# Proof Over Promises | Deepthi Kumar (o1Labs) at ETHConf

- Channel: [ETHGlobal](https://streameth.org/ethglobal)
- Date: 2026-07-09
- Duration: 19:47
- Topics: O(1) Labs, Deepthi Kumar, Mina Protocol, ETHCom, zero knowledge proofs, ZK, verifiability, verifiable computation, enterprise workflows, policy enforcement, AI agents, agentic systems, AI threat actors, zero days, treasury transfers, institutional finance, identity verification, Google Wallet, Microsoft, EU digital identity, SNARKs, privacy, scalability, ZK rollups, cryptographic receipts, audit trails, proof aggregation, composability, vendor integrations, agent delegation, trust
- Watch: https://streameth.org/watch/yt-nuHqSE44OBk
- YouTube: https://www.youtube.com/watch?v=nuHqSE44OBk

## Description

In this talk, Deepthi Kumar, co-CEO of O(1) Labs, the team behind Mina Protocol, makes the case for moving from trust based to proof based enterprise systems. She frames the central problem facing every business in the coming years: increasing reliance on systems, vendors, and AI agents that you cannot actually check. Her thesis fits in one line: make trust verifiable. Rather than handing over more promises about what systems can do, organizations should hand over more proofs. She notes that zero knowledge proofs have quietly gone mainstream beyond cryptography papers and private tokens, with Google shipping ZK age verification in Google Wallet, Microsoft integrating ZK based credentials into its identity offering, and the EU mandating ZK based digital identity for member states.

Deepthi explains that ZK has historically been used for two purposes (privacy and scalability through ZK rollups), but proposes a third use case: verifiability, which is an inherent property of ZK rather than something bolted on. A SNARK at its core is verifiable computation. She argues this should be applied to enterprise workflows where policies govern high stakes operations, citing two converging forces driving the urgency: AI as an offensive weapon (with the first AI generated zero days being reported and threat actors using AI across attack lifecycles), and AI as an actor (agents acting on our behalf at a scale existing guardrails cannot match). The proposed shift is from a model where you trust systems to act, to one where the highstakes action is logged until a proof shows up validating that all policies were enforced. She describes ZK as a "knot tree" that cannot be bribed. The mechanics are three steps: encode existing policy into a ZK circuit running in parallel to existing enforcement, verify the proof against the expected statement, then execute the operation. Benefits include cryptographic receipts, per action audit trails (finer than sample checks), and composability across multi organization workflows through proof aggregation. She closes by emphasizing this is additive not replacement (preserving security heterogeneity), incrementally adoptable starting with one narrow high value flow, and applicable to institutional finance money movement, AI agent identity and delegation, and any vendor integration with policy enforcement requirements.

00:00 Introduction
00:10 Why You Cannot Check Your Vendors and Agents
00:54 Proof Over Promises
01:33 ZK Has Quietly Gone Mainstream
02:17 Google, Microsoft, and the EU on ZK
02:46 The Two Established Uses: Privacy and Scalability
03:42 Proposing Verifiability as the Third Use Case
04:17 Why Verifiability Is Inherent to ZK
05:05 Why Now: Two Forces Pushing in the Same Direction
05:50 AI on the Offense
06:26 AI as an Actor on Your Behalf
07:01 Shifting From Trust to Proof Authorization
07:47 ZK as a Knot Tree That Cannot Be Bribed
08:22 Treasury Transfer as a Concrete Example
09:24 Verifiable Workflow in Three Steps
10:03 Encoding Policy Into a ZK Circuit
10:44 Verifying the Proof Statement
11:27 Cryptographic Receipts
11:54 Per Action Audit Trails
12:19 Composability Across Multi Organization Workflows
13:44 Not Rip and Replace, Additive Layer
14:51 Why ZK Adoption Can Be Incremental
15:45 Applications: Institutional Finance Money Movement
16:27 AI Agents Proving They Followed Policy
16:53 Agent Identity and Delegation Chains
17:21 Enterprise Operations and Vendor Integrations
18:02 Key Takeaway: Add a ZK Layer to High Stakes Operations
19:01 Closing

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _  

*ETHConf 2026*
ETHConf is a 3 day event for founders, industry leaders, and builders who are excited about the possibilities of building on top of Ethereum.

Connect with 2,000+ top innovators in crypto, finance, technology, and policy at our inaugural three-day event packed with showcases, demos, partnerships, and conversations shaping the future of the global economy.

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _  

✅ *Follow ETHConf*
X: https://x.com/ethconf
Website: https://ethconf.com

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _  

🎟️ Get your early bird tickets for ETHConf 2027: https://ethconf.com/2027#tickets
🎤 View the full ETHConf 2026 Speaker Schedule: https://ethconf.com/schedule

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _