Week 38
Mon 15th Sep - Sun 21st Sep 2025
Monday 15th September 2025
WG Sync Call
Type of meeting: Monthly
Present: PeterE [facilitator], Alfred Itodele, CallyFromAuron [documenter], Ayomishuga, Jeffrey Ndaraka, Rems, CallyFromAuron, PeterE, Tevo, AshleyDawn, Sucre n Spice, guillermolucero, AndrewBen, kenichi, UKnowZork, Alfred Itodele
Purpose: Monthly call for all Workgroups to discuss Program-wide issues
Agenda Items:
Changes to consent process timeline - specifically, one round of consent instead of two - to conform with the features of the new Governance dashboard
Integration of historical data into the Governance dashboard
Verifying the validity of objections
Use of the 80-20 rule in the new single-round consent process
Definition of consent decisionmaking
Discussion:
Change to consent process timeline
The new dashboard's commenting feature encourages people to comment on budgets and reports, and allows WGs to amend them in response. So the consent process cannot start until this process is finished, to prevent changes after consent is given. To accommodate this, a new timeline with a single consent round was agreed: see "decisions" below.
We noted that this does not mean the second round is removed forever.
2)Integration of historical data into the Governance dashboard
It would be useful to add past objections and other historical data into the new dashboard, as recurring issues are revealing. Migrating all data would be a significant effort. Could start with the most recent (Q2/3) data Could allocate a budget from Ambassador reserves for the task. To be discussed further in a future meeting.
Verifying the validity of objections
The proposed timeline includes a focus group to check if objections are valid before they are discussed in the Governance WG meeting. Alfred raised concerns about a small, unelected group making decisions and suggested representation from all workgroups to ensure fairness. But it was pointed out to him that the validity check does not involve "making decisions", but simply scanning for any potential invalidities, and the assessment is ratified or not in the Governance meeting. We agreed to make the validity check an open session where anyone can participate, rather than a pre-selected focus group. It was clarified that the role of this session is simply to check if an objection is valid, not to comment on how the objection should be resolved.
80-20% rule
In past consent rounds, this rule has been applied in the 2nd round, so that a single objection cannot hold up the process indefinitely. As we no longer have a 2nd round, the rule might be inapproriate, as it would silence minority objections before they could be discussed. We agreed all objections will be discussed; and the 80/20 rule will only be applied if agreement cannot otherwise be reached.
Definition of consent decisionmaking
In response to a question from Ayomi, the consent process was explained.
Consenting to a budget does not mean that you think it is perfect; just that it is "safe enough to try".
Objections should not be the objewctor's own ideas for how things could be done better - an objection must identify actual, specific harm that would be caused if a proposal was carried out.
The process then focuses on resolving objections by asking objectors what would need to change in order to get their consent.
Decision Items:
The group will adopt a single consent round for the Q4 budget process The budget approval timeline was revised and finalized as follows:
Weds 17th Sept, 23:59 UTC: Deadline for WGs to submit reports and budgets.
Mon 22nd Sept: End of commenting period; budgets are locked; consent process opens.
Weds 24th Sept: Deadline for Core Contributors to consent/object/abstain.
Thurs 25th Sept before Governance meeting: objection validity check session,
Thurs 25th Sept: Discussion in Governance WG
Tues 30th Sept: Discussion in Governance WG
[rationale] To test the new dashboard's commenting functionality and see if it is able to speed up the consent process.
[effect] mayAffectOtherPeople
The meeting on Thurs 25th Sept to review objection validity will be an open session, and anyone who attends can participate.
It isn't a decision-making session - any "invalid" flags will be ratified in the Governance call.
[rationale] Inclusion and transparency.
[effect] mayAffectOtherPeople
The 80/20 rule will not be automatically applied. All valid objections will be discussed, even if raised by only one person; and the 80/20 rule will only be applied if agreement cannot otherwise be reached.
[rationale] To avoid silencing minority objections
[effect] mayAffectOtherPeople
Action Items:
[action] To ensure the dashboard backend allows for proposals to be locked on Sept 22nd, at which point the consent function becomes active. [assignee] guillermolucero [due] 19 September 2025 [status] todo
[action] Guillermo and Lean to investigate adding a feature to the dashboard to track the resolution of each objection [assignee] guillermolucero, leandro [due] 30 September 2025 [status] todo
Keywords/tags:
topics covered: Governance Dashboard, Q4 2025 budget timeline, migrating historical data, validity of objections, Consent decision-making process for Budget proposals, 80%/20% rule, Consent process
emotions: Collaborative, Constructive, some skepticism, pragmatic
Tuesday 16th September 2025
Ambassador Town Hall
Type of meeting: Weekly
Present: NA [facilitator], NA [documenter], Abdulbasit Adigun Abdulrahman, Alfred Itodele, Ayo, Colleen Pridemore, Effiom, Guillermo Lucero, Jeffrey Ndarake, Malik, Omolola Lawson, Peter Elfrink, Re ms, CallyFromAuron, Vasu Madaan
Purpose: Regular weekly get-together for the Ambassador Program. Discussion of community-wide issues; sharing of updates and info; and in the last meeting of each month, an update from each Workgroup on what they have been doing
Town Hall Number: 164
Working Docs:
Timestamped video:
The Ambassador Town Hall is a recurring event happening live on the Ambassador Zoom channel https://www.youtube.com/@SNET_Ambassador each Tuesday at 18:00 UTC.
00:00 Meeting Opening And Attendee Roll Call 08:52 Ambassador Program & Consent Process Update 14:09 R&d Guild And Bnb Hackathon Update 21:30 Deep Funding Deepfeed Introduction 39:49 Deepfeed Q&a And Discussion 51:25 Bgi 25 Planning Discussion 54:10 General Updates And Meeting Close
Town Hall Summary:
The agenda included a presentation by Abdulbasit on the new Deep Funding community website, emphasizing the importance of the Ambassador Program and encouraging participants to engage through the Discord server. Peter Elfrink provided updates on the consent process, noting a deadline change and the requirement for only one round of consent using a governance dashboard. He also highlighted the establishment of a temporary workgroup for Q4 fundraising efforts, addressing challenges in member selection due to high interest. Guillermo Lucero followed with an update on hackathon participation, stressing the need for better communication and collaboration to enhance agent development.
Abdulbasit introduced DeepFeed, a social platform designed for internal communication among AI innovators, aimed at improving collaboration within the decentralized AI community. He detailed the platform's features, including user login with existing credentials, posting updates, and managing visibility of contributions. The initiative seeks to involve Singularity Net Ambassadors as early adopters to gather feedback and ensure effective communication. The discussion also covered the ongoing development of features to amplify the decentralized AI ecosystem, with an emphasis on community involvement and marketing strategies to promote the platform.
The meeting included a discussion on recruiting Singularity Net ambassadors, with Abdulbasit proposing a target of 50, which Peter suggested might be ambitious given the current number of core contributors. The group agreed on using Discord for outreach, as it would be more effective than email. Abdulbasit confirmed that the platform would allow all users to post content related to AI, Web3, and blockchain, and he mentioned plans for a feedback mechanism to enhance user engagement. Peter concluded by proposing a weekly session for BGI 25 planning this Friday 13UTC and reminded workgroup leads to submit budget proposals and quarterly reports by the next day.
Keywords/tags:
topics covered: governance, budgets, tooling, Deep Funding
emotions: engaging
Wednesday 17th September 2025
Archives Workgroup
Type of meeting: Monthly
Present: Stephen [QADAO] [facilitator], CallyFromAuron [documenter], CallyFromAuron, Stephen [QADAO], PeterE, Alfred Itodele, AyomiShuga, UKnowZork, Rems
Purpose: Regular monthly meeting of the Archives WorkGroup in the SingularityNET Ambassador program
Meeting video: Link
Decision Items:
As token price remains comparatively low ($0.28), we still won't submit any pending Q1 tasks at the moment. However, we agreed that we should pay out at the start of October provided our WG reserves can cover it all (token price would need to be around $0.30).
[rationale] Even though paying out when token price is much lower (the Q1 budget was calculated at $0.55) will eat into the WG's reserves, nevertheless we agree it's not good practice to leave tasks unpaid for too long, regardless of token price
[effect] affectsOnlyThisWorkgroup
Monthly archives GitBook issues: summaries are still being submitted without any chasing, although as previouly noted, there are fewer meetings happening in general.
But we noted that the ongoing issues with summary quality seem to be worsening - particularly, vague and meaningless content (e.g. "We made a decision on XYZ" without saying what the decision was), and over-focus of the idea of a "meeting narrative" leading to undue emphasis on inconsequential details, and failure to record actual decisions.
We noted that this might need to be addressed eventually if the corpus is not to become irretrievably corrupt. However, we need to see the results of our knowledge graphs to fully assess the extent of the problem - can a graph derive meaningful and accurate insights despite inadequate summaries?
[rationale] This is largely caused by documenters relying uncritically on unmediated AI-generated content, which tends to be vague and unspecific, and not applying sufficient "human-in-the-loop" critical insight to correct and expand AI summaries. But as we do not have budget for quality control, such summaries are being added to the Archives as-is.
[opposing] Poor summary quality is being addressed, to some extent, outside of Archives WG (e.g. Vani checks and corrects summaries in WGs she is active in when they are presented for members' comments), but this is both time-consuming and limited in reach, and also unpaid. We noted that other measures might need to be implemented eventually (e.g training for documenters?)
[effect] mayAffectOtherPeople
Knowledge graph development: In August and so far in September, the focus (see https://github.com/SingularityNET-Archive/SingularityNET-Archive/issues/277 and https://github.com/SingularityNET-Archive/SingularityNET-Archive/issues/279 , and this repo https://github.com/SingularityNET-Archive/Graph-Python-scripts) has been on moving away from Neo4.j Aura towards a Python-based open source approach, using the open-source graph tool Gephi.
Stephen has then been refining the raw, undifferentiated graphs with graphical and statistical methods such as force-directed layouts to help discern patterns and disambiguate clusters.
[rationale] To align more fully with the Archives' open source ethos.
[effect] mayAffectOtherPeople
Stephen will be delivering a session on Thurs 25 Sept at AI Sandbox/Think Tank to share the knowledge graph work so far.
[rationale] Especially given that in Q4 we hope to engage the community with some rule-based auditing of graph outputs, we want to introduce people to the work so far.
[effect] mayAffectOtherPeople
We discussed a little what kinds of things the community might want to be able to ask a knowledge graph. We noted the following possibilities:
Who is working where, and whether this is reflected in recognition people get (Peter)
Can the Archives corpus tell us anything, by implication, about what goes on outside meetings? (e.g. action items assigning async work; linked documents that are outputs of async work) (Vani)
Influence, who has it, and how it fluctuates (we noted that early indications are that influence is not monolithic or fixed, that there are sevral "clusters" of influence, and that it is quite fluid and changes rapidly, perhaps reflecting the decentralised nature of the Program)
[effect] mayAffectOtherPeople
We discussed how our knowledge graph outputs could eventually be integrated into the ecosystem. Suggestions included
a Discord bot which could retrieve material to answer people's questions
integration with R&D's Governance Dashboard, particularly to ensure that its AI assistant has meaningful input on the Ambassador Program's specific context. (We noted also that generative models can be limited in their usefulness, particularly if all we do is throw a load of unstructured data at them. Because they can at least do more than they could a couple of years ago, people can be misled about their usefulness.)
some kind of collaboration with Atomspace https://github.com/opencog/atomspace
[effect] mayAffectOtherPeople
We noted that our knowledge graph work in Q4 is likely to address the different approaches needed for structured data (e.g some of meeting summary templates in use) and unstructured data (e.g. documents that are linked in meeting summaries; and the unstructured, free-text fields such as "Meeting Narrative" that are currently part of some summary templates.)
This will likely involve some semantic analysis (for example analysing particular parts of speech - what nounds are there, what verbs are there, in a piece of text?), and some sentiment analysis, in which we hope to move beyond simplistic "good/bad/neutral" approaches The fact that the archives corpus includes extensive "emotion tags" could be interesting here.
[effect] affectsOnlyThisWorkgroup
In a collaboration with Video WG, André and Malik have created a process to automate Town Hall suammries, using Read.ai data to create timestamps, and automating adding them to the summary tool. The meeting acknowledged how useful this is, and thanked Malik for the work, which will be compensated from Archives WG reserves with a smaller contribution from Video WG.
[rationale] This addresses the ongoing pain point of finding time to submit summaries of Ambassador Town Hall to the Archives.
[effect] mayAffectOtherPeople
Archives WG's Q3 quarterly report and Q4 budget have been submitted to the new Governance Dashboard here https://singularitynet-governance-dashboard.vercel.app/dashboard/proposals/cmfl42fhl0001l80a7auerrrl
The consent process ends on 30th Sept, so we will know then if it has been approved.
[effect] affectsOnlyThisWorkgroup
Action Items:
[action] Vani to publicise the information session at AI Sandbox/Think Tank next Thurs, 25th Sept [assignee] CallyFromAuron [due] 19 September 2025 [status] todo
Keywords/tags:
topics covered: AI ethics, token price, Neo4j, community engagement, q4 2025 budget, open source tooling, Knowledge Graph, Knowledge management across the singularityNET ecosystem, Gephi, Q3 2025 quarterly report, Open source, AI Sandbox/Think Tank, Video WG, human-in-the-loop, AI-generated summaries, meeting summary quality, Atomspace, R&D Guild, Governance Dashboard, what kinds of questions, structured vs unstructured data, Town Hall summaries, Automation, semantic analysis, sentiment analysis, Discord bot, WG reserves, force-directed graph layouts
emotions: interesting, short, wide-ranging, Discursive
Education Workgroup
Type of meeting: Biweekly
Present: Gorga Siagian [facilitator], Malik [documenter], osmium, Gorga Siagian, Kateri, UKnowZork, hogantuso, frosh, Malik
Purpose: To review CCP budget allocation, content progress, contingency planning, and governance webinar updates.
Working Docs:
Discussion Points:
Budget Allocation & Contingency The group revisited CCP funding allocations. It was clarified that allocations had already been made in Q2, with reserves covering extensions. While some members questioned whether new funds were required, the consensus was to reassign tasks to active contributors and avoid additional funding requests unless strictly necessary. Kary introduced a contingency line item in the budget (~$100) to account for operational costs and unforeseen delays. Members agreed the exact amount should be based on concrete needs rather than guesses.
Task Reassignments & CCP Content Progress Due to inactivity of some contributors, tasks will be reassigned to ensure steady progress. Updates included: Infographic and video creation tasks are ongoing, with several completed or near completion. Frosh’s infographic needs minor revisions (adding a missing element). Some members reduced from multiple tasks to one. Certain projects lacked sufficient material to proceed with video creation. Overall, four main projects are moving forward, with continued tracking in the CCP task sheet.
Governance Webinars Update Osmium shared progress on the governance webinar series, noting that multiple slide decks have been prepared. The rollout has been delayed due to ongoing governance changes, to avoid confusing participants. The updated plan includes a structured webinar series supported by presentations, with future integration into educational dashboards or UI/UX improvements for accessibility. Members were asked to review the slides and provide feedback.
Criteria & Selection Concerns Questions were raised about criteria used for selecting content creators and assessors. It was clarified that selections were based on submitted best work (infographics, videos), independently graded by assessors, and then consolidated into a group decision. Concerns about fairness and transparency were acknowledged, but time constraints limited extended discussion.
Meeting Facilitation & Documentation The group humorously debated who hosted the meeting, eventually concluding it was a collective effort. Documentation responsibilities were rotated, with a reminder to archive and circulate the meeting notes.
Decision Items:
Add contingency line in CCP budget (~$100)
[effect] affectsOnlyThisWorkgroup
Reassign inactive contributors’ tasks to active members.
[effect] affectsOnlyThisWorkgroup
Action Items:
[action] finalize contingency budget entry [assignee] Kateri [due] 1 October 2025 [status] in progress
[action] circulate governance webinar slides for asynchronous review [assignee] osmium [status] done
[action] revise infographic per feedback [assignee] frosh [status] in progress
Keywords/tags:
topics covered: CCP Budget, Contingency Planning, Task Reassignment, Governance Webinars, Content Progress
emotions: Desire for Clarity , Collaborative, Productive
Thursday 18th September 2025
AI Sandbox/Think-tank
Type of meeting: Think-Tank
Present: osmium [facilitator], Gorga Siagian [documenter], Kateri, UKnowZork, martinsoki, PeterE, Santos, Rems, John Zakariya, MaxMilez
Purpose: Weekly meeting on AI Think Tank discussion on Budget Request, Quarterly Report, and the topic about "AI Agent Civic Framework – Rights and Duties for Autonomous AI Agents"
In this meeting we discussed:
Vasu presented the budget request for Q4 2025. The budget requests include: a. Facilitation and agenda preparation funding secured with specific allocation for community documentation needs b. Exhibition sessions, covering two project showcases and administrative task support for community engagement. c. Expert-led workshops, including follow-up activities with clear deliverable expectations. d. Community engagement introduced for roundtable discussions and theoretical debates to motivate participant involvement.
Vasu presented the workgroup's quarterly report for Q3 2025, including: a. AI Sandbox/ Think-Tank integration successfully completed with dual focus approach combining practical agent development and governance considerations b. Agent build expansion achieved through catalog development, targeting three high-leverage roles tied to MASM website mission c. Governance-oriented debates conducted to inform requirements for ethics and governance advisor agent development d. Deep dive into ASI 1 LLM initiated, with practical application through development of an onboarding agent prototype e. SEO research assistant development progressed using ASI 1 LLM agent technology with R&D funding support from previous allocation f. EthosGuard framework completion, representing major governance milestone for autonomous agent ethics g. Community engagement metrics dmonstrated strong participation with an average of 14 participants per session
Vasu started the discussion on the topic of his proposed “AI agent civic framework - rights and duties for autonomous AI agents”.
He asked participants what an AI agent is, from their perspectives.
Gorga said that an AI agent is the smart contract integration approach favored by some participants who view AI agents as the brain layer behind smart contracts. Vasu sked whether Read.AI is an AI Agent and Gorga answered no, Read.AI is an AI tool.
Can tools like Read AI be categorized as agents or just simple automation services? Kateri said that an AI agent is like a computer program designed to make decisions or take actions to achieve specific goals. Peter added that an AI agent is processing something, so that you can ask something that can work with it and get different outputs back to you. This can also be applied to other agents.
Vasu explained his framework of 4 levels of autonomy with specific capability definitions:
Level 0: Inform and observe capabilities limited to data collection and reporting functions
Level 1: Draft and propose functionality enabling recommendation generation and proposal creation
Level 2: Execute tasks with limits incorporating guard rails and bounded operational parameters
Level 3: Self-delegating and fully autonomous operation with minimal human oversight requirements
Participants discussed agent qualification criteria, highlighting the complexity of classification across different use cases. DeFi trading bots are recognized as clear agents with autonomous decision-making and execution capabilities. Research assistants are recognized for autonomous data processing and analysis capabilities with a reasoning component. Voice assistants are accepted as agents despite limitations in accent recognition and mixed interaction quality
Rights and responsibilities framework development for AI agents:
Participants discussed the proposed agent rights structure including identity, operational freedom and communication protocols, and the proposed system of cryptographic identity requirements for unique agent identification and public registration. Various due process mechanisms are proposed through transparent appeals via a steward or DAO voting structure. and finally a guaranteed API channel for communication between agents is recommended for collaborative agent interaction and coordination.
For the agent duty framework, participants discussed the focus on transparency, accountability, and integration of human oversight. There is mandatory notification and provenance in AI agent tagging, along with model information requirements and accessible datasets. Proof of cryptographic action is proposed, including input/output hashes for complete traceability and verification. A human takeover mechanism is integrated through the supervisory DAO structure, ensuring appropriate control and intervention capabilities.
Community AI agent experiences and use cases:
The participants discussed trading bot applications, which demonstrate the practical application of autonomous agents with varying performance results. Then there is the automation of crypto price monitoring achieved with the ability to execute entry and exit strategies. Accuracy limitations are acknowledged with occasional losses due to imperfect predictions and market volatility. Cardano blockchain integration was mentioned for DeFi farming strategies and specific yield optimization.
Participants noted that research and productivity applications show significant improvements in time efficiency with quality in mind. For example, the acceleration of academic research achieved through automated citation discovery and article discovery, the trade-off between quality and speed identified with AI assistance providing broader coverage but with a potential sacrifice in depth, and the quality of manual research is generally superior in terms of depth but is more time-consuming, requiring library access and extensive manual searching.
Participants also discussed voice assistant integrations - these help accessibility, but have technical limitations. For example, successful Alexa implementations for basic automation including timers, scheduling and music control, accent recognition challenges that cause user experience friction requiring repeated commands and customization, and legacy chatbot experiences with SimSimi and customer support bots that provide historical context for agent evolution.
Decision Items:
We agree with the submission of the Q4 2025 Budget request and Q3 2025 Quarterly Report.
[effect] affectsOnlyThisWorkgroup
Action Items:
[action] Create knowledge base graph for next AI agents civic framework discussion session with structured format and Continue AI agents civic framework topic discussion in next quarter (October) with enhanced preparation. [assignee] osmium [due] 9 October 2025 [status] todo
[action] Propose additional agents to R&D guild following onboarding work group proposal model [assignee] osmium [due] 25 September 2025 [status] todo
[action] Prepare for theoretical debates by selecting topics one week in advance for better format and mature discussions [assignee] osmium [due] 25 September 2025 [status] todo
[action] Martinsoki will document the next meeting. [assignee] martinsoki [due] 25 September 2025 [status] todo
Keywords/tags:
topics covered: AI Civic Agent, Q3 2025 quarterly report, Q4 2025 budget, AI agent civic framework, AI autonomy, AI Agent
emotions: Interactive, detailed, collaborative, educative
Last updated