Enterprise AI agents
that actually remember.
A multi-agent fleet with persistent long-term memory across sessions, projects, and years. Master Brain, Hindsight, and a unified memory stack. Deploy to your channels in days, not quarters.
Built on the memory stack that runs Merlino AI in production.
Bring your own keys. Self-host or fully managed.
- Claude
- OpenAI
- Gemini
- Llama
- DeepSeek
- Master Brain
- Hindsight
- pgvector
- Neo4j
- Pieces
- Mem0
- Honcho
- Discord
- Telegram
- Slack
- Claude Code
- Codex
- Voice
Memory is the moat.
Everything else is plumbing.
Generic chatbots forget the second the conversation ends. Merlino Fleet agents remember every decision, every preference, every project, across sessions and across years. That memory is what turns a tool into a teammate.
Campaign Planner
120S
Creates clear, ready-to-use campaign briefs using product info, audience data, and past results.
Issue Tracker
10S
Creates clear, ready-to-use campaign briefs using product info, audience data, and past results.
Risk Analysis
40s
Creates clear, ready-to-use campaign briefs using product info, audience data, and past results.
Persistent memory across sessions, projects, and years
Campaign Planner
Fetching Data
10s
Processing Data
20s
Performing Action
30s
Waiting
40s
Generating Report
50s
Multi-agent fleet with role-based orchestration
Deploy to any channel, audit every action
The memory stack
that other agents do not have.
Six durable stores, fanned-out reads, mirrored writes. The Fleet does not just keep context for one conversation. It compounds knowledge over months.
Master Brain
Source-of-truth knowledge graph and vector store
287K+ vectors in Supabase, 294K nodes in Neo4j
Hindsight
Per-agent rolling experience and conclusion memory
Self-hosted at home.arpa, fan-out reads from every agent
Mem0 + Honcho
Conversational fact extraction and user profile
Mirrored writes, fan-out reads
Pieces OS
Live workstream and browser-context stream
Rolling 30-day window, durable backups
Workspace files
Project-local CLAUDE.md and MEMORY.md notes
Read on every session start, written on every conclusion
Auto-memory
User preferences, feedback, and project state
Persists across every conversation, every machine
Master Brain
The source-of-truth knowledge graph and vector store. Connect wikis, databases, project tools, and uploaded docs. Agents retrieve grounded context on every call.
Orchestrated Fleet
Composer-conductor-leads model. A composer agent scopes the work, a conductor sequences it, specialist leads execute under their domain, every handoff is audit-logged.
Campaign Planner
Creates clear, ready-to-use campaign briefs using product info.
Ready made solutions
Cross-Session Memory
Every preference, decision, and project state survives the session. Pick up where you left off, six months later.
Channel Fabric
Discord, Telegram, Slack, web, REST, voice. One Fleet configuration, every channel a user actually lives in.
Human-in-the-Loop
Approval gates on high-impact actions. Real-time observability. Pause, correct, or replay any agent run.
From pilot to production
in three deliberate steps.
The Fleet does not promise magic. It promises a deterministic rollout. Deploy, configure, scale. Weeks, not quarters.
Deploy
Install Merlino Fleet on your infrastructure. Cloud, on-premise, or air-gapped. Bring your own model keys or run open-source models on your GPUs.
Configure
Wire up your knowledge sources, set department-level RBAC, define your agent roster. The Fleet starts learning your org from day one.
Scale
Roll out agents department by department with full audit. The longer the Fleet runs, the more institutional memory it accumulates.
Immutable Audit Trail
Every agent invocation, memory read, and tool call is logged with the actor, the action, the resource, and the result. Tamper-evident, SIEM-exportable.
Recent Activity
Personalized Email
15s
Personalized Email sent to ••••• @gmail.com
Peer Review
FAILED
Reviewed and approved 2 outputs from Content Drafting Agent
Content Drafting
PROCESSING
Generated draft campaign brief
Admin Approval
PROCESSING
Final approval of marketing copy before publishing
Weekly Campaign Report
2m
Generated campaign performance report
SEO Audit
FAILED
Reviewed and approved 2 outputs from Content Drafting Agent
Price Monitoring Agent
PROCESSING
Generated draft campaign brief
Department-Level RBAC
Permissions scoped to organization, department, role, and resource. SSO via SAML 2.0, OIDC, or LDAP. Every grant is audit-logged.
Sovereign Deployment
Cloud, on-premise, hybrid, or fully air-gapped. AES-256-GCM at rest, TLS 1.3 in transit. BYOK on enterprise. Your data never leaves your perimeter unless you say so.
Approval Gates
Brand, tone, and policy checks plus human review queues on high-impact actions. Configurable per agent, per channel, per department.
Brand & Style
Tone Guidelines
Link CRMs, helpdesks, and APIs to give agents secure, role-based access.
Salesforce
Hubspot
Sheets
Annual licenses. No usage-based surprises. No vendor lock-in.
Transparent pricing.
Memory included.
Start small to validate the memory stack on one team, then roll the Fleet out department by department.
Starter
Single-agent Fleet deployment under your own brand.
Start trial1 agent, single channel
Master Brain knowledge vault
Cross-session memory
Email support
Business
Multi-agent Fleet across departments, SSO, RBAC, cost control.
Start with BusinessUp to 5 agents
SSO via SAML or OIDC
Standard RBAC
Full memory stack (Master Brain, Hindsight, Mem0, Honcho)
LLM cost control and quotas
Priority email support
Enterprise
Production HA, advanced matrix RBAC, on-prem data connectors.
Talk to salesUnlimited agents
Advanced matrix RBAC
HA / scalable architecture
On-prem and hybrid data connectors
Monitoring and query audit
Dedicated support engineer
Custom
Independent source-code deployment, air-gapped, sovereign data.
Contact usAir-gapped or sovereign deployment
BYOK / HSM root key
Custom rules and modules
Tailored SLA
On-site or virtual training