Max Semenchuk / Simulation Studio
Max shared Simulation Studio (simstudio.w3i.network/reports), a social simulation engine he has been building. The system is fed structured profile data — primarily LinkedIn bios, company connections, and specializations — from a database of roughly 1,000 Web3 network contacts. A user selects a segment (e.g., “public goods,” “educators”), writes a scenario prompt, and the engine runs a structured discussion through AI agents representing those profiles. The output shows vote distributions, group clustering around diverging opinions, and a summary of the simulated discussion.
Max framed the primary use case as pre-conference agenda design and governance modeling: you can cluster participants, stress-test a policy proposal, or run a product question through a synthetic audience before the real meeting. The system also has a smart-suggest feature that identifies the top 20 most relevant profiles for a given question, since including the full database scales cost and runtime proportionally — most of his previous runs cost between 3 cents and $1.50. He also described using it to create a digital twin of a specific counterparty to review a B2B proposal before sending. Max validated the engine by uploading transcripts of past real conversations and measuring how accurately the model predicted group clustering — finding roughly three times better accuracy than random assignment. Brandon connected this to Gregory Landua’s earlier experiments with political agents, and Max confirmed he had reviewed Gregory’s repository as context.
The broader application Max described: a “virtual parliament” model where committees self-arrange around topics, vote, and surface conflicts — a form of synthetic governance that a human can observe and intervene in if something goes wrong. He suggested Regen Tokenomics data (up to 100 registered participants with partial LinkedIn coverage) as a candidate dataset for a future run. Additional intern candidates from his network are available at w3i.network/connect/catalog.
Christian / Developer Internship Program
Christian announced that two interns with developer and AI/ML backgrounds will join the group in June for a 12-week program. He received approximately 100 applications, is using AI to screen down to 8 finalists, then will select 2 and share their CVs and cover letters with the group before an introductory call. He asked for participation from Max and Brandon, and both confirmed they were in.
The program’s goal is to build out what Christian called a Regen software development kit: tools that connect Regen Ledger and Regen Marketplace to applications on Ethereum and Base. A concrete near-term target is integrating eco-credit retirement into a space exploration game being developed by Dan Pittman, a former Regen team member, who has long wanted in-game rewards tied to ecological impact. Christian also noted that a significant portion of the work may simply be documentation — making the existing Regen Compute tooling clear enough that external developers can plug in without needing a walkthrough. Lance’s use of the Regen Compute API for his own project was cited as an example of functionality that already exists but isn’t broadly surfaced.
Max advised structuring the program as weekly sprints with planning at the start and retrospective at the end of each week — short cycles are recommended for new teams. He emphasized the role of a facilitator who acts as a coach rather than a director, and suggested that the interns’ cover letters would reveal how much structure versus autonomy they prefer. Christian agreed to share materials in advance and hold an intro call before the sprint program begins.
Brandon Kelly / Eco Wealth (Vealth)
Brandon shared Eco Wealth, operating under the domain vealth.net, a proof-and-coordination layer for ecological work. The system has three current layers: Packets — readable PDFs and briefs scoped to buildings, food ecosystems, and stewardship opportunities; Receipts — proof artifacts that log what work happened, what evidence exists, what changed, and what next action was unlocked; and Credits — a V0 internal, non-tradable proof-of-stewardship token used to track useful ecological work within the system.
Brandon’s near-term go-to-market is construction safety companies. His background is as a safety coordinator, and he described a workflow where he assesses a site, identifies what needs to be fixed and by whom, and produces a report — with an eco-credit retirement attached to the deliverable. The longer vision is a machine that makes ecological labor opportunities legible and fundable across buildings, food, and environments, drawing on the Living Building Challenge, regenerative agriculture, and the UN’s SDG framework as target states. Brandon is Living Future accredited and cited personal familiarity with the Bullitt Center (a Living Building Challenge project in Seattle) through prior TerraGenesis/Regen Network consulting work.
Brandon also discussed a research protocol he encountered with 27 adapters and many AI providers competing on benchmark problems, and raised whether a proof-of-environmental-work analog could function similarly — paying agents or humans partially for progress on known ecological problems rather than requiring full solutions. He acknowledged the product management challenge: scope versus near-term revenue versus long-term vision are pulling in different directions, and he is currently building via ChatGPT and Claude without using an IDE. Max’s recommendation was to maintain a task log, close one thing before opening another, and identify the single killer feature worth shipping first.
Group / Regen Compute Enterprise Sales
With 36 current subscribers, Regen Compute is stable but growth requires a clearer marketing approach. Brandon suggested targeting enterprises directly — cold outreach to sustainability teams, positioning the compute-offset angle as additive to rather than a replacement for existing voluntary carbon commitments. Christian pushed back: a sustainability officer at a large company already has a voluntary carbon program that covers their whole operation; segmenting compute hours into a separate, smaller program adds friction without obvious benefit, and the platform risks being dismissed before the conversation starts.
Christian’s read is that retail users — individuals who are abstractly concerned about AI energy use — are the stronger product-market fit for Regen Compute, because they are not already inside a corporate carbon program and respond to direct framing around AI’s ecological cost. Brandon countered that Regen Compute’s regenerative credit pipeline is differentiated from standard offset products and could give a corporate sustainability team a unique, additive story for shareholders. Both agreed the pitch needed more work and that an institutional outreach workflow hadn’t been implemented yet.