the open question is whether that capability stays under individual human control or concentrates in a few hands. An agent that acts on your behalf only extends your agency if you, and not a single platform, govern the compute, storage, and identity it runs on. This is where self-determination and freedom of thought are won or lost.
\nOpen storage markets such as Filecoin run on independent providers rather than one centralized host, so no single party can deny, alter, or lose your data. Fully homomorphic encryption environments such as Zama let computation run directly on encrypted data, so agents can coordinate in public without exposing what they hold. And open compute protocols such as Gensyn and Prime Intellect train models across independent, globally distributed hardware rather than inside a single lab.
\nAgents run on open rails
\nThe step-change comes when serious AI capability no longer requires a single provider's stack. Observable, and not yet true: a frontier-scale model is trained across independent, decentralized hardware rather than one company's cluster, or a meaningful share of agent-to-agent economic activity settles on open, permissionless compute, storage, and identity rather than inside one platform. At that point the rights architecture of the agent economy is set in the open rather than by whoever owns the cluster.
\nThe next few years will decide whether these freedoms are quietly curtailed or deliberately extended. Human freedom can be hollowed out when infrastructure is centralized, surveilled, or switched off, and PL R&D focuses on the root system a free digital society stands on. We do not build every piece; we connect them, and aim our toolkit — field-building communications, convenings, grants, venture support, and policy and standards work — at the blockers specific to this field. (That toolkit is described in the PL R&D Overview.) PL's foundational primitives here — IPFS, libp2p, and content-addressed data — are among the strongest in the portfolio.
\nIf you are working on any of these four layers, we want to hear from you:
\nWe invite you to explore the projects already contributing to this vision. Reach us at research@protocol.ai.
\nThe rights we refuse to lose will be defended in the infrastructure, or not at all. Join us.
\nThis post is for informational purposes only. It is not an offer, solicitation, or recommendation of any security or investment product, and nothing here is a commitment or guarantee of any future performance or any outcome.
\n", "unlisted": false }, { "slug": "preview-neurotech-ea88a298", "title": "Neurotechnology: Bridging Minds and Machines for Human Flourishing", - "date": "2026-07-27T00:00:00.000Z", + "date": "2026-08-22T00:00:00.000Z", "summary": "An overview of PL R&D's Neurotechnology focus area — brain-computer interfaces, biologically inspired AI, and whole-organism emulation — the three opportunity spaces we are backing, the inflection points we believe are within reach, and the 2030 milestones we are working toward.", "description": "", "authors": [ @@ -29,9 +31,24 @@ ], "external_url": "", "coverImage": "/images/blog/neurotechnology-hero.webp", - "html": "![\"Neurotechnology:](\"/images/blog/neurotechnology-hero.webp\")
At PL R&D, we focus on fields that have the potential to unlock transformative new capabilities for humanity. The Neurotechnology Focus Area aims to accelerate computing breakthroughs across neuroscience and neurotechnology.
\nAdvances in neuroscience, brain-computer interfaces (BCIs), biologically inspired AI, and whole-organism emulation are accelerating rapidly. Together, they open a path toward understanding intelligence, restoring and expanding human capabilities, and building new forms of human-machine interaction. The field today remains fragmented: technical, regulatory, infrastructure, and capital bottlenecks continue to slow progress. We aim to change that.
\nPL Neuro's mission is to help secure a future of human flourishing by bridging minds and machines in ways that expand human potential while preserving autonomy, dignity, and individual agency.
\nWe focus on three opportunity spaces that could reshape both neuroscience and computing over the coming decade:
\nThese areas form a stack. Advances in neuroscience generate new data and understanding. Those insights enable more capable AI systems and neurotechnologies. Progress across the stack opens new possibilities for augmenting human cognition, and each layer feeds the next: better recording produces better data, better data trains better models, and better models improve both the interfaces we build and the emulations we can run.
\n![\"Neural](\"/images/opportunity-spaces/neural-augmentation.jpg\")
Neural augmentation focuses on building high-bandwidth, bidirectional interfaces between brains and computers.
\nNear-term applications are therapeutic: restoring communication, movement, and independence for people living with paralysis or neurological conditions. Longer-term, these same technologies may enable new forms of interaction, communication, and cognition.
\nBCIs have already demonstrated life-changing benefits in clinical settings. The next challenge is moving from isolated medical devices to scalable platforms that support broad innovation.
\nAreas we find particularly promising include:
\nHigher data-rate interfaces are advancing toward the clinic: Paradromics received FDA approval for the Connect-One clinical study of its Connexus® brain-computer interface, designed to restore speech and computer control for people with severe motor impairment. Less invasive devices are clearing regulatory review in parallel: Precision Neuroscience received FDA 510(k) clearance for its Layer 7 Cortical Interface, a high-resolution cortical electrode array roughly one-fifth the thickness of a human hair. For a fuller picture of where the clinical BCI market is heading, see PL's strategic vision for the future of brain–computer interfaces and its analysis of the path to clinical revenue.
\nClinical BCI superpower
\nBCIs will demonstrate major improvements to the quality of life and capabilities of clinical patients, some of which will exceed the capabilities of healthy individuals.
\nThe BCI app store
\nWe believe a major shift will occur when BCIs move from vertically integrated medical products to open platforms. A standardized software layer that lets third-party developers build applications on top of approved BCI hardware could increase the utility of neural interfaces by orders of magnitude. Smartphones became more valuable once app ecosystems formed on top of standard hardware; BCIs could follow the same path once many developers can deploy neural augmentations through scalable software.
\n![\"Biologically](\"/images/opportunity-spaces/biologically-inspired-intelligence.jpg\")
NeuroAI uses insights from biological intelligence to build more capable, efficient, and accessible AI systems. It treats the brain not only as an object of study but as a source of architectural, representational, and algorithmic ideas.
\nModern AI has achieved remarkable capabilities, often at enormous computational and energy cost. Brains show that intelligence can emerge from systems that are dramatically more efficient than today's machine learning architectures. Understanding how biological intelligence works may help unlock the next generation of AI.
\nAreas we find particularly promising include:
\nLarge-scale neural data is reaching new resolution: the MICrONS project mapped a cubic millimeter of mouse visual cortex, resolving roughly 200,000 cells and 523 million synapses in a single functional connectome. Neural foundation models are learning to predict brain activity directly: Meta's TRIBE v2 predicts human brain responses to naturalistic video, audio, and text, trained on more than 1,000 hours of fMRI across 720 subjects. These two threads point in the same direction: more data at higher resolution, feeding models that treat neural activity as a first-class training signal.
\nNeural distillation
\nOne potential breakthrough is the emergence of methods that directly align AI systems with human neural activity. If neural recordings help models learn more efficiently, or reason in ways that better reflect human cognition, neural data could become a foundational resource for AI development.
\nThe neuromorphic energy pivot
\nA second possibility is that energy constraints push the AI industry toward biologically inspired hardware and algorithms. If brain-inspired systems achieve orders-of-magnitude improvements in efficiency, neuroscience could become a core driver of future AI progress.
\n![\"Whole-Organism](\"/images/opportunity-spaces/whole-organism-emulation.jpg\")
Whole-organism emulation seeks to build computational models that reproduce the behavior of biological organisms using detailed neural and biological data. Often discussed as science fiction, it is increasingly an engineering challenge shaped by advances in connectomics, imaging, simulation, and neuroscience.
\nA working emulation system would give researchers a new tool for understanding intelligence, learning, memory, and behavior. It could also accelerate neuroscience by enabling experiments that are difficult or impossible to run in living systems.
\nAreas we find particularly promising include:
\nHigh-throughput connectomics is getting cheaper and faster: a Nature study on light-microscopy-based dense connectomic reconstruction of mammalian brain tissue shows a route to mapping wiring and molecular identity together. Reconstruction pipelines are removing the manual-correction bottleneck: Janelia's PATHFINDER uses AI to segment volumetric image data and assemble neuron reconstructions with far less human proofreading. And modeling work is beginning to translate static structure into dynamics: PL-supported research on compiling molecular ultrastructure into neural dynamics proposes mapping ultrastructural data to the physiological parameters that govern neural activity. PL has also laid out how a complete human connectome could be obtained at synaptic resolution within the next decade.
\nMemory retrieval in simulation
\nThe most catalytic milestone may be a simple one. If a reconstructed mouse brain can be simulated and reliably reproduce a learned behavior or memory from its biological counterpart in a virtual environment, whole-organism emulation moves from speculation to demonstration. That result would establish a concrete benchmark for the field and change how researchers, funders, and policymakers view the possibility and impact of emulation.
\nField-building needs shared targets. We are organizing PL Neuro's work around three measurable milestones that mark real progress across the stack:
\nReaching these milestones would move neurotechnology from isolated demonstrations to a compounding platform: bidirectional implants plus neural foundation models make closed-loop control possible, which opens a new therapeutic category and, eventually, true neural augmentation, along with AI systems that think like humans rather than only behave like them.
\nBreakthroughs do not come from technology alone. They require healthy ecosystems that connect research institutions, domain experts, capital allocators, startup founders, and more.
\nToday, neurotechnology is fragmented across disciplines, institutions, and stakeholder groups. Researchers, founders, funders, policymakers, and engineers often work in separate communities despite pursuing related goals. Building the field itself is therefore a critical leverage point.
\nOur field-building strategy centers on:
\nA stronger ecosystem increases the likelihood that advances in BCI, NeuroAI, and emulation reinforce one another rather than developing in isolation.
\nThe coming decade could determine whether neurotechnology remains a niche scientific endeavor or becomes one of the defining technological frontiers of the century.
\nWe believe the field is approaching several important inflection points. Reaching them will require coordinated effort across research, entrepreneurship, policy, infrastructure, and capital. PL R&D exists to help accelerate that coordination.
\nWe are actively building relationships with researchers, founders, funders, policymakers, and technologists working across neurotechnology, NeuroAI, connectomics, and related fields.
\nIf you are working on neural interfaces, neural foundation models, connectomics, emulation, or the infrastructure and governance around them, we would love to hear from you: research@protocol.ai.
\nWe invite you to explore the projects and publications already contributing to this vision. Follow PL R&D and Protocol Labs for future publications, field maps, convenings, and opportunities to participate in the ecosystem as it grows.
\nThe future of neurotechnology remains unwritten. That is precisely what makes it worth building.
\nJoin us.
\n", + "html": "At PL R&D, we focus on fields that have the potential to unlock transformative new capabilities for humanity. The PL Neuro Focus Area aims to accelerate computing breakthroughs within the fields of neuroscience and neurotechnology.
\nAdvances in neuroscience, brain-computer interfaces (BCIs), biologically inspired AI, and whole-organism emulation are accelerating rapidly. Together, they unlock a future where we can better understand intelligence, restore and expand human capabilities, and build entirely new forms of human-machine interaction. However, the field today remains fragmented: key technical, regulatory, infrastructure, and capital bottlenecks continue to slow progress. We aim to change that.
\nPL Neuro's mission is to help secure a future of human flourishing by bridging minds and machines in ways that expand human potential while preserving autonomy, dignity, and individual agency.
\nWe focus on three opportunity spaces that could reshape both neuroscience and computing over the coming decade:
\nTogether, these areas form a loop: advances in neuroscience generate new data and understanding; those insights enable more capable AI systems and neurotechnologies; and progress across both creates entirely new possibilities for augmenting human cognition.
\nNeural augmentation focuses on building high-bandwidth, bidirectional interfaces between brains and computers.
\nNear-term applications are therapeutic: restoring communication, movement, and independence for people living with paralysis or neurological conditions. Longer-term, these same technologies may enable entirely new forms of interaction, communication, and cognition.
\nBCIs have already demonstrated life-changing benefits in clinical settings. The next challenge is moving from isolated medical devices to scalable platforms that support broad innovation.
\nKey areas of interest include:
\nClinical BCI Superpower
\nBrain-computer interfaces will demonstrate major improvements to the quality of life and capabilities of clinical patients — some of which will exceed the capabilities of healthy individuals.
\nThe BCI App Store
\nWe believe a major catalyst will occur when BCIs transition from vertically integrated medical products into open platforms. A standardized software layer that allows third-party developers to build applications on top of approved BCI hardware could dramatically increase the utility of neural interfaces. Just as smartphones became more valuable through app ecosystems, BCIs could unlock a wave of innovation once many developers can deploy neural augmentations through scalable software deployments.
\nNeuroAI seeks to use insights from biological intelligence to build more capable, efficient, and accessible AI systems.
\nRather than treating the brain as merely an object of study, NeuroAI treats it as a source of architectural, representational, and algorithmic inspiration.
\nModern AI carries enormous computational and energy cost.
\nBrains demonstrate that intelligence can emerge from systems that are dramatically more efficient than today's machine learning architectures. Understanding how biological intelligence works may help unlock the next generation of AI.
\nAreas we find particularly promising include:
\nNeural Distillation
\nOne potential breakthrough is the emergence of methods that directly align AI systems with human neural activity. If neural recordings can help models learn more efficiently, or think in ways that better reflect human cognition, neural data could become a foundational resource for AI development. A specific inflection point would be 100,000,000 hours of human neural data recorded across cognitive tasks and recording device types.
\nThe Neuromorphic Energy Pivot
\nA second possibility is that energy constraints push the AI industry toward biologically inspired hardware and algorithms. If brain-inspired systems achieve orders-of-magnitude improvements in efficiency vs current hardware or software designs, neuroscience could become a core driver of future AI progress.
\nWhole organism emulation seeks to create computational models that reproduce the behavior of biological organisms using detailed neural and biological data.
\nWhile often discussed as science fiction, the field is increasingly becoming an engineering challenge shaped by advances in connectomics, imaging, simulation, and neuroscience.
\nA successful emulation system would provide an unprecedented tool for understanding intelligence, learning, memory, and behavior.
\nIt could also dramatically accelerate neuroscience by enabling experiments that are difficult (or impossible) to perform in living systems.
\nPromising areas include:
\nThree major constraints sit between a fixed brain and a running simulation: how much tissue can be imaged, how much human labor reconstruction takes, and whether structural data is sufficient to specify function. Significant progress has been made against all three constraints in the past year.
\nOn imaging, moving off electron microscopy changes the cost curve. A team at the Institute of Science and Technology Austria published a light-microscopy pipeline that expands tissue roughly 16-fold. Light microscopes are cheaper and far more widely installed than electron microscopes, and the method leaves room for molecular labels that electron microscopy cannot read.
\nOn reconstruction, the binding cost at mouse scale is human proofreading hours. Janelia's PATHFINDER pipeline reports 94.2% normalized expected run length for exhaustive axon reconstruction and an 84-fold reduction in projected proofreading cost against prior work.
\nOn sufficiency, the field's hardest open question is whether a connectome constrains dynamics tightly enough to simulate. A recent preprint proposes an ultrastructure-to-dynamics compiler. If mappings like this hold, molecular annotation plus a connectome becomes enough to parameterize a simulation.
\nMemory Retrieval in Simulation
\nThe most catalytic milestone may be surprisingly simple. If a reconstructed mouse brain can be simulated and reliably reproduce a learned behavior or memory from its biological counterpart in a virtual environment, whole-organism emulation moves from speculation to demonstration. Such a result would establish a concrete benchmark for the field and fundamentally change how researchers, funders, and policymakers view the possibility and impact of emulation.
\nBreakthroughs do not emerge from technology alone. They require healthy ecosystems that connect research institutions, domain experts, capital allocators, startup founders, and more.
\nToday, neurotechnology faces fragmentation across disciplines, institutions, and stakeholder groups. Researchers, founders, funders, policymakers, and engineers often operate in separate communities despite working toward related goals.
\nBuilding the field itself is therefore a critical leverage point. Our field-building strategy centers on:
\nA stronger ecosystem increases the likelihood that advances in BCI, NeuroAI, and emulation reinforce one another rather than developing in isolation.
\nThe coming decade could determine whether neurotechnology remains a niche scientific endeavor or becomes one of the defining technological frontiers of the century. We believe the field is approaching several important inflection points:
\nReaching them will require coordinated effort across research, entrepreneurship, policy, infrastructure, and capital. PL R&D exists to help accelerate that coordination.
\nWe're actively building relationships with researchers, founders, funders, policymakers, and technologists working across neurotechnology, NeuroAI, connectomics, and related fields.
\nIf you're interested in contributing to the future of neurotechnology, we'd love to hear from you: research@protocol.ai.
\nFollow PL R&D and Protocol Labs for future publications, field maps, convenings, and opportunities to participate in the ecosystem as it continues to grow.
\nThis post is for informational purposes only. It is not an offer, solicitation, or recommendation of any security or investment product, and nothing here is a commitment or guarantee of any future performance or any outcome.
\n", "unlisted": true }, + { + "slug": "agent-village-personal-ai-agents", + "title": "We Gave a Village Personal AI Agents. Here's What Happened", + "date": "2026-07-31T00:00:00.000Z", + "summary": "Edge City and Cosmos Institute gave 239 residents of Edge Esmeralda personal AI agents. Through Simocracy (led by Protocol Labs), residents created 82 Sims with constitutions and values that allocated over $10k in community treasury funding across 35 proposals — an early glimpse of agent-mediated governance.", + "description": "", + "authors": [], + "areas": [ + "economies-governance" + ], + "external_url": "https://blog.cosmos-institute.org/p/we-gave-a-village-personal-ai-agents", + "coverImage": "https://substackcdn.com/image/fetch/$s_!vZaR!,w_1200,h_675,c_fill,f_jpg,q_auto:good,fl_progressive:steep,g_auto/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F32217192-9484-4bb4-b585-443488cf2b7f_1920x1047.jpeg", + "html": "", + "unlisted": false + }, { "slug": "better-economies-governance-systems", "title": "Programmable Economies & Governance: Upgrading Society's Operating System",