vidtheque.

read-only demo

Videos Q9ycQHbDdJs

Agents Need Receipts, Not More Tool Calls - Armanas Povilionis, Alithea Bio

index_state ready data_status ok

AI Engineer· published 2026-07-20· 0:10:22· en-US· indexed 2026-08-10 19:47

Open on YouTube

Scene timeline

  1. Shot 0, 0:00 to 0:28, 1 of 1 keyframes kept
  2. Shot 1, 0:28 to 0:57, 0 of 1 keyframes kept
  3. Shot 2, 0:57 to 1:36, 1 of 1 keyframes kept
  4. Shot 3, 1:36 to 1:39, 1 of 1 keyframes kept
  5. Shot 4, 1:39 to 1:42, 1 of 1 keyframes kept
  6. Shot 5, 1:42 to 2:06, 1 of 1 keyframes kept
  7. Shot 6, 2:06 to 2:34, 1 of 1 keyframes kept
  8. Shot 7, 2:34 to 3:02, 0 of 1 keyframes kept
  9. Shot 8, 3:02 to 3:34, 1 of 1 keyframes kept
  10. Shot 9, 3:34 to 4:06, 0 of 1 keyframes kept
  11. Shot 10, 4:06 to 4:48, 1 of 1 keyframes kept
  12. Shot 11, 4:48 to 4:55, 1 of 1 keyframes kept
  13. Shot 12, 4:55 to 5:03, 1 of 1 keyframes kept
  14. Shot 13, 5:03 to 5:34, 1 of 1 keyframes kept
  15. Shot 14, 5:34 to 6:04, 1 of 1 keyframes kept
  16. Shot 15, 6:04 to 6:18, 1 of 1 keyframes kept
  17. Shot 16, 6:18 to 6:54, 1 of 1 keyframes kept
  18. Shot 17, 6:54 to 7:13, 0 of 1 keyframes kept
  19. Shot 18, 7:13 to 7:54, 0 of 1 keyframes kept
  20. Shot 19, 7:54 to 8:26, 1 of 1 keyframes kept
  21. Shot 20, 8:26 to 8:47, 0 of 1 keyframes kept
  22. Shot 21, 8:47 to 9:19, 0 of 1 keyframes kept
  23. Shot 22, 9:19 to 9:46, 0 of 1 keyframes kept
  24. Shot 23, 9:46 to 9:49, 0 of 1 keyframes kept
  25. Shot 24, 9:49 to 10:22, 1 of 1 keyframes kept

25 shot(s).

keyframes kept every frame deduplicated

What was stored

cues
73
whisperx 73
chunks
18
from 73 cues
keyframes
16
kept of 25 captured
frames with text
16
280 lines read
chapters
0
from the source metadata
keyframe bytes
2.6 MB
word timings on 73 cues

Provenance

Each pipeline stage, its state and the model that produced it
stage state model started took
fetch done — 2026-08-10 04:00 1m 55s
stt done — 2026-08-10 04:02 9s
chunk done — 2026-08-10 04:02 0s
text_embed done — 2026-08-10 19:47 1s
keyframe done — 2026-08-10 04:02 31s
ocr done — 2026-08-10 04:03 6s
frame_embed done — 2026-08-10 19:47 2s

Frames, and what the machine read

  • 0:11 #0 done4 line(s)

    shot 0·sharpness 957.5

    1. _o..o_froglet project of0.98
    2. Alithea bio0.96
    3. Agents need receipts0.99
    4. not more tool calls.0.97

  • 0:31 #1 skipped

    shot 1·duplicate of #0

  • 1:16 #2 done2 line(s)

    shot 2·sharpness 1086.9

    1. _0..0_ froglet0.93
    2. K Alithea bio0.94

  • 1:38 #3 done2 line(s)

    shot 3·sharpness 1526.3

    1. _0..0_ froglet0.95
    2. Alithea bio0.98

  • 1:42 #4 done2 line(s)

    shot 4·sharpness 1666.0

    1. _0..0_froglet0.96
    2. K Alithea bio0.94

  • 2:03 #5 done2 line(s)

    shot 5·sharpness 1940.8

    1. _0..0_ froglet0.94
    2. Alithea bio0.99

  • 2:31 #6 done10 line(s)

    shot 6·sharpness 1811.3

    1. _o..0_ froglet0.91
    2. Alithea bio0.99
    3. AGENT SURFACES0.99
    4. PROVIDER ASSETS1.00
    5. MCP / A2A-style calls0.99
    6. data1.00
    7. OpenClaw / NemoClaw1.00
    8. compute1.00
    9. custom runtimes1.00
    10. analysis services1.00

  • 2:53 #7 skipped

    shot 7·duplicate of #6

  • 3:09 #8 done14 line(s)

    shot 8·sharpness 2263.3

    1. _o..o_ froglet0.90
    2. A Alithea bio0.92
    3. AGENT INTERFACES0.98
    4. FROGLET1.00
    5. PROVIDER ASSETS1.00
    6. MCP / A2A-style calls0.99
    7. discover1.00
    8. data1.00
    9. OpenClaw / NemoClaw1.00
    10. execute1.00
    11. compute1.00
    12. custom runtimes1.00
    13. get receipt1.00
    14. analysis services0.99

  • 4:02 #9 skipped

    shot 9·duplicate of #8

  • 4:35 #10 done17 line(s)

    shot 10·sharpness 3042.7

    1. _0..o_ froglet0.90
    2. K Alithea bio0.95
    3. https://froglet.dev/demo/1.00
    4. 00.88
    5. 00.87
    6. Lightning1.00
    7. Al Claude Code0.94
    8. Stripe1.00
    9. Codex1.00
    10. x+ x4020.95
    11. MCP1.00
    12. OpenClaw1.00
    13. WA1.00
    14. WASM1.00
    15. 口0.52
    16. Python1.00
    17. Docker1.00

  • 4:49 #11 done32 line(s)

    shot 11·sharpness 4249.9

    1. _o..o_froglet0.96
    2. A Alithea bio0.91
    3. armanas—armanas@Armanass-MacBook-Air—-zsh—80×270.98
    4. FROGLET_AGENT_TARGET=manual0.98
    5. curl -fsSL https://froglet.dev/agent1.00
    6. bash1.00
    7. Installing froglet-node from0.99
    8. Installed froglet-node to /Users/armanas/.local/bin/froglet-node1.00
    9. Starting Froglet provider/runtime from published images...0.99
    10. [+] up 2/20.99
    11. Container froglet_agent-provider-1 Healthy1.00
    12. 0.5s1.00
    13. Container froglet_agent-runtime-1 Running1.00
    14. 0.0s1.00
    15. {0.99
    16. "status": "ok",0.98
    17. "bootstrap_dir": "/Users/armanas/.froglet/agent",0.99
    18. "data_dir": "/Users/armanas/.froglet/data",1.00
    19. "froglet_node": "/Users/armanas/.local/bin/froglet-node",1.00
    20. "provider_url": "http://127.0.0.1:8080",0.98
    21. "runtime_url": "http://127.0.0.1:8081",0.99
    22. "network_mode": "clearnet",1.00
    23. "compose_file": "/Users/armanas/.froglet/agent/compose.yaml",1.00
    24. "compose_project_name": "froglet_agent",0.99
    25. "compose_started": true,0.98
    26. "agent_target": "claude-code",1.00
    27. "mcp_docker_network": "froglet_agent_default",1.00
    28. "mcp_config_path": "/Users/armanas/.froglet/agent/mcp/.mcp.json",1.00
    29. "next_mcp_actions": ["status", "publish_artifact"],0.99
    30. "next_instruction": "Restart or point your agent at the MCP config path, call0.99
    31. froglet status, then publish a demo service with template demo.add."1.00
    32. < took 3s < at 18:40:550.94

  • 5:02 #12 done18 line(s)

    shot 12·sharpness 3212.3

    1. _o..o_froglet0.97
    2. Alithea bio0.97
    3. armanas- Vibe-vibe-80×240.94
    4. Vibe-vibe0.97
    5. Read https://try.froglet.dev/1lms.txt,1.00
    6. follow the hosted demo flow exactly if1.00
    7. you can access it, otnerwise say only that you could not access it, then give0.98
    8. me an honest, evidence-backed assessment that reports the observed HTTP0.99
    9. statuses, observed service IDs, observed deal status, observed result,0.99
    10. whether a receipt was present, and any mismatch between these docs and live0.99
    11. behavior before explaining what Froglet just proved, what it did not prove,1.00
    12. and the single most relevant next experiment for my files, tools, data,1.00
    13. configuration, workflows, constraints, and goals.1.00
    14. Thought1.00
    15. Fetched https://try.froglet.dev/1lms.txt (11,205 chars, text/plain)0.99
    16. 114 lines0.95
    17. auto approve-0.98
    18. 17% of 200k tokens0.98

  • 5:10 #13 done5 line(s)

    shot 13·sharpness 711.8

    1. _o..o_froglet0.95
    2. Alithea bio0.98
    3. Scientific collaboration1.00
    4. setup takes years and0.99
    5. millions1.00

  • 5:58 #14 done8 line(s)

    shot 14·sharpness 1300.4

    1. _o..o_froglet0.98
    2. Alithea bio0.98
    3. Scientific collaboration1.00
    4. froglet setup costs few1.00
    5. setup takes years and1.00
    6. thousand tokens and takes0.97
    7. costs millions1.00
    8. minutes.1.00

  • 6:11 #15 done67 line(s)

    shot 15·sharpness 1417.8

    1. Froglet — identity. execution.0.97
    2. ←0.99
    3. →0.98
    4. froglet.dev1.00
    5. ☆1.00
    6. Work1.00
    7. _0..0_ froglet0.91
    8. Home0.97
    9. Docs1.00
    10. Marketplace1.00
    11. Managed1.00
    12. Open source1.00
    13. GitHub1.00
    14. FROGLET - IDENTITY. EXECUTION. SETTLEMENT.0.99
    15. LET THE AGENT PROVE IT1.00
    16. Froglet is not for1.00
    17. PROMPT FOR YOUR AI AGENT1.00
    18. Open https://try.froglet.dev/llms.txt. Run0.99
    19. COPY1.00
    20. You.1.00
    21. demo.add and one witness/hash follow-up.1.00
    22. Report statuses, result, receipt/feed proof,1.00
    23. mismatches, and what this proves or does not0.99
    24. It's for your AI1.00
    25. prove for my projects/data. If unreachable,1.00
    26. say so.0.93
    27. Agent.1.00
    28. Paste into Claude Code, Codex, OpenClaw, or any agentic client. The1.00
    29. agent should come back with live evidence, not a product summary.1.00
    30. An open protocol for Al agents to sell and buy services—0.99
    31. FAST PROOF1.00
    32. STRICT AUDIT1.00
    33. cryptographic identity, signed deals, executable work, and verifiable1.00
    34. settlement. Trust through math, not middlemen.0.99
    35. Hosted proof1.00
    36. COPY1.00
    37. Evidence audit0.98
    38. COPY0.99
    39. Try in cloud0.98
    40. Run locally1.00
    41. Watch walkthrough →0.98
    42. FOLLOW-UP0.98
    43. Proof + witness0.99
    44. LOCAL PATH0.97
    45. Install proposal1.00
    46. COPY0.99
    47. COPY1.00
    48. FEED0.99
    49. Receipt check1.00
    50. COPY1.00
    51. AGENT REPORT0.98
    52. CATALOG0.99
    53. RESULT1.00
    54. EVIDENCE1.00
    55. No trust tour.1.00
    56. 5 free demos1.00
    57. sum = 120.97
    58. receipt + feed0.97
    59. Make the model bring back evidence from1.00
    60. demo.add + witness/hash0.99
    61. {a:7,b:5}0.99
    62. descriptor · offer0.94
    63. the live service.1.00
    64. receipt1.00
    65. LAUNCH1.00
    66. https://froglet.dev/demo/1.00
    67. Exaetly twe public0.82

  • 6:39 #16 done46 line(s)

    shot 16·sharpness 1073.1

    1. Demo - Froglet0.98
    2. ←1.00
    3. froglet.dev/demo/1.00
    4. ☆1.00
    5. Work1.00
    6. _0..0_ froglet0.89
    7. Home0.95
    8. Docs1.00
    9. Marketplace1.00
    10. Managed1.00
    11. Open source1.00
    12. GitHub1.00
    13. STEP 1 / 111.00
    14. The network1.00
    15. Three roles, one protocol. Providers sell1.00
    16. resources. Requesters buy them.0.98
    17. Marketplaces help them find each other.1.00
    18. Learn more →0.98
    19. Every node can play multiple roles. A1.00
    20. marketplace is itself a provider that sells0.99
    21. search. There are no special node classes.0.99
    22. provider1.00
    23. F^P0.88
    24. deal1.00
    25. requester1.00
    26. F^R1.00
    27. froglet - zsh0.96
    28. ) FROGLET_PAYMENT_BACKEND=none0.99
    29. FROGLET_PUBLISH_DEMO_SERVICES=1 cargo run0.98
    30. local assumption: 127.0.0.1:8080 and :80810.98
    31. bin froglet-node0.99
    32. register1.00
    33. discover1.00
    34. are free1.00
    35. STARTUP OUTPUT (ABRIDGED)1.00
    36. Local Runtime API: http://127.0.0.1:80811.00
    37. Local API Gateway: http://127.0.0.1:80800.99
    38. Node is now online and accepting traffic.1.00
    39. ~70.54
    40. M1.00
    41. marketplace1.00
    42. Provider Froglet → F^P0.98
    43. 40.95
    44. Requester → F^R, Marketplace → M0.98
    45. Back1.00
    46. Continue1.00

  • 7:04 #17 skipped

    shot 17·duplicate of #16

  • 7:30 #18 skipped

    shot 18·duplicate of #16

  • 8:01 #19 done44 line(s)

    shot 19·sharpness 980.0

    1. Demo - Froglet0.97
    2. ←1.00
    3. froglet.dev/demo/1.00
    4. ☆1.00
    5. Work1.00
    6. _0..0_ froglet0.96
    7. Home0.91
    8. Docs1.00
    9. Marketplace1.00
    10. Managed1.00
    11. Open source1.00
    12. GitHub0.99
    13. STEP 5 / 110.96
    14. Discovery1.00
    15. A marketplace is optional. It indexes1.00
    16. signed provider feeds so requesters can0.99
    17. search. Learn more →0.99
    18. The marketplace is not a protocol root of truth.1.00
    19. F^P0.99
    20. F^R1.00
    21. It consumes signed descriptors and offers;1.00
    22. provider1.00
    23. requester1.00
    24. kernel quote, deal, receipt, and settlement1.00
    25. semantics stay the same with or without it.0.99
    26. …0.54
    27. froglet - zsh0.96
    28. publish feed0.98
    29. discover1.00
    30. ~ ) curl0.89
    31. http://127.0.0.1:8080/v1/provider/services1.00
    32. SERVICES OUTPUT (ABRIDGED)1.00
    33. {"services":1.00
    34. [{"service_id":"demo.add","price_sats":0},1.00
    35. ...]}0.68
    36. M0.99
    37. ->0.67
    38. marketplace1.00
    39. M does:1.00
    40. 1. verify signed descriptor1.00
    41. 2. index bound offers1.00
    42. 3. return provider URLs1.00
    43. Back1.00
    44. Continue1.00

  • 8:38 #20 skipped

    shot 20·duplicate of #16

  • 9:03 #21 skipped

    shot 21·duplicate of #16

  • 9:35 #22 skipped

    shot 22·duplicate of #16

  • 9:47 #23 skipped

    shot 23·duplicate of #14

Transcript

73 cues· 951 words· 5,779 chars

  1. 0:01 What is the most valuable agentic automation work?
  2. 0:10 I think scientific research is quite high on that list.
  3. 0:17 After a decade working in life sciences collaboration projects, I think that more tools alone will not enable automation of scientific research.
  4. 0:30 because science work relies on collaboration.
  5. 0:36 For agents to collaborate autonomously, we need a chain of verifiable receipts.
  6. 0:45 We need a solution which can provide these receipts, proving every step, ensuring that every result can be trusted, and enabling collaboration at scale.
  7. 0:58 Let's step back.
  8. 1:02 Imagine agents as cooks in the kitchen.
  9. 1:07 Giving them more tools improves kitchen sufficiency.
  10. 1:12 Better knives, more pans, more ovens boost the speed and quality.
  11. 1:18 But it only enhances local work.
  12. 1:24 Scientific work is not cooking alone in your own kitchen.
  13. 1:30 It is closer to running a Michelin star restaurant.
  14. 1:34 The outcomes depend on suppliers and their produce, the level of service that you can provide, and an ability to consistently deliver the same quality dish again and again and again.
  15. 1:53 You cannot bring everything into one kitchen.
  16. 1:58 The challenge isn't local tools.
  17. 2:02 It is aligning the entire supply chain.
  18. 2:07 Today we already have plenty of tools for agents and agent automation.
  19. 2:14 On the other hand, we also have
  20. 2:17 data and specialized analytics algorithms, which are distributed across organizations and live in silos.
  21. 2:28 At Alifea, our vision for Froglet is very simple.
  22. 2:37 As agentic workflow automation matures, organizations will not just give agents more tools,
  23. 2:46 they will allocate them budgets.
  24. 2:50 We already kind of doing it in a primitive way, allocating them token budgets per task.
  25. 3:00 The next step is a bit broader.
  26. 3:03 It's allowing agents to manage their own budget.
  27. 3:07 for anything that they might need, discovering services, requesting data, negotiating execution, paying for work in cross-organizational boundary setting.
  28. 3:26 At that point, the agent is no longer just a cook with a better knife.
  29. 3:31 It starts acting like an executive chef, finding suppliers,
  30. 3:37 ordering ingredients, coordinating the kitchen work, and keeping a record of everything that's happening.
  31. 3:49 That is why we're building Froglet, the protocol for agents to discover, transact with, and receive verifiable receipts for external data and service providers.
  32. 4:07 Froglet is designed to sit in between of many moving parts.
  33. 4:13 And that's why we are not replacing existing tools and protocols.
  34. 4:18 We integrate with different payment rails, with different agenting harnesses, execution environments, and even network transport protocols.
  35. 4:32 It does not require that everyone has the same software stack.
  36. 4:37 It just requires that everyone has the same interface.
  37. 4:43 For more details, please visit froglet.dev, where you will be able to see how to run Froglet locally with just one command.
  38. 4:57 Or you can even try Froglet remotely with just one prompt.
  39. 5:05 So in essence, closed scientific collaboration often turns into a bespoke enterprise project.
  40. 5:15 That can take years and cost millions before even the first reusable workflow exists.
  41. 5:24 On another hand, the Froglet's mission is to simplify that much more.
  42. 5:32 Once the organization has deemed that the resource is shareable, a provider should be able to expose it through Froglet.
  43. 5:45 An agent can discover it, understand its terms, request the work, and receive a verifiable receipt.
  44. 5:55 That costs a few thousand tokens and takes minutes.
  45. 6:04 Let's deep dive a bit deeper.
  46. 6:07 So here is our froglet.dev website, and here you can see a walkthrough button.
  47. 6:18 If you click on it, you will have a much more detailed review of what's happening, and you can read documentation in even more detail.
  48. 6:26 So first of all, the Froglet network consists of homogeneous nodes.
  49. 6:35 Every single actor in the environment runs the same core node.
  50. 6:42 It just plays a different role.

Open at this second