Videos Fu45geO3zX8
Agents Need Receipts, Not More Tool Calls - Armanas Povilionis, Alithea Bio
Scene timeline
72 shot(s).
keyframes kept every frame deduplicated
What was stored
- cues
- 152
- whisperx 152
- chunks
- 34
- from 152 cues
- keyframes
- 54
- kept of 72 captured
- frames with text
- 54
- 1,691 lines read
- chapters
- 0
- from the source metadata
- keyframe bytes
- 8.7 MB
- word timings on 152 cues
Provenance
| stage | state | model | started | took |
|---|---|---|---|---|
fetch |
done | — | 2026-08-11 00:48 | 1m 19s |
stt |
done | — | 2026-08-11 00:50 | 17s |
chunk |
done | — | 2026-08-11 00:50 | 0s |
text_embed |
done | — | 2026-08-11 00:50 | 1s |
keyframe |
done | — | 2026-08-11 00:50 | 1m 04s |
ocr |
done | — | 2026-08-11 00:51 | 55s |
frame_embed |
done | — | 2026-08-11 00:52 | 9s |
Frames, and what the machine read
-
- _o..o_ froglet project of Alithea bio0.93
- Agents need receipts,0.99
- not more tool calls.1.00
-
- _0..0_ froglet0.96
- Alithea bio0.96
-
- _0..0_ froglet0.96
- Alithea bio0.99
-
- _0..0_froglet0.98
- Alithea bio0.99
-
- _0..0_ froglet0.94
- Alithea bio0.99
- AGENT SURFACES0.99
- PROVIDER ASSETS1.00
- MCP / A2A-style calls0.98
- data1.00
- OpenClaw / NemoClaw1.00
- compute1.00
- custom runtimes1.00
- analysis services1.00
-
- _0..0_ froglet0.96
- Alithea bio0.98
- AGENT INTERFACES1.00
- FROGLET1.00
- PROVIDER ASSETS1.00
- MCP / A2A-style calls0.98
- discover1.00
- data0.99
- OpenClaw / NemoClaw0.98
- execute1.00
- compute1.00
- custom runtimes1.00
- get receipt1.00
- analysis services1.00
-
- _0..0_ froglet0.94
- Alithea bio0.99
- https://froglet.dev/demo/1.00
- 00.97
- 00.96
- Lightning1.00
- Al Claude Code0.99
- Stripe1.00
- Codex1.00
- x+ x4020.96
- MCP1.00
- OpenClaw1.00
- WASM1.00
- 00.66
- Python1.00
- Docker1.00
-
- _o..o_ froglet0.90
- Alithea bio0.99
- armanas - armanas@Armanass-MacBook-Air--zsh-80×270.96
- FROGLET_AGENT_TARGET=manual0.99
- curl -fsSL https://froglet.dev/agent1.00
- bash1.00
- Installing froglet-node from1.00
- Installed froglet-node to /Users/armanas/.local/bin/froglet-node0.99
- Starting Froglet provider/runtime from published images...1.00
- [+] up 2/20.96
- Container froglet_agent-provider-1 Healthy0.99
- 0.5s1.00
- Container froglet_agent-runtime-1 Running1.00
- 0.0s1.00
- "status": "ok",0.98
- "bootstrap_dir": "/Users/armanas/.froglet/agent",0.99
- "data_dir": "/Users/armanas/.froglet/data"1.00
- "froglet_node": "/Users/armanas/.local/bin/froglet-node",1.00
- "provider_url": "http://127.0.0.1:8080",0.99
- "runtime_url": "http://127.0.0.1:8081",1.00
- "network_mode": "clearnet",1.00
- "compose_file": "/Users/armanas/.froglet/agent/compose.yaml",1.00
- "compose_project_name": "froglet_agent",1.00
- "compose_started": true,1.00
- "agent_target": "claude-code",1.00
- "mcp_docker_network": "froglet_agent_default",0.99
- "mcp_config_path": "/Users/armanas/.froglet/agent/mcp/.mcp.json",1.00
- "next_mcp_actions": ["status", "publish_artifact"],0.99
- "next_instruction": "Restart or point your agent at the MCP config path, call1.00
- froglet status, then publish a demo service with template demo.add."1.00
- < took 3s < at 18:40:550.96
-
- _o..o_froglet0.93
- Alithea bio0.97
- armanas- Vibe- vibe -80×240.93
- Vibe-vibe0.95
- Read https://try.froglet.dev/1lms.txt,0.99
- follow the hosted demo flow exactly if0.99
- you can access it, otnerwise say only that you could not access it, then give0.99
- me an honest, evidence-backed assessment that reports the observed HTTP0.99
- statuses, observed service IDs, observed deal status, observed result,1.00
- whether a receipt was present, and any mismatch between these docs and live0.99
- behavior before explaining what Froglet just proved, what it did not prove,1.00
- and the single most relevant next experiment for my files, tools, data,1.00
- configuration, workflows, constraints, and goals.0.99
- Thought0.99
- Fetched https://try.froglet.dev/llms.txt (11,205 chars, text/plain)1.00
- 114 lines0.97
- auto approve0.98
- >0.67
- 17% of 200k tokens0.98
-
- _0..0_ froglet0.91
- Alithea bio0.96
- Scientific collaboration1.00
- setup takes years and0.99
- millions1.00
-
- Chrome1.00
- File1.00
- Edit1.00
- View1.00
- History1.00
- Bookmarks1.00
- Profiles0.96
- Tab1.00
- Window1.00
- Help1.00
- 90.78
- ①0.73
- 国0.65
- 96%0.81
- G0.65
- Fri Jun 26 18:520.99
- Froglet-Al-Engineer-Researc1.00
- x0.60
- Froglet- identity. execution0.96
- x0.79
- ←1.00
- →0.94
- docs.google.com/presentation/d/1stTpyl3X4diA3yzawusrYeZLhXrOV73j/edit?slide=id.g3ef25a767c8_0_465#slide=id.g3ef25a767c81.00
- Ask Google1.00
- ☆1.00
- Work1.00
- Froglet-Al-Engineer-Research-Collaboration-v31.00
- .PPTX0.99
- ☆0.99
- 目0.87
- Slideshow1.00
- Share0.95
- File Edit View Insert Format1.00
- Slide1.00
- Arrange Tools0.98
- Help1.00
- a + -0.60
- @ Fit0.95
- A0.56
- T0.71
- 田0.94
- Background1.00
- Layout1.00
- Theme1.00
- Transition1.00
- ^0.69
- 11.00
- 21.00
- 31.00
- 51.00
- 61.00
- 81.00
- 91.00
- 101.00
- 111.00
- 121.00
- 131.00
- _o..0_froglet0.92
- AAlithea bio0.91
- 81.00
- Slide1.00
- Image1.00
- 91.00
- Scientific collaboration1.00
- setup takes years and0.99
- thousand tokens and takes1.00
- froglet setup costs few1.00
- Media1.00
- costsmillions1.00
- minutes.1.00
- 口0.92
- 200m0.71
- Uploads1.00
- 101.00
- Record1.00
- 111.00
- Speaker1.00
- 121.00
- Enhance this slide ×0.97
- 品0.89
- <0.97
- That costs few thousand tokens and take minutes1.00
-
- Chrome1.00
- File1.00
- Edit1.00
- View1.00
- History1.00
- Bookmarks1.00
- Profiles0.94
- Tab1.00
- Window0.94
- Help1.00
- 90.82
- ①0.56
- 96%0.81
- G0.71
- Fri Jun 26 18:530.99
- Froglet-Al-Engineer-Researc1.00
- x0.67
- Froglet — identity. execution.0.98
- x0.70
- +0.99
- ←1.00
- C0.52
- froglet.dev1.00
- ☆0.99
- Work1.00
- _0..0_ froglet0.95
- Home1.00
- Docs1.00
- Marketplace1.00
- Managed1.00
- Open source1.00
- 口0.51
- GitHub0.99
- 40.65
- FROGLET - IDENTITY. EXECUTION. SETTLEMENT.0.99
- LET THE AGENT PROVE IT1.00
- Froglet is not for1.00
- PROMPT FOR YOUR AI AGENT0.98
- Open https://try.froglet.dev/llms.txt. Run0.99
- COPY1.00
- You.1.00
- demo.add and one witness/hash follow-up.1.00
- Report statuses, result, receipt/feed proof,1.00
- mismatches, and what this proves or does not1.00
- It's for your AI0.99
- prove for my projects/data. If unreachable,1.00
- say so.1.00
- Agent.1.00
- Paste into Claude Code, Codex, OpenClaw, or any agentic client. The1.00
- agent should come back with live evidence, not a product summary.1.00
- An open protocol for Al agents to sell and buy services—1.00
- FAST PROOF0.99
- STRICT AUDIT1.00
- zoom0.89
- cryptographic identity, signed deals, executable work, and verifiable1.00
- settlement. Trust through math, not middlemen.1.00
- Hosted proof0.96
- COPY1.00
- Evidence audit1.00
- COPY1.00
- FOLLOW-UP1.00
- LOCAL PATH0.98
- Try in cloud1.00
- Run locally0.97
- Watch walkthrough →0.99
- Proof + witness1.00
- Install proposal0.99
- COPY1.00
- COPW0.87
- FEED1.00
- Receipt check0.99
- COPY1.00
- AGENT REPORT1.00
- CATALOG1.00
- RESULT1.00
- EVIDENCE1.00
- No trust tour.1.00
- 5 free demos1.00
- sum = 120.99
- receipt + feed0.99
- Make the model bring back evidence from0.99
- demo.add + witness/hash1.00
- {a:7,b:5}1.00
- descriptor · offer0.94
- the live service.1.00
- receipt1.00
-
- Chrome1.00
- File1.00
- Edit1.00
- View1.00
- History1.00
- Bookmarks1.00
- Profiles1.00
- Tab1.00
- Window1.00
- Help1.00
- 90.88
- ①0.66
- 96%0.83
- G0.65
- Fri Jun 26 18:530.98
- Froglet-Al-Engineer-Researc1.00
- x0.67
- Froglet — identity. execution.0.98
- x0.73
- +0.98
- ←1.00
- →0.70
- froglet.dev1.00
- ☆0.99
- Work1.00
- _0..0_ froglet0.92
- Home1.00
- Docs1.00
- Marketplace1.00
- Managed1.00
- Open source1.00
- C0.80
- GitHub0.98
- You.1.00
- demo.add and one witness/hash follow-up.1.00
- Report statuses, result, receipt/feed proof,1.00
- mismatches, and what this proves or does not1.00
- It's for your AI0.99
- prove for my projects/data. If unreachable,1.00
- say so.1.00
- Agent.1.00
- Paste into Claude Code, Codex, OpenClaw, or any agentic client. The1.00
- agent should come back with live evidence, not a product summary.1.00
- An open protocol for Al agents to sell and buy services—0.99
- FAST PROOF1.00
- STRICT AUDIT1.00
- cryptographic identity, signed deals, executable work, and verifiable1.00
- Hosted proof1.00
- 40.94
- Evidence audit1.00
- COPY0.99
- COPY1.00
- settlement. Trust through math, not middlemen.1.00
- FOLLOW-UP1.00
- LOCAL PATH1.00
- Try in cloud1.00
- Run locally0.98
- Watch walkthrough →0.99
- Proof + witness0.99
- Install proposal0.98
- COPY0.99
- COPY1.00
- FEED0.88
- zoom0.84
- Receipt check0.99
- COPY0.99
- AGENT REPORT1.00
- CATALOG1.00
- RESULT1.00
- EVIDENCE1.00
- No trust tour.1.00
- 5 free demos0.99
- sum = 120.98
- receipt + feed1.00
- Make the model bring back evidence from0.98
- demo.add + witness/hash1.00
- {a:7,b:5}1.00
- descriptor · offe:0.92
- the live service.0.99
- receipt1.00
- LAUNCH1.00
- Exactly two public0.99
-
- Chrome1.00
- File1.00
- Edit1.00
- View1.00
- History1.00
- Bookmarks1.00
- Profiles1.00
- Tab1.00
- Window1.00
- Help1.00
- 90.88
- ①0.89
- 96%0.83
- G0.65
- Fri Jun 26 18:530.96
- Froglet-Al-Engineer-Researc1.00
- x0.67
- Froglet — identity. execution.0.96
- x0.73
- +0.98
- ←1.00
- →0.97
- froglet.dev1.00
- ☆1.00
- Work1.00
- _0..0_ froglet0.95
- Home1.00
- Docs1.00
- Marketplace1.00
- Managed1.00
- Open source1.00
- 口0.62
- C0.80
- GitHub0.98
- settlement. Trust through math, not middlemen.0.99
- FOLLOW-UP0.98
- LOCAL PATH0.97
- Try in cloud1.00
- Run locally1.00
- Watch walkthrough →0.98
- Proof + witness0.99
- Install proposal1.00
- COPY1.00
- COPY1.00
- FEED0.98
- Receipt check0.98
- COPY1.00
- 41.00
- AGENT REPORT1.00
- CATALOG1.00
- RESULT1.00
- EVIDENCE1.00
- No trust tour.1.00
- 5 free demos0.99
- sum = 120.98
- receipt + feed0.97
- Make the model bring back evidence from1.00
- demo.add + witness/hash1.00
- {a:7,b:5}1.00
- descriptor · offer0.96
- the live service.1.00
- receipt1.00
- zoom0.92
- LAUNCH1.00
- Exactly two public1.00
- entry points.1.00
- Froglet doesn't apologize for constraints, it names them. Pick the one that1.00
- matches your intent— hosted trial or local install. No third option.0.99
- 01· HOSTED0.96
- B2 - LOCAL0.88
- Try in cloud0.97
- Run locally0.99
-
- Chrome1.00
- File1.00
- Edit1.00
- View1.00
- History1.00
- Bookmarks1.00
- Profiles0.94
- Tab1.00
- Window1.00
- Help1.00
- 90.82
- ①0.63
- 96%0.82
- G0.68
- Fri Jun 26 18:530.98
- Froglet-Al-Engineer-Researc1.00
- x0.67
- Froglet — identity. execution.0.99
- x0.70
- +0.99
- ←1.00
- C0.52
- froglet.dev1.00
- ☆0.99
- Work1.00
- _0..0_0.93
- froglet1.00
- Home1.00
- Docs1.00
- Marketplace1.00
- Managed1.00
- Open source1.00
- GitHub0.99
- FROGLET - IDENTITY. EXECUTION. SETTLEMENT.0.99
- LET THE AGENT PROVE IT0.99
- Froglet is not for1.00
- PROMPT FOR YOUR AI AGENT0.99
- Open https://try.froglet.dev/llms.txt. Run0.98
- COPY1.00
- You.1.00
- demo.add and one witness/hash follow-up.1.00
- Report statuses, result, receipt/feed proof,1.00
- mismatches, and what this proves or does not1.00
- It's for your AI1.00
- prove for my projects/data. If unreachable,0.99
- say so.1.00
- Agent.1.00
- Paste into Claude Code, Codex, OpenClaw, or any agentic client. The0.99
- agent should come back with live evidence, not a product summary.1.00
- cryptographic identity, signed deals, executable work, and verifiable1.00
- An open protocol for Al agents to sell and buy services—0.99
- FAST PROOF0.99
- Hosted proof1.00
- STRICT AUDIT1.00
- Evidence audit1.00
- COPY1.00
- COPY1.00
- zoom0.88
- settlement. Trust through math, not middlemen.0.99
- 40.93
- FOLLOW-UP1.00
- LOCAL PATH0.99
- Try in cloud1.00
- Run locally1.00
- Watch walkthrough →0.98
- Proof + witness0.97
- Install proposal1.00
- COPY1.00
- COPY1.00
- FEED1.00
- Receipt check0.97
- COPY1.00
- AGENT REPORT1.00
- CATALOG1.00
- RESULT1.00
- EVIDENCE1.00
- No trust tour.1.00
- 5 free demos0.99
- sum = 120.98
- receipt + feed1.00
- Make the model bring back evidence from0.98
- demo.add + witness/hash0.98
- {a:7,b:5}0.99
- descriptor · offer0.95
- the live service.1.00
- receipt1.00
-
- Chrome1.00
- File1.00
- Edit1.00
- View1.00
- History1.00
- Bookmarks1.00
- Profiles1.00
- Tab1.00
- Window1.00
- Help1.00
- 90.82
- ①0.85
- 96%0.86
- G0.68
- Fri Jun 26 18:530.98
- Froglet-Al-Engineer-Researc1.00
- Demo - Froglet0.94
- x0.55
- +0.99
- ←1.00
- →0.94
- froglet.dev/demo/1.00
- ☆0.99
- 白0.58
- Work1.00
- _0..0_ froglet0.93
- Home1.00
- Docs1.00
- Marketplace1.00
- Managed1.00
- Open source1.00
- G0.65
- OGitHub0.93
- STEP 1 / 110.96
- The network1.00
- Three roles, one protocol. Providers sell1.00
- resources. Requesters buy them.1.00
- Marketplaces help them find each other.0.99
- Learn more →0.98
- Every node can play multiple roles. A1.00
- provider1.00
- F^P1.00
- deal0.89
- requester1.00
- F^R1.00
- marketplace is itself a provider that sells1.00
- search. There are no special node classes.0.99
- froglet - zsh0.98
- register1.00
- discover1.00
- zoom0.95
- ~ ) FROGLET_PAYMENT_BACKEND=none0.96
- FROGLET_PUBLISH_DEMO_SERVICES=1 cargo run --0.97
- bin froglet-node1.00
- local assumption: 127.0.0.1:8080 and :80810.99
- are free0.94
- STARTUP OUTPUT (ABRIDGED)0.99
- M1.00
- Local Runtime API: http://127.0.0.1:80810.97
- marketplace1.00
- Local API Gateway: http://127.0.0.1:80801.00
- Node is now online and accepting traffic.1.00
- ->0.53
- Provider Froglet → F^P0.98
- Requester → F^R,0.99
- Marketplace → M0.99
- 40.94
- Back1.00
- Continue1.00
Transcript
152 cues· 2,152 words· 11,811 chars
- 0:00 Hello, everybody.
- 0:02 I'm Armanos.
- 0:03 I work at Alifea Bio, where we focus on life sciences topics.
- 0:07 More specifically, we work on immuno-oncology and immunopathodontics.
- 0:13 While we work on science topics, we use ourselves and provide our clients agentic automation or agentic AI solutions.
- 0:23 However, we found a certain gap, which we would like to discuss today.
- 0:29 And we have a solution which we're working on, which we would like to share.
- 0:34 It's called Froglet.
- 0:37 But to start with, I would like to ask, what do you think is most valuable work that agentic automation can do?
- 0:49 My guess, scientific research automation would be quite high up that list.
- 0:56 However, from my experience, adding more tools to agents will not suffice.
- 1:03 because science and especially lives in especially in life sciences is inherently collaborative process and in order to enable agents to collaborate
- 1:18 We need a way to have a verifiable chain of receipts.
- 1:24 What do I mean?
- 1:25 We need a solution which can provide these receipts, proving every step, ensuring that every result can be trusted and enabling repeatability as well as collaboration at scale.
- 1:41 Having this thesis in mind, let's go to the next step.
- 1:45 So
- 1:47 Stepping back, we imagine agents as cooks in the kitchen.
- 1:54 Giving them more tools or better tools improves kitchen's efficiency.
- 2:03 Better knives, more pans, more ovens, definitely boosts the speed and quality, but it only enhances the local work.
- 2:14 On the other hand, scientific work is not cooking alone in your own kitchen.
- 2:21 It is closer to running a Michelin star restaurant.
- 2:26 The outcome depends on providers and quality of their produce.
- 2:36 high level services, as well as an ability to consistently deliver the same quality dish again and again and again.
- 2:45 You cannot bring everything into one kitchen.
- 2:50 The challenge isn't local.
- 2:51 It's not local tools.
- 2:53 The challenge is aligning the entire supply chain that it's repeatable and consistent.
- 3:02 Today, we have already agents with plenty of tools and increasing in volume, number, and quality.
- 3:12 And we also have on our hand data and specialized analytics algorithms and compute, which are distributed across different organizations and siloed.
- 3:26 So our vision is that whenever AI agent workflow automation matures, organizations will give to the agents not only tools, they will give them budgets.
- 3:43 We're already kind of doing this in a primitive form by allocating tokens for token budgets for a specific task.
- 3:54 But what I'm talking is a bit broader.
- 3:57 It's allowing agents to manage their own budgets to achieve their own goals, whether it's to discover services, request data, negotiate execution terms, or pay for work across organizational boundaries.
- 4:13 At that point, an agent is no longer just a cook with a better knife.
- 4:19 It starts to act like a chef.
- 4:25 like an executive chef, finding suppliers, ordering ingredients, coordinating work at the kitchen, and finally keeping the record of what has happened.
- 4:38 That is why we are building Froglet, an open-source protocol for agents to discover, transact with, and receive verifiable receipts from external data and service providers.
- 4:56 For more detailed information, obviously, go to our page, froglet.dev, where you will find general descriptions on how it works, what it integrates with.
- 5:09 Froglet is designed to be in between the moving parts.
- 5:14 It's not designed to replace.
- 5:15 It's designed to integrate.
- 5:17 It integrates with different payment rails.
- 5:19 It integrates with different agent harnesses, execution environments, and even transport protocols.
- 5:27 And it doesn't require that every node in the network uses the same software stack.
- 5:33 That's a beauty.
- 5:34 We are not forcing everyone to use the same, to work the same way.
- 5:38 We're just asking everybody to use the same interface.
- 5:44 In the webpage, you will find how to run Froglet with just, how to install it locally just with one command, or even how to run it remotely just with one prompt.
- 6:00 So in essence, what we're saying is closed source scientific collaboration often turns into a bespoke project, enterprise project, that can take years and cost millions before any reusable workflow exists.
loading