Mirrored from filpgf.io — ProPGF Batch 3 (Karma program 1479, application 6a293f9add6e8419c8d7c434, status: pending). Contact details redacted; canonical application lives on filpgf.io. 1.1 Project Name INTO Protocol: Filecoin-backed Integrity Layer for AI Agent Workflows 1.2 Project Github https://github.com/intoprotocol/INTO 1.3 Project Website https://intoprotocol.com 1.4 Team Lead/Point of Contact SliverC 1.5 Category [ "RFPs (Coming Soon)" ] 1.6 Open Source Status Partial 2.1 Project Summary INTO(Integrity Network for Trusted Outputs) Protocol is a Filecoin-backed infrastructure layer for AI-generated work integrity. The first phase will build INTACT Layer, a tamper-evident trace, proof, backup, and verification MVP for AI agent workflows. AI agents are increasingly being used to generate documents, make recommendations, operate business workflows, retrieve enterprise data, and complete multi-step tasks. However, most AI-generated work today lacks a standard way to prove what happened: what inputs were used, what files were referenced, what steps were executed, what output was generated, and whether the final evidence was preserved without tampering. INTACT(Integrity layer for trusted agentic compute traces) Layer solves this by allowing AI applications to submit standardized workflow traces, generate trusted output proof packages, back up evidence files through Filecoin-backed storage, and verify proof records through an API or public verification page. The grant-funded scope is intentionally focused on the core infrastructure: workflow trace schema, proof package generation, Filecoin-backed evidence backup, verification API, and 2–3 pilot integrations with AI startup partners. Enterprise dashboards, RWA modules, advanced compliance reporting, private deployments, billing systems, and customized enterprise modules are planned as future commercial extensions and are not included in the first grant scope. By integrating with real AI agent startups, INTO Protocol will demonstrate how Filecoin-backed storage can become a practical trust layer for AI-generated work, creating reusable infrastructure for AI workflow integrity, evidence preservation, and verification. 2.2 Who does this work support? [ "Pods", "Onramps", "Storage Providers", "Network Infrastructure", "Other" ] 2.3 Total Funding Requested (USD) 138,000 2.4 Milestones & Budget [ { "title": "M1: Protocol Design, Workflow Trace Schema & Technical Architecture", "description": "This milestone defines the technical foundation of INTO Protocol and the first grant-funded module, INTACT Layer. The goal is to turn the concept of “AI-generated work integrity” into a concrete architecture that AI applications can integrate with.\n\nThe focus is not to build a full enterprise product yet. Instead, this stage will define the core protocol components: standardized AI workflow trace schema, trusted output proof package structure, verification API design, and Filecoin-backed evidence backup flow.\n\nThis milestone is important because AI agent workflows are highly variable. Before implementation, the project must define a reusable structure for recording workflow events, hashing inputs and outputs, linking evidence files, and generating proof packages that can later be verified by other systems.", "dueDate": "2026-08-31", "fundingRequested": "20,700", "completionCriteria": "- INTO Protocol technical architecture document\n- INTACT Layer workflow trace schema v0.1\n- Trusted output proof package schema v0.1\n- API specification for trace submission, proof generation, evidence backup, and verification\n- Filecoin-backed storage integration design\n- Data model for tasks, agents, workflow events, files, proof packages, and storage records\n- MVP implementation roadmap\n- Technical risk assessment\n- Initial GitHub repository with README, roadmap, schema drafts, and open-source plan" }, { "title": "M2: INTACT Layer Core & Output Integrity API", "description": "This milestone builds the core INTACT Layer service. The goal is to enable AI applications to submit workflow traces and generate trusted output proof packages for individual AI agent tasks.\n\nThe implementation will focus on a simple developer-facing API rather than a heavy enterprise dashboard. AI startups should be able to integrate by submitting task metadata, input hashes, output hashes, file references, workflow events, agent IDs, timestamps, and task IDs.\n\nBy the end of this milestone, INTO Protocol should be able to receive AI workflow trace data, normalize it, record integrity metadata, and generate machine-readable proof packages that can be reused for verification and storage backup.", "dueDate": "2026-10-31", "fundingRequested": "48,300", "completionCriteria": "- AI agent workflow data can be submitted through API\n- The system can generate a structured proof package for each task\n- Input hash, output hash, metadata, timestamps, and workflow trace are recorded\n- Proof package output is machine-readable and reusable for verification\n- Developers can integrate the API using documentation or sample code\n- At least one internal demo workflow can generate a complete proof package" }, { "title": "M3: Filecoin-backed Evid …[truncated] 3.1 Impact pathway Output: The grant will deliver INTACT Layer, the first MVP of INTO Protocol. This includes standardized AI workflow trace schema, trusted output proof package generation, Filecoin-backed evidence backup, verification API, public verification page, developer documentation, open-source release, and 2–3 AI Agent startup pilot integrations. Outcome: AI startups will be able to integrate a reusable proof and trace layer into their AI agent workflows. Instead of generating outputs that are difficult to verify later, their systems will be able to produce structured workflow traces, proof packages, and Filecoin-backed evidence records. Application developers will gain a practical way to show what an AI agent did, what data or files were referenced, what output was generated, and whether the related evidence was preserved. Pilot integrations will test this model in real AI startup workflows. Impact: INTO Protocol creates a Filecoin-backed trust layer for AI-generated work. It turns Filecoin’s verifiable and durable storage properties into a practical infrastructure component for AI workflow integrity. Over time, this can help Filecoin capture recurring storage demand from AI agent workflows, AI-generated work products, enterprise AI systems, and proof-based AI applications. It also creates reusable schemas, proof package formats, verification APIs, and integration patterns for future AI infrastructure products built on Filecoin. 3.2 Verification metrics |Metric |Data source | How it's measured| Target (end of grant) | |AI Agent Startup integrations |partner confiramtion |AI startup workflows integrated | 2+| |onchain paid storage |storage records |onchain source |depends on clients | INTO Protocol will be verified through technical delivery, open-source release, pilot integrations, proof package generation, Filecoin-backed storage usage, and verification API usage. 3.3 References I'm operating AI tutorial and education communities with several thousand followers, have access to many AI developers, AI builders, and AI agent startup founders. 4.1 Monthly Operating Burn [ "$10-$100K (small team)" ] 4.2 What % of total team monthly burn depends on this grant? 80% 4.3 If this grant is not awarded, what happens? Our team is confident that we can complete the development of this business line within six months and successfully connect with at least 2 AI Agent startups to purchase our services. 4.4 Core Team We are an independent three-person technical team led by Sliver Chuan, with 2–3 years of hands-on experience building and delivering AI workflow products. Our core strength is not only AI development, but AI implementation. We have worked on local AI deployment, customized AI agent systems, token/model routing tools, AI workflow automation, enterprise AI process delivery, and full AI solution implementation. We have helped teams move from AI ideas and demos into real workflows that users can operate. We have provided technical services to AI infrastructure and AI workflow projects, including Olares-related technical work. Through this work, we gained experience with local-first AI infrastructure, private deployment, data sovereignty, user-controlled compute, application isolation, and AI workflows running in user-owned environments. This background is directly relevant to INTO Protocol because the project is also about trust in AI-generated work: how AI workflows are recorded, how evidence is preserved, and how outputs can be verified. We have also worked with AI implementation and productization projects such as AI customer acquisition systems, customized intelligent-agent platforms, local deployment, and enterprise AI workflow delivery. Related project materials describe AI implementation as embedding AI into real business workflows, configuring accounts, models, networks, and permissions, and transforming high-frequency tasks into AI workflows and digital employee assets. This experience gives us a practical understanding of how AI agents operate in customer-facing environments. In addition to technical delivery, we operate AI education and tutorial communities with several thousand followers. This gives us direct access to AI builders, AI developers, and AI agent startup teams. We have already spoken with AI agent companies about the need for proof, traceability, and output verification, and several teams have expressed strong interest in trying this direction. We believe INTO Protocol is a natural extension of our previous work. After helping teams build AI workflows, we see the next missing layer: those workflows need standardized traces, trusted output proof packages, Filecoin-backed evidence backup, and verification APIs. This is exactly what the INTACT Layer MVP will deliver. 4.5 Has your team received a ProPGF grant or funding from PLFIF before? [ "No" ] 5.1 Key risks & dependencies The project could become too customized for each pilot.-- However, The grant scope requires pilot integrations but also requires reusable schemas, API docs, developer guides, and sample integrations. Pilot-specific work will be abstracted into standard workflow trace types and proof package formats. Anything else you want to share that we didn't ask? My team is well-positioned to build this because we have practical AI development experience, AI education/community reach, prior AI workflow deployment experience, and access to AI agent startups. Based on earlier market research, AI agent companies have shown strong interest in trying and potentially adopting this type of proof and trace layer. The first phase is intentionally narrow. It will not include enterprise-grade dashboards, RWA modules, advanced compliance reporting, billing, private deployment, or custom enterprise products. Those are future commercial extensions. The grant scope will deliver the reusable foundation: workflow trace schema, trusted output proof package, Filecoin-backed evidence backup, verification API, open-source release, and 2–3 pilot integrations. Contributing to Core Infrastructure? N/A — INTO Protocol is not a Filecoin core protocol implementation or client dependency. INTO Protocol is an AI infrastructure project built on top of Filecoin-backed storage. The first grant-funded phase will build INTACT Layer, a tamper-evident trace, proof, backup, and verification layer for AI agent workflows. It helps AI applications generate trusted output proof packages and preserve evidence files through Filecoin-backed storage. Objective 1 Direct Objective 2 Indirect Objective 3 Direct Open Source Context INTO Protocol will be partially open source during the first phase. The grant-funded INTACT Layer MVP will open source the workflow trace schema, trusted output proof package schema, verification API specification, developer documentation, sample integration code, and selected core components for generating and verifying proof packages. Some components may remain closed source or restricted in the initial phase, including private pilot integration details, security-sensitive implementation logic, abuse-prevention logic, private customer data, and managed service operations. The goal is to make the core proof, trace, and verification model reusable for AI developers and Filecoin builders, while keeping sensitive pilot and operational components secure.