London Voice

chevron-icon
Back
Top
chevron-icon
project-presentation-img
matt_factorylabs
Project Owner

London Voice

Expert Rating

n/a
  • Proposal for BGI Nexus 1
  • Funding Request $50,000 USD
  • Funding Pools Beneficial AI Solutions
  • Total 3 Milestones

Overview

London Voice is an AI-powered, decentralized governance application designed to capture authentic voices from local communities and help them effect meaningful change. By gathering nuanced, ground-level perspectives, London Voice merges voting and AI-driven sentiment analysis to spotlight issues that truly matter. Participants retain ownership of their data, contributing pseudo-anonymously through Soulbound Tokens (SBTs). The result is a collective, community-trained AI “voice” that can inform better policy decisions, amplify underrepresented perspectives, and support real-world interventions

Proposal Description

How Our Project Will Contribute To The Growth Of The Decentralized AI Platform

London Voice aligns with BGI’s mission by using AI and blockchain to enable inclusive civic engagement and ethical data use. It addresses gaps in representation, empowering communities to express real concerns and derive shared solutions. By providing transparent, data-driven insights to local decision-makers, it accelerates social innovation, fosters equity, and demonstrates how AI can ethically enhance civic well-being.

Our Team

Our team at Factory Labs unites blockchain experts, AI researchers, and civic-tech innovators. Led by Dr. Nick Almond (PhD, governance researcher) and co-founders with deep domain expertise in smart contracts, Web3 identity, and decentralized data platforms, we blend academic rigor with real-world product delivery. Supporting members include designers, community managers, and social scientists committed to ethical, inclusive technology solutions.

AI services (New or Existing)

Hypergraph RAG pipeline

Type

New AI service

Purpose

Hypergraph RAG pipeline ingests token-weighted votes and free-form text into a semantic hypergraph. It uses retrieval-augmented generation to create a “community voice” LLM reflecting actual community data. Structured (votes) and unstructured (stories) inputs link in real time ensuring relevant context. This grounds AI outputs in validated feedback reducing errors and building an ever-evolving community-driven knowledge base.

AI inputs

- Quadratic voting data (token-weighted) - Free-form text (stories proposals) - Semantic hypergraph (topics & co-preferences) - Ongoing updates from assemblies/user feedback

AI outputs

- Contextual AI responses for top-voted issues - Community-informed LLM “voice” - Summaries and proposals from real-time data - Draft policy ideas tied to shared sentimen

Consensus Insights Engine

Type

New AI service

Purpose

Consensus Insights Engine analyzes hypergraph data for co-preference patterns alignments and priority clusters. By detecting connected topics and overlaps in voting behavior or text feedback it reveals major points of agreement or conflict. Summaries highlight urgent concerns guiding stakeholders on where the community converges or diverges. These insights shape impactful broadly supported interventions.

AI inputs

- Hypergraph data (votes + text) - Co-preference metrics (adjacency overlaps) - Tags timestamps relevant metadata - Ongoing updates from RAG pipeline

AI outputs

- Ranked lists of top priorities - Cluster maps of shared themes - Conflict alerts needing resolution - Data-driven recommendations for policy

Semantic Similarity Binary

How it will be used

We use this service to compare short text snippets gathered from community inputs (e.g. proposals or feedback) and determine if they convey similar ideas. By returning a binary score (1 = high similarity 0 = low) we can cluster related suggestions reduce duplicates and streamline consensus-building in our platform.

Hate Speech Detection

How it will be used

We integrate Hate Speech Detection to ensure user-submitted content remains respectful. Any flagged text (e.g. “hate” “abuse” “spam”) triggers our moderation workflow aligning the platform with ethical guidelines. This maintains a safe environment where all voices can be heard constructively.

Multilingual Speech Recognition

How it will be used

We employ Multilingual Speech Recognition to transcribe audio content during live assemblies or voice submissions. The service detects the spoken language automatically (e.g. English French German Chinese) and outputs text in the desired language. This fosters inclusivity for non-native speakers and streamlines voice-based voting or testimony.

Company Name (if applicable)

Factory Labs Ltd

The core problem we are aiming to solve

Democracies worldwide struggle with disconnection between official decision-makers and local community voices. Voters typically only engage in broad elections every few years, leaving a gap where urgent, nuanced issues aren’t addressed. This fosters disengagement, especially among younger demographics, and leads to policies misaligned with real community needs. Existing systems often overlook or under-represent local voices, amplifying frustration, apathy, and inequality in civic life.

Our specific solution to this problem

London Voice is designed to capture local sentiment continuously, rather than once every election cycle.

  • Data Gathering: Citizens share their issues, ideas, and context via a Web3-enabled application, “Influence,” with zero gas fees and intuitive UX.
  • Quadratic Voting & SBTs: Voting occurs via Soulbound Tokens for pseudo-anonymous, Sybil-resistant participation. Quadratic weighting ensures each vote reflects genuine priority rather than raw financial power.
  • AI Synthesis: The text-based feedback and voting data feed into large language models fine-tuned to represent the community’s collective sentiment. This “community-trained AI” surfaces shared priorities and suggests potential solutions or policy interventions.
  • Data Ownership & Rewards: Participants retain rights over their data; no personal identifiers are tracked. Those who provide richer, context-specific input can earn tokenized rewards (“vote mining”).
  • Community Governance: The resulting AI outputs—policy proposals, solution ideas, or requests for support—are governed by a DAO-like structure, allowing residents to refine or veto AI-driven insights.
  • London Voice: A community governed agent, that is both a representitive and lobbyist for the local population.

By continuously listening to localized feedback, London Voice bridges the gap between policy-makers and underrepresented voices, generating actionable, data-backed proposals that can catalyze real-world change.

Project details

Background & Motivation

Contemporary democracies often capture public sentiment only through sporadic elections, leaving urgent local issues to accumulate without structured attention. Studies indicate that fewer than 4% of people in the UK feel truly represented by national decision-makers—a sentiment magnified in urban centers like London, where economic disparities and demographic diversity collide. Younger populations, in particular, consistently show lower voter turnout, underscoring a sense of powerlessness in traditional political processes.

Communities on the ground face profound and evolving challenges. From housing crises to the lack of accessible healthcare, these issues affect everyday lives in complex, rapidly changing ways. However, the mechanisms for ongoing input—via town hall meetings, sporadic surveys, or comment periods—often fail to harness the rich context and nuanced perspectives of local residents. As a result, critical interventions can be delayed or overlooked. Meanwhile, data that could inform bold, evidence-based policies remain siloed in disconnected systems, limiting politicians’ ability to craft responsive solutions.

Against this backdrop, emerging technologies offer a fresh opportunity. AI and blockchain, particularly when combined, promise decentralized, participatory models that move beyond annual or quadrennial votes. By leveraging AI for large-scale data analysis and blockchain for secure, transparent governance, it becomes possible to capture community sentiment continuously—providing a living map of local priorities that leaders and planners can reference at any time. This is not just about digital novelty; it’s about restoring agency to citizens who feel unheard, and ensuring their insights actively shape policy, urban development, and community initiatives.

London Voice arises from this aspiration. Our core motivation is to create a scalable, human-centered platform that regularly assimilates people’s lived experiences, organizes them into meaningful policy prompts, and presents them back to the community and decision-makers as actionable recommendations. By blending democratic ideals with AI’s power to synthesize vast input, we envision a future where any neighborhood—like Barking or Archway—can rally around a shared digital “voice” capable of influencing resource allocation, urban planning, and civic engagement on an ongoing basis. In doing so, we aim to demonstrate how modern technology, carefully designed with transparency and inclusivity at heart, can mend the rift between local populations and the governance structures meant to serve them.

The Vision

London Voice envisions a future where AI-driven governance and human-centered design intersect to create an inclusive, transparent forum for collective decision-making. By leveraging Web3 technologies, we ensure:

  • Pseudonymity: Protecting participants’ personal data through Soulbound Tokens (SBTs) that grant voting rights without tying votes to real-world identities.
  • Open Participation: Anyone with an SBT can contribute issues and perspectives, bridging the gap between offline community activism and digital democracy tools.
  • Ongoing Adaptation: The AI agent, tuned to reflect local sentiments, is not static. It evolves as conditions change, building a “living snapshot” of collective beliefs and aims.

Pilot Projects & Real-World Interventions

  • Barking Voice: Our first major intervention focuses on a real community group in Barking lobbying local developers. Although the area’s density is increasing, vital community services (e.g., healthcare, green spaces) are not expanding at pace.
    • Local Impact: By gathering residents’ direct feedback—needs, fears, aspirations—London Voice will synthesize these into a coherent community mandate. This can be presented to developers and local authorities, increasing pressure to include essential community amenities.

Evidence of Established Community Engagement and Need for London Voice

We have presented London Voice recently to the Thames Life community group, which would be our touch point with the community. Here are some select quotes from the community, which is primed and ready to be activated with this tool. Recorded and transcribed directly from our presentation event.

Quotes from the Barking Community Presentation

Kevin on Needing Reliable Information

“The most important thing, I think for us lot… is information. You can’t change or measure anything without good information we can rely on.”

Suggested Tag: #TrustedInformation


Alan on Feeling Gaslit by the Council

“I’m fed up with being gaslit by the council all the time… we need to keep at them, or they’re just going to ride over us.”

Suggested Tag: #Accountability


Eric on Youth Opportunities

“As a young person, for me the most important thing is more opportunities for younger people.”

Suggested Tag: #YouthEngagement


Pete on “Consultation” Fatigue

“They never do anything for what we want. They ask us to go to their consultations, but they’ve already made the plan… They just want our names so they can say, ‘Oh look, we talked to them.’”

Suggested Tag: #MeaningfulConsultation


Local Resident on Spatial Ownership

“I live north of the A13. I think it’s about people having ownership of spatial things that affect them.”

Suggested Tag: #SpatialOwnership


Resident on Green Spaces & Community Control

“We need to protect our communities… When they’re building all these high-rises and there’s no gardens for people. We need to change the mindset of developers.”

Suggested Tag: #EnvironmentalPriority


Community Member’s Endorsement of London Voice

“We should get behind this. I think it’s a great idea. If you can do it, you’ve got an awesome tool.”

Suggested Tag: #CommunitySupport

By capturing and synthesizing direct feedback like this, London Voice can pinpoint urgent community concerns (lack of reliable information, mistrust of local authorities, unmet youth needs, etc.) and aggregate them into actionable focus areas. These quotes demonstrate both the demand for inclusive, continuous engagement and the enthusiasm for a tool that amplifies resident voices.

Using consultations like this, London Voice can generate the semantic tags that define community voting options. For example:

  • Archway Voice (Islington Council & Architecture Association): We plan to activate a partnership with Islington Council and the Architecture Association to create Archway Voice. This initiative will similarly leverage relationships with existing grassroots organizations, enabling them to share hyper-local insights and shape development decisions.
    • Collaboration Model: Students, activists, and council representatives convene assemblies, feed input into our Influence app, and collectively fine-tune a localized AI model. The outputs—policy proposals, design suggestions, or resource allocation plans—can then be championed by local governance bodies.

How London Voice Works

  1. Submission Phase: Residents or community members submit issues and ideas—ranging from the cost of living to the need for a new community health clinic.
  2. Voting & Data Collection: Quadratic Voting ensures balanced representation of intensity (not just majority rule). SBT-based identity gating prevents spam and manipulations.
  3. AI Synthesis: We utilize large language models (LLMs) to interpret and aggregate these votes and text-based inputs, producing a ranked list of community priorities and an accessible summary of underlying sentiment.
  4. Agent Governance: The community collectively governs the AI’s parameters and how it may propose solutions. If the AI’s responses or generated proposals conflict with community values, participants can adjust training data, add disclaimers, or set new alignment criteria.

Benefits & Impact

  • Local Empowerment: London Voice fosters a culture where local decisions are shaped by lived experiences rather than top-down mandates.
  • Evidence-Based Lobbying: By capturing genuine community sentiment, the AI agent provides a credible narrative that policymakers and developers must consider.
  • Transparent Decision-Making: All voting records, discussion prompts, and AI fine-tuning steps are publicly available, building trust and accountability.

Data contributions are pseudo-anonymous, and no personal identifiers are stored. Each participant controls how their data is used for model training. The codebase is fully open-sourced under GPLv3, allowing external audits of data handling procedures and model behaviors. In this version team members will retain full moderation rights of the platform, to allow data that breachers a set of clear ethical guidelines to be removed from the platform. In future iterations this moderation function can be decentralised to a set of community moderators.

Future Expansion & Tokenized Rewards

Although we are not implementing direct tokenized rewards in this proposal, we anticipate that future iterations may reward participants for high-quality or in-depth contributions. This aligns with our long-term vision of a people-owned AI, where those who provide valuable data are fairly acknowledged and potentially compensated.

Conclusion

By bringing together communities, developers, councils, and AI specialists, London Voice stands as a practical demonstration of how emerging technology can inform, empower, and unify. Whether it’s addressing the lack of green spaces in Barking or co-creating an architectural blueprint for Archway, London Voice aims to channel collective intelligence into tangible, real-world change—one neighborhood at a time.

Existing resources

We are extending and customizing our existing “Influence” dApp for user onboarding, quadratic voting, and data collection. We also build upon well-established LLM frameworks (e.g., open-source large language models), ensuring minimal overhead and leveraging proven technology.

Open Source Licensing

Custom

We will use GPLv3 for the entire codebase, including smart contracts, front-end, and supporting libraries, with no proprietary exceptions.

Links and references

Factory Labs: https://ldnvoice.factorylabs.org/

Additional videos

Voice App Preview: https://youtu.be/XSwPnEp8VPw

Was there any event, initiative or publication that motivated you to register/submit this proposal?

A personal referral

Describe the particulars.

A refferal from Esther Galfalvi - Decentralisation Program Lead.

Proposal Video

Placeholder for Spotlight Day Pitch-presentations. Video's will be added by the DF team when available.

  • Total Milestones

    3

  • Total Budget

    $50,000 USD

  • Last Updated

    24 Feb 2025

Milestone 1 - Prototype & AI Inference Demonstration

Description

This milestone focuses on building and showcasing the core AI agent’s functionality using historic and synthetic datasets. We will demonstrate how the system constructs a semantic hypergraph of co-preferences and generates insights sentiment snapshots and early-stage solution proposals. Core AI & Graph Module Implement the AI workflow for processing votes and text data forming a semantic hypergraph that captures co-preferences. Data Integration & Synthetic Demo Load synthetic or anonymized historic data to simulate community inputs. Validate the inference pipeline producing example outputs such as ranked priorities and brief solution sketches. UI & Technical Documentation Release a limited-access front-end showing how data is ingested and visualized. Provide a short written overview of the model architecture inference process and early results.

Deliverables

Prototype Environment: A testbed application demonstrating ingestion of voting data text comments and AI inference. Tech Walkthrough: Short video or screen-share session illustrating end-to-end data flow. Report: A brief technical document summarizing the approach hypergraph construction key metrics and early insights.

Budget

$15,000 USD

Success Criterion

The AI pipeline successfully parses synthetic/historic data and outputs coherent preference graphs. A minimum of one demonstration session (video or live) showcasing the prototype to stakeholders. Positive feedback from internal testers validating the system’s readiness for real-world data.

Milestone 2 - Real-World Pilot & Community Engagement

Description

This milestone delivers a functioning instance of London Voice for a live community—focusing on a group in Barking London. Participants will use the app to express local concerns (like inadequate healthcare or green spaces) while the AI synthesizes these inputs into actionable outputs. Public-Facing Application Complete user-friendly front-end with SBT-based authentication quadratic voting and intuitive submission forms. Community Onboarding Partner with Barking community activists to distribute SBTs run QV sessions and gather detailed feedback. AI & Semantic Hypergraph in Action Ingest real community data into the AI pipeline generating localized priority lists and solution hints (e.g. arguments for more healthcare or community spaces). Refinement & Feedback Loop Collect user feedback on usability and alignment of AI outputs refining the system’s interface and model prompts as needed.

Deliverables

Live dApp accessible to the Barking community complete with robust user onboarding. Community Data & Reports: Real-world QV results co-preference mappings and an AI-generated summary of top local issues. User Feedback: Structured responses and testimonials from local organizers confirming the tool’s practical relevance.

Budget

$20,000 USD

Success Criterion

At least 50–100 Barking residents successfully onboarded and submitting votes/comments. Demonstrable alignment of AI recommendations with genuine community concerns (e.g., identification of local healthcare gaps). Written and video feedback from at least one local activist or group leader confirming usefulness for advocacy.

Milestone 3 - Archway Expansion & BGI Nexus Outreach

Description

Building on the Barking pilot this milestone expands London Voice to additional areas (including Archway in partnership with Islington Council and the Architecture Association) and demonstrates an aggregated “meta-vote” that integrates insights from multiple neighborhoods. It also includes broader dissemination and reporting of findings including a BGI Nexus community intervention. Archway Deployment: Launch a localized instance (“Archway Voice”) with community partners (e.g. local groups council representatives) collecting distinct issues and data. City-Wide Meta-Vote: Implement functionality to collate and compare data across Barking Archway and potentially other neighborhoods providing a London-wide “meta” perspective. Dissemination & Reporting Produce a detailed open report/whitepaper capturing the methodology pilot outcomes community feedback and lessons learned. Conduct a BGI Nexus community intervention or workshop showcasing how the platform aligns with beneficial AI principles and fosters social impact. Scaling & Documentation: Provide how-to guides and modular technical documentation for prospective adoptors ensuring ease of replication.

Deliverables

Multiple Local Realms: At least two distinct realms (Barking + Archway) operational each generating their own data sets and AI outputs. Aggregated Meta-Vote: A city-wide or cross-realm dashboard illustrating shared and divergent priorities. BGI Nexus Event: At least one workshop or showcase targeting the BGI community demonstrating how London Voice furthers ethical and beneficial AI use cases. Comprehensive Report: An open-access document summarizing technical architecture community outcomes ethical considerations and future roadmap.

Budget

$15,000 USD

Success Criterion

Verified usage by at least two communities beyond Barking (e.g., Archway, plus one more if time/resources permit). A finalized city-level “meta-vote” analysis that merges different neighborhoods’ inputs into a holistic priority overview. Positive engagement at the BGI Nexus event—feedback from attendees and stakeholders acknowledging the project’s impact and alignment with beneficial AI goals. Completion and publication of a final report, accessible to external researchers, policy-makers, and the broader Web3/AI community.

Join the Discussion (0)

Expert Ratings

Reviews & Ratings

    No Reviews Avaliable

    Check back later by refreshing the page.

feedback_icon