MENTATRequest a pilot

Stellar Blockchain Infrastructure Agent

Your Stellar ops team is one @mention away.

Tag Mentat in Slack, Teams or the terminal. It reads your Stellar Core and RPC nodes, computes the plan, and waits for your engineers to approve it.

The problem

Teams hesitate to run infrastructure they never learned.

“Integrating Stellar into Spiko’s stack was the most challenging network integration we’ve done so far.”
Spiko engineering blog, February 2026 · now the largest tokenized fund issuer on Stellar
  • 59%

    of institutional investors cite integration challenges in tokenization

    EY, Jan 2026
  • 58%

    want practical training on custody and settlement before they move

    EY, Jan 2026
  • 41%

    of the largest firms name operational complexity as their top challenge

    Broadridge, May 2026
  • 42%

    of financial institutions cite skills gaps as a barrier to digital assets

    PYMNTS and FIS, Oct 2026

01 · Try it

Stellar infrastructure might be hard sometimes. Let Mentat carry it.

Tag it in the tool your team already uses. Mentat reads, diagnoses and plans; changes run only after a named engineer approves. Pick a channel.

Example conversation. When the plan arrives, press Approve.

02 · Away from your desk

It keeps working while you are out.

Mentat watches your nodes around the clock. When something needs a decision, it reaches you on your phone with the cause, the fix and the cost. Approve from the notification, or let it handle what is already in its scope.

  • Capacity, before users notice
  • Upgrades that slipped through
  • Contract storage about to expire
  • Backup providers that drift
  • Disk and archive runway
Saturday, October 3122:47

03 · Security

AI as your teammate, not your replacement.

Rest easy. Mentat is built by a team that already runs SOC 2 and ISO 27001 compliant, bank-grade infrastructure, and it is built to the same standards. The model never touches a shell, a key or a node directly.

  • No keys, ever

    Signing stays in your HSM. Mentat only prepares unsigned transactions.

  • Two person rule

    High risk plans need two named approvers. Only the approved hash runs.

  • Your network, your model

    Runs on premises with the model you choose, even fully offline.

  • Signed audit log

    Every question, plan and command, append only and exportable.

  • Read only mode

    Give auditors and regulators evidence with no write access at all.

  • Where you already talk

    Slack, Teams, Google Chat, terminal, email, Claude and ChatGPT through MCP.

04 · Questions

What security teams ask first.

Can Mentat move funds or sign transactions?

No. It never holds keys. When a contract’s storage needs extending, it prepares an unsigned transaction and places it in your signing queue.

What can it do without approval?

Read only checks, plus a short list of pre approved runbooks you switch on yourself, such as restarting a stalled RPC node. Everything else waits as a plan.

Where does it run?

Inside your infrastructure, next to your nodes, with the language model you choose. Nothing has to leave your network.

Does my team need Stellar experience?

No. Engineers who run Linux and EVM nodes ask in their own terms. Mentat maps each concept to Stellar and shows the command it will run.

What happens if a plan fails?

Every plan carries its rollback. The executor stops at the first failed health check, and the audit log shows which step ran.

MENTATMENTATMENTAT

05 · Pilot

Bring your nodes. We bring the runbooks.

We are onboarding a small group of institutions for the Protocol 30 cycle. Write with your network, node count and timeline.

Pilots start read only. Nothing changes on your nodes until you turn on approvals.