Add live organizational context to stateful agents.
Connect LangGraph (@LangChain) to Roster so agents can resolve who owns, approves, reviews, handles, or receives an escalation without hard-coding people into graph logic.
https://t.co/3GU7VoKqZd
@levie Always insightful, thanks for sharing. Roster helps FDE move AI agents from demos into accountable enterprise workflows by giving every agent and automation a governed way to resolve who should approve, own, review, handle, or receive work. https://t.co/ZmWSEv4iOV
How do today’s platforms answer: “Who should act on this?”
A category-by-category look at agent orchestration, workflow automation, iPaaS, BPM/ITSM, service catalogs, authorization engines, and DIY approaches — and where Roster fits.
https://t.co/bBFPhiPhY3
A full tour of the Roster web UI and its admin features — the console for a Participant Resolution Agent. Model who handles what, connect your directory, resolve who should act, and govern every bit of it.
Roster is a participant resolution agent providing the organizational harness for AI workflows. Roster resolves natural language queries down to the right active users, groups, and roles using your existing directory data.
Start for free: https://t.co/42XVLzfEnV
00:00 What Roster is — the admin console
00:25 Dashboard & reporting
00:51 The data model: projects & participants
01:32 Labels & delegations
02:06 Directory connectors
02:27 Resolve requests & the audit trail
02:54 Identity & access
03:28 Observability: model runs & audit events
03:47 Admin settings & governance
04:02 Playground
04:22 Wrap-up
Agents know how to act; Roster resolves who should own, approve, or sign off — from local Docker tests to the same call in CI. ✅Free deployment: https://t.co/mMdCP2DCIt
Roster in Claude Cowork: Your AI agent knows how — Roster resolves who should actually sign off. Watch the approval chain rewrite itself in real time.
🚀Get started: https://t.co/VoSgD0qw3K
A full tour of the Roster web UI and its admin features — the console for a Participant Resolution Agent. Model who handles what, connect your directory, resolve who should act, and govern every bit of it.
Roster is a participant resolution agent providing the organizational harness for AI workflows. Roster resolves natural language queries down to the right active users, groups, and roles using your existing directory data.
Start for free: https://t.co/42XVLzfEnV
00:00 What Roster is — the admin console
00:25 Dashboard & reporting
00:51 The data model: projects & participants
01:32 Labels & delegations
02:06 Directory connectors
02:27 Resolve requests & the audit trail
02:54 Identity & access
03:28 Observability: model runs & audit events
03:47 Admin settings & governance
04:02 Playground
04:22 Wrap-up
Stop hardcoding your enterprise org chart into AI prompts. Roster lets your AI agents figure out who needs to approve, act, or be notified using your existing directory data via MCP, CLI, REST or Web UI.
https://t.co/VoSgD0qw3K
Roster is a Participant Resolution Agent for agentic workflows. It gives your agents one neutral primitive — resolve — that turns a plain-language question into the right active user, group, or role, in real time, from the directory data you already have. Delegations and a full audit trail are built in.
No hard-coded org charts. No new system of record to maintain.
→ One Docker container. Reach it over MCP, CLI, REST or Web UI.
→ Bring your own model. Vendor-neutral. On-prem or your cloud.
→ Built on the routing & delegation engine Advantys has run in production for 20+ years, across 500+ organizations.
Agents know how. Roster resolves who.
https://t.co/42XVLzfEnV
THE HYPE DIED. THE HARD PART STARTED.
95% of enterprise AI pilots show zero measurable P&L impact (MIT). 42% of companies abandoned most of their AI projects in 2025 (S&P Global). The bubble talk is earned.
But look at what's actually failing — it's rarely the model. It's org debt: approval chains that predate the tool, procurement that outlasts model generations, an ROI bar that finally got real.
The gap was never "does it work." It's deployment. Bolt AI on, and the write-off is inevitable. Rebuild who does what around it, and the numbers start compounding.
Humans are both sides of that ledger — the friction, and the only real source of judgment. The orgs seeing real ROI are the ones that decided, precisely, where a human still belongs.
Roster is built on 20+ years running business-critical workflows for Fortune 1000 companies — not a lab experiment, a tool for the teams in the trenches. It resolves who acts next when the approver list breaks, with an audit trail behind every call.
Agents know how. Roster resolves who.
🚀Early Access this month.
💰 Pricing follows the same logic: pay for resolvable humans, nothing else. Resolutions, agents, and service accounts are unlimited — Free included. No metered token bill.
⚙️ That's Roster. One container, your infra or your cloud. Bring your own model — swap it, downgrade it, run open-weight. Your directory data never leaves.
Cost and control are the same problem. If your org context never leaves your infra, and never touches one vendor's black-box model, you're not exposed to either side of that trade.
🎙️ Alex Karp, on CNBC: frontier labs charge you for tokens that create no value, then absorb your data into a model only they control. His framing: a tax on every company's private knowledge.
Two data points from this week, read together.
📊 @Gartner_inc : the AI spending reckoning is here. Enterprises are scrutinizing bills after 2 years of blind adoption — some paying $200-500/dev/month in tokens for returns that don't show up. 👇
@levie We completely agree. Involving the right people at the right time is critical for agentic workflows. That’s why we’re building a participant resolution agent that provides the organizational harness AI workflows need.