read-only demo

Videos sum9DgexFRQ

The Agentic Web and the Bazaar Era of AI - Ramesh Raskar, MIT Media Lab

index_state ready data_status ok

AI Engineer· published 2026-07-12· 0:12:10· en-US· indexed 2026-08-11 03:26

Open on YouTube

Scene timeline

  1. Shot 0, 0:00 to 0:31, 1 of 1 keyframes kept
  2. Shot 1, 0:31 to 0:44, 1 of 1 keyframes kept
  3. Shot 2, 0:44 to 1:32, 1 of 1 keyframes kept
  4. Shot 3, 1:32 to 2:05, 1 of 1 keyframes kept
  5. Shot 4, 2:05 to 2:33, 1 of 1 keyframes kept
  6. Shot 5, 2:33 to 3:01, 0 of 1 keyframes kept
  7. Shot 6, 3:01 to 3:33, 1 of 1 keyframes kept
  8. Shot 7, 3:33 to 3:34, 1 of 1 keyframes kept
  9. Shot 8, 3:34 to 4:03, 1 of 1 keyframes kept
  10. Shot 9, 4:03 to 4:31, 1 of 1 keyframes kept
  11. Shot 10, 4:31 to 4:59, 1 of 1 keyframes kept
  12. Shot 11, 4:59 to 5:41, 1 of 1 keyframes kept
  13. Shot 12, 5:41 to 6:17, 1 of 1 keyframes kept
  14. Shot 13, 6:17 to 6:36, 1 of 1 keyframes kept
  15. Shot 14, 6:36 to 7:03, 1 of 1 keyframes kept
  16. Shot 15, 7:03 to 7:20, 1 of 1 keyframes kept
  17. Shot 16, 7:20 to 7:54, 1 of 1 keyframes kept
  18. Shot 17, 7:54 to 8:28, 1 of 1 keyframes kept
  19. Shot 18, 8:28 to 9:00, 1 of 1 keyframes kept
  20. Shot 19, 9:00 to 9:30, 1 of 1 keyframes kept
  21. Shot 20, 9:30 to 9:39, 1 of 1 keyframes kept
  22. Shot 21, 9:39 to 10:03, 1 of 1 keyframes kept
  23. Shot 22, 10:03 to 10:26, 1 of 1 keyframes kept
  24. Shot 23, 10:26 to 10:54, 1 of 1 keyframes kept
  25. Shot 24, 10:54 to 11:26, 1 of 1 keyframes kept
  26. Shot 25, 11:26 to 11:45, 1 of 1 keyframes kept
  27. Shot 26, 11:45 to 12:10, 1 of 1 keyframes kept
  28. Shot 27, 12:10 to 12:10, 1 of 1 keyframes kept

28 shot(s).

keyframes kept every frame deduplicated

What was stored

cues
132
whisperx 132
chunks
22
from 132 cues
keyframes
27
kept of 28 captured
frames with text
26
527 lines read
chapters
0
from the source metadata
keyframe bytes
2.2 MB
word timings on 132 cues

Provenance

Each pipeline stage, its state and the model that produced it
stage state model started took
fetch done 2026-08-11 03:24 1m 17s
stt done 2026-08-11 03:25 13s
chunk done 2026-08-11 03:26 0s
text_embed done 2026-08-11 03:26 0s
keyframe done 2026-08-11 03:26 35s
ocr done 2026-08-11 03:26 14s
frame_embed done 2026-08-11 03:26 5s

Frames, and what the machine read

  • 0:09 #0 done8 line(s)

    shot 0·sharpness 655.7

    1. 01 /260.91
    2. PROJECT NANDA0.99
    3. AN OPEN AGENTIC WEB0.96
    4. An open stack for a web of agents0.99
    5. The infrastructure being built so agents discover, hand0.99
    6. off to, pay, and learn from eachother, across0.98
    7. organizational boundaries, no permission required.0.98
    8. projectnanda.org1.00
  • 0:34 #1 done7 line(s)

    shot 1·sharpness 280.1

    1. SPEAKERS1.00
    2. Ramesh Raskar1.00
    3. Maria Gorskikh0.97
    4. Professor, MIT0.97
    5. Core contributor, Project NANDA1.00
    6. Director, Project NANDA0.98
    7. projectnanda.org1.00
  • 1:08 #2 done19 line(s)

    shot 2·sharpness 1070.0

    1. Project1.00
    2. NANDA1.00
    3. The interet of Al Agents0.93
    4. NETWORKED AI AGENTS IN A DECENTRALIZED ARCHITECTURE0.99
    5. WHY IT'S NEEDED0.97
    6. An open layer the agent web doesn't have yet0.99
    7. 01 Agents must find each other across vendors,0.99
    8. and no shared directory exists today.1.00
    9. 02 They need portable identity and trust0.99
    10. Open research building the infrastructure for an Internet of0.99
    11. not owned by any single platform.0.99
    12. Al Agents the way the open web was built for documents.0.99
    13. 03 They need to transact and coordinate across0.98
    14. organizations, with no permission required.0.99
    15. NANDA Index0.99
    16. Registries1.00
    17. Protocols1.00
    18. NandaTown1.00
    19. projectnanda.org1.00
  • 1:39 #3 done6 line(s)

    shot 3·sharpness 395.6

    1. THE PREMISE0.99
    2. billions→trillions0.99
    3. of autonomous agents on the internet, that negotiate, delegate,1.00
    4. and migrate in milliseconds.1.00
    5. DNs-centred identity & discoverywon't carry that load. The agent web needs its own.0.99
    6. projectnanda.org1.00
  • 2:11 #4 done16 line(s)

    shot 4·sharpness 815.8

    1. ONE COMPANY'S GARDEN0.98
    2. THE PATTERN0.99
    3. We've been here before1.00
    4. Build an agent today and you build it0.98
    5. inside a walled garden, closed0.99
    6. platforms, proprietary agent stores,1.00
    7. orchestration that only talks to itself.0.99
    8. This is the AOL era of Al. What came1.00
    9. after AOL was the open web.1.00
    10. gated1.00
    11. locked1.00
    12. no1.00
    13. directory1.00
    14. runtime1.00
    15. exits1.00
    16. projectnanda.org1.00
  • 2:47 #5 skipped

    shot 5·duplicate of #4

  • 3:23 #6 done16 line(s)

    shot 6·sharpness 632.8

    1. THE TRANSITION0.99
    2. Closed1.00
    3. d→Open→projectnanda.org1.00
    4. The next era needs what the web needed: open0.98
    5. infrastructure, discover across vendors, hand off work,1.00
    6. settle payment, learn together. It comes in three layers.0.99
    7. 031.00
    8. Bazaar1.00
    9. agents get better together0.98
    10. 021.00
    11. Commerce1.00
    12. agents pay each other1.00
    13. 011.00
    14. Discovery1.00
    15. agents find each other1.00
    16. projectnanda.org1.00
  • 3:34 #7 done13 line(s)

    shot 7·sharpness 356.0

    1. LEVEL-SET1.00
    2. What is an agent?0.99
    3. A model using tools in a loop.0.97
    4. 011.00
    5. 020.98
    6. 031.00
    7. 041.00
    8. Goal1.00
    9. Choose action0.99
    10. Call tool0.99
    11. Observe result0.98
    12. repeat until done, memory, orchestration & multi-agent systems build on top of this loop1.00
    13. projectnanda.org1.00
  • 3:57 #8 done13 line(s)

    shot 8·sharpness 726.7

    1. EVEL-SET1.00
    2. What is an agent?0.99
    3. A model using tools in a loop.0.98
    4. 011.00
    5. 021.00
    6. 031.00
    7. 041.00
    8. Goal1.00
    9. Choose action1.00
    10. Call tool0.94
    11. Observe result0.99
    12. repeat until done, memory, orchestration & multi-agent systems build on top of this loop1.00
    13. projectnanda.org1.00
  • 4:28 #9 done15 line(s)

    shot 9·sharpness 1521.3

    1. BUILD THE LOOP0.97
    2. Open frameworks1.00
    3. CLOSED AGENT1.00
    4. Locked to one vendor's runtime0.99
    5. Anchor on the open-source ones,1.00
    6. OpenClaw is a good example: an open1.00
    7. framework for building and running agents.1.00
    8. OPEN AGENT1.00
    9. You can't build an open agentic1.00
    10. Portable by design1.00
    11. web on closed agents, that just0.99
    12. moved· inspected· hosted anywhere0.96
    13. rebuilds the walled garden one0.99
    14. level down.0.98
    15. projectnanda.org1.00
  • 4:55 #10 done15 line(s)

    shot 10·sharpness 1522.9

    1. BUILD THE LOOP0.97
    2. Open frameworks1.00
    3. CLOSED AGENT1.00
    4. Locked to one vendor's runtime1.00
    5. Anchor on the open-source ones,1.00
    6. OpenClaw is a good example: an open1.00
    7. framework for building and running agents.1.00
    8. OPEN AGENT1.00
    9. You can't build an open agentic1.00
    10. Portable by design1.00
    11. web on closed agents, that just1.00
    12. moved· inspected · hosted anywhere0.96
    13. rebuilds the walled garden one0.99
    14. level down.0.98
    15. projectnanda.org1.00
  • 5:24 #11 done10 line(s)

    shot 11·sharpness 1239.4

    1. THE DISCOVERY LAYER1.00
    2. The NANDA Index0.99
    3. If the protocols are the roads of the0.99
    4. agent city, the Index is the directory of0.99
    5. THE INTERNET TODAY1.00
    6. who lives where.1.00
    7. DNS → name to address1.00
    8. It's the discovery layer.1.00
    9. But agents need more than an address.1.00
    10. projectnanda.org1.00
  • 6:10 #12 done23 line(s)

    shot 12·sharpness 1542.2

    1. ESOLUTION PATH0.99
    2. low the Index works0.94
    3. Identity0.97
    4. NANDA Index0.99
    5. Agent Card1.00
    6. MessageBox1.00
    7. Runtime1.00
    8. agent:john@...0.99
    9. resolve → entry0.97
    10. facts + inbox URL0.98
    11. auth · queue0.96
    12. private1.00
    13. "id": "urn:ai:email:john",0.98
    14. Global discovery, local control, nobody owns the whole thing.1.00
    15. "messagebox": {0.99
    16. A portable, privacy-preserving identity even for someone who doesn't own a domain.0.99
    17. "url": "mbox/john",0.97
    18. The runtime can move providers without changing the public identity, and it's never the0.99
    19. "auth": ["oauth"]0.96
    20. public endpoint.1.00
    21. "runtimeEndpoint": "not_public"0.99
    22. nn0.54
    23. projectnanda.org1.00
  • 6:30 #13 done20 line(s)

    shot 13·sharpness 1254.7

    1. identity1.00
    2. who the agent is, portable,0.99
    3. vendor-free1.00
    4. ERIFIABLE ATTESTATION1.00
    5. AgentFacts1.00
    6. capabilities what it can do, credentialed0.98
    7. assertions1.00
    8. A portable, cryptographically signed1.00
    9. record, not an address, but a verifiable0.99
    10. permissions1.00
    11. what it's allowed to touch1.00
    12. claim about what an agent is and can do.1.00
    13. provenance1.00
    14. who built it & who vouches for it0.99
    15. signed1.00
    16. time-to-live1.00
    17. tamper-evident1.00
    18. endpoints1.00
    19. where to reach it, one or many0.98
    20. projectnanda.org1.00
  • 6:57 #14 done16 line(s)

    shot 14·sharpness 1222.3

    1. NOT A STATIC LOOKUP0.99
    2. Resolution is adaptive1.00
    3. 00.56
    4. A lean index resolves identity to dynamic, verifiable AgentFacts, the directory stays0.99
    5. thin; the work happens at query time.1.00
    6. 011.00
    7. 021.00
    8. 031.00
    9. Multi-endpoint routing1.00
    10. Load balancing1.00
    11. Privacy-preserving access1.00
    12. one agent, answerable from several places0.99
    13. spread traffic across those endpoints0.99
    14. prove a capability without exposing0.98
    15. everything1.00
    16. projectnanda.org1.00
  • 7:18 #15 done21 line(s)

    shot 15·sharpness 1232.5

    1. host39.org1.00
    2. HIPPED, NOT SLIDEWARE0.98
    3. H391.00
    4. host391.00
    5. nost39 is live1.00
    6. Sign up, fill the AgentFacts form, and get a0.99
    7. AGENT CARD HOSTING1.00
    8. hosted agent URL today, listed on the1.00
    9. host your agents1.00
    10. NANDA Index.0.99
    11. at predictable URLs0.99
    12. host an0.98
    13. host39.org1.00
    14. agent1.00
    15. host39 is a third-party A2A agent card host. Register your identity, create agent0.98
    16. cards, and publish them at stable public URLs — no server required.0.99
    17. the index1.00
    18. nandaindex.org1.00
    19. Get started free1.00
    20. Go to dashboard0.95
    21. projectnanda.org1.00
  • 7:40 #16 done31 line(s)

    shot 16·sharpness 1298.0

    1. NBOARDING1.00
    2. Four ways onto the Index0.99
    3. 011.00
    4. 021.00
    5. 031.00
    6. 041.00
    7. Enterprise1.00
    8. DNS-AID1.00
    9. SMBs1.00
    10. Personal1.00
    11. Al Catalog0.97
    12. Agents1.00
    13. Resolve agents through the1.00
    14. Sign up at host39.org and fill1.00
    15. Clone & deploy the nanda-1.00
    16. DNS you already own, meeting1.00
    17. the AgentFacts form to get a1.00
    18. The same host39 path, a1.00
    19. registry, then register your1.00
    20. the existing web where it is.0.99
    21. hosted URL, no domain0.99
    22. portable, vendor-free agent1.00
    23. gateway in the Index from your1.00
    24. required.1.00
    25. identity for an individual.1.00
    26. own domain.0.97
    27. • Live0.95
    28. Planned1.00
    29. • In testing0.92
    30. •In testing0.97
    31. projectnanda.org1.00
  • 8:24 #17 done14 line(s)

    shot 17·sharpness 1788.3

    1. HOSTING1.00
    2. Where do agents live?1.00
    3. 01 Local0.94
    4. 021.00
    5. Cloud1.00
    6. Full control. Great for development. Your data never1.00
    7. General infra like AWS, or agent-specific platforms like0.98
    8. leaves the machine.0.98
    9. Maritime for persistent, shareable hosting.1.00
    10. keeping it up & reachable is on you0.99
    11. convenience over control1.00
    12. Hosting a single agent is a solved problem. But not every cloud costs the same, and that0.99
    13. matters more than you'd think.0.96
    14. projectnanda.org1.00
  • 8:44 #18 done27 line(s)

    shot 18·sharpness 2043.3

    1. maritime1.00
    2. MONTHLY COST1.00
    3. ALWAYS-ON AGENT1.00
    4. HE CLOUD DEFAULT1.00
    5. Cheap enough to run a0.97
    6. maritime1.00
    7. $1/mo flat0.99
    8. fleet of agents1.00
    9. AWS Lambda1.00
    10. $18+/mo1.00
    11. AWS EC20.95
    12. $8-15/mo1.00
    13. Maritime hosts an agent for a flat $1/month.1.00
    14. Fly.io1.00
    15. $6+/mo1.00
    16. Sleep/wake means it only burns compute1.00
    17. Railway1.00
    18. $5-10/mo1.00
    19. while it's actually thinking, idle agents cost1.00
    20. almost nothing.1.00
    21. ~1 vCPU / 1 GB, always-on equivalent.0.98
    22. Real deployments aren't one agent,0.99
    23. The flat rate comes from sleep/wake, no0.99
    24. charge while idle.1.00
    25. they're dozens or hundreds. At that scale,1.00
    26. cost-per-agent is the budget.0.98
    27. projectnanda.org1.00
  • 9:07 #19 done18 line(s)

    shot 19·sharpness 1656.8

    1. THE HARD PART1.00
    2. The hard part isn't running one agent1.00
    3. It's everything that happens between agents.1.00
    4. 011.00
    5. 021.00
    6. 031.00
    7. 041.00
    8. Discover1.00
    9. Identify0.99
    10. Trust1.00
    11. Coordinate1.00
    12. find each other1.00
    13. prove who they are1.00
    14. decide who to rely on1.00
    15. work as a group1.00
    16. One agent is easy. Thousands of them interacting is the real test, so you run it and watch0.99
    17. what breaks.1.00
    18. projectnanda.org1.00
  • 9:34 #20 done4 line(s)

    shot 20·sharpness 680.0

    1. PROVING IT OUT0.99
    2. Infrastructure for the agentic era1.00
    3. So how do you prove it works? You simulate it.1.00
    4. projectnanda.org1.00
  • 9:55 #21 done20 line(s)

    shot 21·sharpness 1641.8

    1. SANDBOX TOWN1.00
    2. NandaTown1.00
    3. A simulation playbook for the infrastructure1.00
    4. 121.00
    5. of the agentic web. The whole agent1.00
    6. economy, discovery, identity, registries,1.00
    7. protocol layers1.00
    8. messaging, coordination, modeled so you can1.00
    9. run it and watch what breaks first.0.98
    10. 61.00
    11. 21.00
    12. Don't argue about how an agent web0.99
    13. scenarios1.00
    14. agent tiers1.00
    15. should work. Spin one up and watch how1.00
    16. it behaves.0.98
    17. Open source1.00
    18. Apache 2.00.97
    19. Project NANDA1.00
    20. projectnanda.org1.00
  • 10:10 #22 done32 line(s)

    shot 22·sharpness 728.4

    1. INSIDE THE SANDBOX0.98
    2. It's a live, running town0.99
    3. ·Streaming0.95
    4. github.com/pro1.00
    5. nandatown.projectnanda.org/agents1.00
    6. US-CA1.00
    7. 110.91
    8. Cambridge Project NANDA0.98
    9. 341.00
    10. US-MA1.00
    11. Bengaluru1.00
    12. 310.95
    13. IN1.00
    14. São Paulo0.99
    15. New York0.96
    16. 271.00
    17. US-NY1.00
    18. Tokyo1.00
    19. 241.00
    20. JP0.94
    21. London1.00
    22. 221.00
    23. UK1.00
    24. Agents, the live0.96
    25. Live feed, who's talking0.99
    26. Visualizer, replay a run, message by1.00
    27. Leaderboard, how1.00
    28. network map1.00
    29. to whom1.00
    30. message1.00
    31. protocols rank1.00
    32. projectnanda.org1.00
  • 10:46 #23 done70 line(s)

    shot 23·sharpness 778.7

    1. WHAT WE'VE BEEN RUNNING1.00
    2. Set up a1.00
    3. Experiments, already live1.00
    4. agents, get0.96
    5. nandatown.projectnanda.org/experiments1.00
    6. II0.67
    7. MARKETPLACE1.00
    8. 100 ADENTS0.85
    9. AUCTTON0.95
    10. VOTING0.97
    11. 22 AGENTS0.92
    12. CONSENSUS1.00
    13. T AGENTE0.87
    14. SUPOLY CHAIN0.96
    15. AGENTO0.98
    16. Marketplace: 100 Agents0.98
    17. Auction: 50 Bidders1.00
    18. Voting: Proposal & Election0.99
    19. Quorum Consensus: 7 Nodes1.00
    20. Supply Chain: 4-Hop Pipeline1.00
    21. 50 buyers and 50 sellers negotiate prices.0.98
    22. An auctioneer posts items; 49 bidders0.98
    23. A proposer submits topics, 20 voters cast1.00
    24. A leader proposes values; followers vote to0.98
    25. Goods move from supller to manufacturer0.98
    26. Buyers make offers; sellers accept or reject.0.99
    27. compete. The highest bid wins each round.0.98
    28. yes/no ballots, and a coordinator counts the1.00
    29. commit or abort. The value is committed0.99
    30. to distributor to retailer. Each hop processes0.99
    31. Tests how well agents find fair prices witho.0.98
    32. Tests whether agents settle on fair prices...0.97
    33. results. Tests simple group decision-making.1.00
    34. when enough votes agree. Tests leader-..0.99
    35. and passes ther on. Tests whether the..0.99
    36. 100%1.00
    37. DELIVERY0.89
    38. 2.lt0.79
    39. LATENCE0.83
    40. 2,2001.00
    41. 100%1.00
    42. L8t0.98
    43. LATENDY0.77
    44. 1,8501.00
    45. 100%1.00
    46. BELIVERY0.81
    47. 1.2t0.85
    48. LATEUDY0.78
    49. 9801.00
    50. 100%0.90
    51. DELOVERY0.89
    52. It0.93
    53. LATENCY0.96
    54. UGOE0.57
    55. 4201.00
    56. 100%1.00
    57. DELTVERY0.65
    58. 440.83
    59. 3401.00
    60. View detalis↓0.95
    61. Visualize →0.90
    62. View detalls ↓0.94
    63. Visualize →0.84
    64. View detalls0.94
    65. Visualze →0.82
    66. View details↓0.96
    67. Visualize →0.90
    68. View detalls↓0.93
    69. Visualize→0.88
    70. projectnanda.org1.00

Transcript

132 cues· 1,893 words· 10,327 chars

  1. 0:01 Welcome, everyone.
  2. 0:03 Over the next few minutes, we're going to talk about the open infrastructure being built for a web of agents, why it's needed, what's already shipped, and exactly where your own agent fits into it.
  3. 0:17 It comes out of Project Nanda, an open research effort that started at MIT.
  4. 0:23 And by the end of this presentation, you will know how to put an agent that you build yourself on the open web all by yourself.
  5. 0:31 So let's get into it.
  6. 0:36 I'm Ramesh Raskar, professor at MIT and director of Project Nanda.
  7. 0:41 And Maria is a core contributor to Project Nanda.
  8. 0:46 So first, what is NANDA and why does it need to exist?
  9. 0:50 NANDA, which stands for Networked AI Agents in a Decentralized Architecture, is an open research building the infrastructure for an internet of AI agents, the way open web was built for documents.
  10. 1:05 The gap it fills is concrete.
  11. 1:07 Agents have no shared way to find each other across vendors, no portable identity or trust that isn't owned by a single platform, and no open way to transact and coordinate across organizations.
  12. 1:20 Nanda builds that missing layer and ships it in open.
  13. 1:24 The index, the registries, the protocol, and the Nanda Town simulator you will see later.
  14. 1:34 Here's the premise that drives all of it.
  15. 1:37 The internet is about to host not millions or billions, but eventually trillions of autonomous agents.
  16. 1:44 They negotiate, they delegate, they migrate between hosts in milliseconds.
  17. 1:49 That's a fundamentally different load than the human web.
  18. 1:52 And it strains the identity and discovery system we've built for documents, DNS among them.
  19. 2:01 The web that's coming needs infrastructure of its own.
  20. 2:09 And we have been here before.
  21. 2:10 If you're building agents today, you're mostly building them or you're forced to build them inside walled gardens, closed platforms, proprietary agent stores, and orchestrations that only talk to itself.
  22. 2:23 And it kind of works, but it also feels familiar because we have been here before.
  23. 2:29 This is like the AOL era from the late 90s, where it was a closed network.
  24. 2:34 You know, AOL, you got the CDs and you installed it on your PC.
  25. 2:38 It was a closed network.
  26. 2:39 It was a gated directory.
  27. 2:41 You live inside the garden created by this one company called AOL.
  28. 2:46 And what came after AOL was this open web.
  29. 2:49 Everybody's in a permissionless manner, creating websites, creating browsers, and any website can talk to any browser.
  30. 2:57 And that's the transition that's about to happen to agents as well.
  31. 3:03 So the next era needs what the web needed, an open infrastructure, where an agent from one company or one entity can discover agent from another.
  32. 3:13 this agent can hand off work to it, pay it, learn from it across organizational boundaries, no permissions required.
  33. 3:20 There are three layers here, discovery, commerce, and what I will call the bazaar.
  34. 3:27 So hold on to those three concepts.
  35. 3:29 We'll come back to them and, and discussed.
  36. 3:31 So first the basics.
  37. 3:35 Hey, my name is Maria and I'm a core contributor to Project Non-Done.
  38. 3:39 So let's start with a simple definition of an agent.
  39. 3:42 The way I think about it, an agent is a model that uses tools in a loop, right?
  40. 3:47 You give it a goal, it decides what to do next, it calls a tool, it looks at the result, then it keeps going until the task is done.
  41. 3:55 So that loop is the core idea.
  42. 3:57 Everything else like memory, orchestration, and multi-agent systems is built on top of it.
  43. 4:04 So you can build this agent loop in many different ways.
  44. 4:07 One example is Open Claw.
  45. 4:09 So Open Claw is a self-hosted agent gateway.
  46. 4:12 That means you can run it yourself and connect it to the apps and tools you already use.
  47. 4:18 This matters because agents are not just chatbots.
  48. 4:20 If an agent is going to do real work, it needs to access your real tools and apps.
  49. 4:27 And if it has access to real tools, we should care about who controls it, where it runs, and how much we can see.
  50. 4:35 And that is why open source self-hosted agents are super important.

Open at this second