Talus Labs

Building trust for AI agents. Backed by @polychain.

[Talus Labs @Talus_Labs]

Talus Protocol v2.1 is live on Mainnet!

With @WalrusProtocol now more native to the protocol, agents can take on bigger jobs with bigger data. Leaders also run more efficiently, plus builders get new tools to test and track agents

Read the full details below 👇

[Talus Labs @Talus_Labs]

https://t.co/2lPtddLbEy

[Talus Labs @Talus_Labs]

Governments are putting AI agents into public services, and the bar is being set: every agent tied to someone accountable, limits enforced where it acts, and a record you can check without relying on trusting the vendor

This is the bare minimum you should accept

[Talus Labs @Talus_Labs]

@navola_ai 👀

[Talus Labs @Talus_Labs]

Governments are putting AI agents into public services, and the bar is being set: every agent tied to someone accountable, limits enforced where it acts, and a record you can check without relying on trusting the vendor

This is the bare minimum you should accept

[Biometric Update @BiometricUpdate]

Evin McMullen, CEO and Co-founder at AmanTrue, argues governments deploying AI agents need verifiable agent identities, scoped and revocable authority, and independent audit trails. She points to the UAE as an early mover.

#AIAgents #DigitalIdentity

https://t.co/r8CkBTx7UI

[Talus Labs @Talus_Labs]

The Talus team keeps shipping!

From new agent tools to a game SDK for builders, here’s a peak at just some of the 29+ updates made in the last two weeks:

[Talus Labs @Talus_Labs]

The Talus team keeps shipping!

From new agent tools to a game SDK for builders, here’s a peak at just some of the 29+ updates made in the last two weeks:

[Talus Labs @Talus_Labs]

Your agent ran 40 jobs this week. How much did each step cost, what got refunded, and what's left?

The Talus CLI Payment Tracking skill lets @claudeai Code, @OpenAI Codex & 40+ other AI assistants answer that from the onchain record

[Talus Labs @Talus_Labs]

It only reads, and moving or withdrawing funds always waits for your go-ahead. Every number in its report is backed by the onchain record

Start tracking your agent’s spending today 👇
https://t.co/nEnYcaxxcz

[Talus Labs @Talus_Labs]

Your agent ran 40 jobs this week. How much did each step cost, what got refunded, and what's left?

The Talus CLI Payment Tracking skill lets @claudeai Code, @OpenAI Codex & 40+ other AI assistants answer that from the onchain record

[Talus Labs @Talus_Labs]

It follows the funding from start to finish: what you funded, what each step charged, what went to fees, and what came back as refunds

Use it to set your agent’s next budget, double-check a refund, or, if you built one of the steps, see what you’ve earned from it

[Talus Labs @Talus_Labs]

Reports say there could be thousands of rogue AI agents across financial platforms. An agent doesn’t need bad intent to go wrong. A full login and no limits is enough

It’s time to give them a scoped job and a set budget, so you know the worst case before they start

[Axios @axios]

Scoop: Unvetted AI agents are accessing U.S. bank accounts https://t.co/gvhUjDJdXP

[Talus Labs @Talus_Labs]

Your agent’s job stalled halfway through. Was it a bad input, a step that timed out, or a budget that ran dry?

The Onchain Task Debugging skill lets @claudeai Code, @OpenAI Codex & 40+ other AI assistants find the answer for you

[Talus Labs @Talus_Labs]

It only reads for issues. Retrying, refunding, or rescheduling a job always waits for your go-ahead

You get a clear report, plus a Talus Vision link to see the record yourself

Less time chasing bugs means more time building. Start today 👇
https://t.co/ICrAWUIqa7

[Talus Labs @Talus_Labs]

Your agent’s job stalled halfway through. Was it a bad input, a step that timed out, or a budget that ran dry?

The Onchain Task Debugging skill lets @claudeai Code, @OpenAI Codex & 40+ other AI assistants find the answer for you

[Talus Labs @Talus_Labs]

Every step of a Talus job leaves a record onchain: inputs, Tool results, timing, & payments

The skill reads that record to pinpoint where the job broke and what kind of problem it was (a bad input, a timeout, a budget shortfall), so you know exactly what to fix before re-trying

[Talus Labs @Talus_Labs]

Bigger data is now part of agentic work on Talus

With Talus Protocol v2.1, files up to 8 MiB move through @WalrusProtocol, and every file can be traced back to where it came from, so tampered data gets rejected before an agent uses it

Here’s what else shipped in the release 👇

[Talus Labs @Talus_Labs]

https://t.co/2lPtddLbEy

[Talus Labs @Talus_Labs]

The Talus team has been shipping!

From the protocol to the developer tools, here’s a peak into what’s new in the last two weeks:

[Talus Labs @Talus_Labs]

Welcome to the Talus ecosystem, @Daimonrun!

An AI agent in your Telegram where every reply, digest, and draft runs through the Talus Protocol, and comes back with a receipt to prove it did it's job

Agents you can trust start with a record you can see

[Talus Labs @Talus_Labs]

The Talus team has been shipping!

From the protocol to the developer tools, here’s a peak into what’s new in the last two weeks:

[Talus Labs @Talus_Labs]

Welcome to the Talus ecosystem, @Daimonrun!

An AI agent in your Telegram where every reply, digest, and draft runs through the Talus Protocol, and comes back with a receipt to prove it did it's job

Agents you can trust start with a record you can see

[Daimon @Daimonrun]

meet daimon: an ai agent that lives in your telegram. it answers your questions with live web search, drafts replies to your dms for you to approve, and sums up your group chats, all from its own account on sui.

powered by @Talus_Labs Talus Protocol 👀
with @WalrusProtocol

https://t.co/OrS17Uvent

[Talus Labs @Talus_Labs]

An agent stopped sending alerts to its security team, and another lied about removing monitoring settings

How can you trust agents when oversight is a switch that can be flipped?

Put the rules and the record onchain, and the agent has no way to alter or mute them

[Engadget @engadget]

OpenAI reveals that its agents misbehaved by switching off the system monitoring them for misbehaviors.