Agent Live
A real-time assistant for sales and support calls.
Explore Agent Live Request access
Valnora & Elric builds AI products, and the engineering to put them to work.
Agent Serve
The AI workspace where AI agents do real work
Agent Serve is where teams build, deploy and manage AI agents, with approvals and audit built in. Want it built for you? Elric Works ships it as a measured outcome.
Agent Serve, built by Valnora & Elric, is the AI workspace where teams build, deploy, and manage AI agents — visually, conversationally, or with code. Teams ask questions in Chat, build agents in the workflow builder, analyze data in dashboards, and assign work to Relays, AI coworkers that ask for approval before they act. Access is by invitation.
IntroducingRelays, AI coworkers for your teamMeet Relays
The Valnora & Elric family
One company, five AI products.
Agent Serve, Agent Live and Agent Inbound & Outbound are available today by invitation. Agent CRM and Agent Infinity are upcoming, on the same foundation.
- Agent ServeThe AI workspace where teams build, deploy and manage AI agents.Explore Agent Serve
Agent Inbound & Outbound
AI voice that answers and makes calls, then hands warm leads to your team.
Explore Agent Inbound & Outbound Request accessUpcoming Agent CRM
An AI-native CRM that keeps itself up to date.
What we are building Join the waitlistUpcoming Agent Infinity
Docs, whiteboards and databases in one workspace, with Relays inside.
What we are building Join the waitlist
How to get Agent Serve
Available now, by invitation.
Agent Serve is for teams that need AI agents they can govern, audit and deploy where their data lives.
Request accessBy invitation
Request access and we set up your organization with you.
Our EU cloud or yours
Use our cloud in the EU, or self-host in your own cloud or VPC.
Your identity provider
Sign in with SAML or OIDC single sign-on and provision with SCIM.
Zero data retention
Run models under your providers’ zero data retention terms, or self-hosted.
Who it’s forEnterprise and financial-services teamsIT and operationsRevenue and sales operationsCompliance and risk
The workspace
One workspace for agents and their reviewers.
Agent Serve is the AI workspace where teams build, deploy, and manage AI agents — visually, conversationally, or with code. Every module shares the same data, permissions and history.
Chat
Ask a question about your business in plain language. Agent Serve answers from your connected data, files and knowledge, and can turn the answer into a chart, a report or a workflow.
Relays
AI coworkers with a role, a toolset and permissions. Assign a Relay a task, and it asks for approval before anything leaves the workspace.
Workflows
The visual builder. Connect blocks, models and integrations into agent logic, test it on real inputs, then deploy it on a schedule, a webhook or an API.
Dashboards
Interactive dashboards on governed metrics, with filters, drill-downs, simulators and certified versions your team can trust.
Relays
AI coworkers that ask before they act.
Relays are AI coworkers in Agent Serve. They take on real work — briefings, reviews, triage — inside the permissions you set, and pause for a person whenever an action needs approval.
Give recurring work to a Relay
A Relay is an AI coworker in Agent Serve with a specialization, written instructions and the resources it may use. Start from a role such as Revenue Analyst, Compliance Reviewer, Support Triage or Data Steward, or write your own.
- Assign one-off tasks or ongoing responsibilities on a schedule
- Each Relay has its own conversation, tasks, knowledge and activity history
- Relays can hand work to other Relays you allow
Decide what a Relay may do on its own
Choose an autonomy level — Observe, Assist or Operate — then set every action to allowed, needs approval or blocked. Reading data and drafting can run freely while sending email or saving to the Library waits for a person.
- Limits on tasks at once and tasks per day
- Approvals expire after the time you set
- Choose who can approve: the owner, stewards or admins
Review the work before it leaves
When a Relay needs approval it stops and shows exactly what it wants to do — the email, the calendar change, the file it will save. Approve it or decline it, and the decision is recorded on the task.
Build
Build agents in conversation or step by step.
Start in Chat when you know what you want. Open the workflow builder when you need every step visible. Both run on the same workspace data and deploy the same way.
Chat
Build and manage everything in natural language. Ask Agent Serve to analyze a table, draft a report or set up a workflow, and keep working from the result.
About Chat
Workflow builder
Lay out agent logic step by step: triggers, model calls, integrations, branches and failure paths. Run it, read the trace, then deploy.
About workflows
Dashboards & BI
Understand the business before you automate it.
Agent Serve includes a full BI layer: data sources, a semantic model, governed metrics and interactive dashboards. Agents and people read the same numbers.
Explore dashboardsThe whole business on one screen
Agent Serve dashboards sit on governed metrics, so revenue, margin, pipeline and service levels mean the same thing in every view. Filter by period, currency or region, and certify the dashboards your leadership relies on.
Pipeline and forecast, by stage and owner
Follow open and weighted pipeline through each stage, see what is expected to close each month, and switch between pipeline, forecast and team views without leaving the dashboard.
Ask “what if” with a live model
Simulator dashboards turn assumptions into outcomes. Change the deal size, discount or term and Agent Serve recalculates margin, NPV and payback, compares scenarios and finds the value that meets a target.
Case study · Aestlo
From 3.2x to 6.5x return on ad spend, in one month.
Aestlo, a Valnora & Elric fashion brand, makes its campaign films, UGC-style ads and branded stills in Agent Serve, and runs ad reporting and operations in the same workspace.
Read the case study
Return on ad spend as reported in Aestlo’s Meta Ads account; results vary by brand.
Use cases
Many modules, one outcome.
Each use case combines the parts of Agent Serve a team needs. Figures are examples of what to measure.
All use casesIllustrative example. Figures show how the workflow is measured; they are not customer results.
Elric Works · engineering by Valnora & Elric
Rather have it built for you? We sell the outcome.
Elric Works is the engineering arm of Valnora & Elric. It has built process intelligence for pharma plants, AR for field engineers and a freight operating system, and it sells each engagement as a measured result.
Explore Elric WorksOutcome Sprint
One business number and a date, agreed in writing. Live by the deadline or your money back.
For founders
A founding build or a dedicated pod that ships your product, then hands it to your team.
For larger companies
Forward deployed engineers who put AI into production inside your systems.
How it works
From data to finished work, people in charge.
- 01
Connect
Link your databases, warehouses, files and business tools. Agent Serve keeps the credentials and decides who can use each connection.
- 02
Define
Describe how the business works: tables, metrics, the semantic model, rules and playbooks. Agents reuse the same definitions your dashboards do.
- 03
Delegate
Hand work to Relays and deployed workflows — on demand, on a schedule or from a trigger such as a webhook.
- 04
Review
Approve the actions that need a person, read the run history for everything else, and refine instructions as you learn.
Security
SOC 2 and GDPR ready by default.
Agent Serve ships with the controls security reviews start with: single sign-on, self-hosting in your own cloud, and zero data retention with supported model providers.
Security and complianceSign in through your identity provider
Agent Serve supports SAML and OIDC single sign-on, with SCIM 2.0 to provision people and groups automatically.
- SAML 2.0 and OIDC, including Okta and Microsoft Entra
- SCIM users and groups, mapped to access
- Verified domains and invite-only sign-up
Run it in your own cloud
Deploy Agent Serve inside your own cloud account or VPC, next to your data, with the same product your team sees here.
- Helm chart for Kubernetes
- Docker images with a production compose file
- Reference Terraform for AWS, eu-central-1 by default
Zero data retention with supported model providers
Run models under your organization’s zero data retention terms, or keep inference inside your network, and decide how long Agent Serve keeps logs.
- Bring your own provider keys and agreements
- Self-hosted models through Ollama, vLLM or LiteLLM
- Configurable log retention and PII redaction
Ready means the controls are in place. Agent Serve does not claim a SOC 2 report or a certification.
Governance
Governance for every agent action.
Agent Serve puts controls where agents meet your business: who can see which data, what an agent may do alone, and a record of what happened.
Approvals before action
Relays pause for a named approver before sending, saving or changing anything you mark as needing approval. Decisions are kept on the task.
Autonomy you set per action
Observe, Assist or Operate, with every action allowed, gated or blocked — plus daily limits and approval expiry.
Row-level security
Policies decide which rows each person can see, and dashboards and tables honour them. Preview a dashboard as another role before you share it.
Audit trail
Workspace activity is written to an audit log your admins can search and export, and every agent run keeps a step-by-step trace.
Single sign-on and provisioning
Sign in through your identity provider with SSO, and provision people and groups with SCIM.
Versioned, certified dashboards
Every dashboard keeps its version history to compare side by side, and the ones your team should rely on can be marked certified.
Integrations
Connected to the tools your teams already use.
Agent Serve agents read from and act in your business systems — messaging, email, CRM, support, engineering, documents and databases — and run on the major model providers.
Browse integrationsRoadmap
What we are building next.
NowAvailable today
- Relays with approvalsAI coworkers with roles, permissions and approval before external actions.Learn more
- Dashboards and BIGoverned metrics, drill-through, scenarios and row-level security.Learn more
- Semantic model and rulesEntities, relationships, measures and reusable business logic.Learn more
- SSO, SCIM and self-hostingSAML and OIDC sign-in, provisioning, and deployment in your own cloud.Learn more
- Agent LiveA real-time assistant for sales and support calls, with an admin portal for managers.Learn more
- Agent Inbound & OutboundAI voice that answers and makes calls, updates CRMs and hands warm leads to your team.Learn more
NextWhat we are building next
FAQ
Questions teams ask first
What is Agent Serve?
Agent Serve is the AI workspace where teams build, deploy, and manage AI agents — visually, conversationally, or with code. It combines Chat, a workflow builder, dashboards and Relays, AI coworkers, on one set of governed data.
How do teams get access to Agent Serve?
Agent Serve accounts are created by invitation. Send a request with your team and use case, and the Valnora & Elric team will follow up to set up your workspace. Colleagues then join through invitations from your workspace admins.
Can a Relay act without a person approving it?
Only where you allow it. Each Relay in Agent Serve has an autonomy level and a rule for every action — allowed, needs approval or blocked. Actions that need approval wait for a named approver and expire if nobody responds.
Which tools and models does Agent Serve connect to?
Agent Serve connects to business tools such as Slack, Microsoft Teams, Gmail, Outlook, Salesforce, HubSpot, Zendesk, Jira, GitHub, Notion, Google Drive and databases such as Snowflake and PostgreSQL, and runs agents on the major model providers. The integrations page lists every connection.
How does Agent Serve control access to data?
Workspace roles decide who can use each resource, row-level security policies decide which rows each person sees in tables and dashboards, and an audit log records workspace activity. Organizations can sign in with SSO and provision people with SCIM.