Delos launches Workers, persistent AI agents that integrate into company org charts with dedicated email and Slack accounts

Delos just launched Workers.

Turns AI agents into always-available digital employees with their own email, phone, Slack, Teams, computer, memory, and company-specific training.

Delos says each worker is trained on one business, so the system learns the company’s routines, tools, vocabulary, permissions, and handoff patterns instead of acting like a generic assistant dropped into a random workplace.

The key point is the persistent identity plus long-term memory, because an AI employee only becomes useful if people can contact it like a coworker and trust it to remember past decisions, open loops, customer context, and internal procedures.

@kimmonismus so used to AI, even humans are looking like AI to me now 😭

@vishalsingh2972 I know that feel. It’s like entering the matrix

@kimmonismus persistent memory is the part everyone else skipped

if they cant remember what they said yesterday theyre not workers, theyre one-off scripts

@kimmonismus Delos as in....

@kimmonismus @kimmonismus sounds like a game changer for projects where you need continuity. no more starting from zero every time.

@kimmonismus You cannot rely on an assistant that forgets 50% of what you told him. Perfect memory is our biggest advantage here

@kimmonismus Meine Agents wenden diese Methode bereits seit zwei Jahren an. Zusätzlich statten wir die relevanten Agenten mit Vision-Fähigkeiten aus, um ein umfassendes Verständnis der visuellen Informationen zu gewährleisten.

@kimmonismus “I’m here to save you from building workflows. Don’t tell me how to work. Just set a goal and I’ll figure out how to reach it.”

Sounds awesome

@kimmonismus Love the vision. Magica's already here — persistent agent, full memory, every tool in one place. Context never drops. Try it.

@rohanpaul_ai Send like a great way to introduce information silos and inefficient information flows.

@rohanpaul_ai giving an ai agent its own slack and email is the part that'll actually break things first,

half of enterprise IT isn't built to treat a bot like a headcount

@kimmonismus I think this is the right framing. The question is less whether it can do one task, and more whether it stays coherent over weeks of interaction.

the hard part here is not “memory”

it is identity + permissioning + state consistency

once an agent has email/slack/phone, it becomes an actor in the org graph: it needs scoped credentials, audit logs, revocation, memory garbage collection, and conflict handling when its stored context disagrees with current source-of-truth

persistent agents are useful only if their memory is inspectable and their authority is bounded

@kimmonismus context resets kill more productivity than people realize tbh. $SENT been tackling this with persistent memory and standing worker identities that carry everything forward between tasks, no starting from scratch every time

@kimmonismus Persistent identity is the front feature. The hidden one is the audit trail. An agent with an email, phone, and Slack handle leaves a footprint. The next operator inherits 4 months of decisions they never made. Who gets to read the agent's memory becomes the org chart.

@rohanpaul_ai This is insane