Use-case recipes

Real jobs, built with almyty — each one names its stack (trigger · models · tools · mode) and walks the concrete steps. Pick the closest recipe, follow it, and swap the pieces for your own APIs and channels.

Every recipe is built from the same primitives — agents, tools, gateways, memory, and the runner. The “apps” are yours: any API you connect becomes tools, any npm package becomes tools, and channels (Slack, Discord, Telegram, webhooks) are the triggers and outputs. Tags below are almyty capabilities; a name like “CRM API” means an API you connect, not a built-in connector.

Support & Sales

• Triage support tickets and draft replies — a real running agent: looks up the order, drafts a reply, a cross-vendor panel checks it, refunds escalate to a human. verification · approvals · memory
• Deflect L1 questions from your docs — answers from your docs in-channel, cites the page, opens a ticket only when it can’t. Slack · Discord · constraints
• Enrich and route inbound leads — enriches, scores against your ICP, writes back to the CRM, pings the right rep. CRM API · Slack · constraints
• Turn call notes into CRM updates — extracts next steps, updates the opportunity, drafts a follow-up for approval. CRM API · approvals

Engineering & DevOps

• Triage and label GitHub issues — labels, links duplicates, routes, requests repros. GitHub API · constraints · memory
• Review pull requests for bugs and policy — flags bugs, missing tests, and policy violations inline; never approves. GitHub API · verification · constraints
• Summarize on-call alerts and suggest a runbook — gathers metrics and logs, correlates with past incidents, posts a summary. monitoring API · Slack · memory
• Orchestrate a multi-vendor coding fleet — one planner driving Claude Code, Codex, and aider on one repo. runner · coding agents

Data, Finance & Ops

• Ask your database questions in plain English — writes and runs read-only queries, answers with the SQL it used. database · Slack · constraints
• Post a nightly data-quality digest — checks freshness and anomalies on a schedule, reports only what’s wrong. scheduled · Slack · memory
• Reconcile Stripe payouts and flag anomalies — matches Stripe against the ledger daily, reports only exceptions. Stripe API · scheduled · constraints
• Cross-vendor contract and risk review — several models argue the risks, a judge writes the memo. debate · multi-model

Enterprise & Governance

• Run a governed refund agent in production — constraints, cross-vendor verify, human approval, full audit. verification · approvals · audit
• Govern AI spend across teams — per-team budgets with soft/hard caps and alerts, enforced at run start. budgets · cost governance
• Enforce approval policies org-wide — action + threshold → required approver, applied across every agent. approvals · policy
• Role-based access for a multi-team org — teams, per-resource roles, least-privilege by default. RBAC · multi-tenant
• Onboard your workforce with SSO and SCIM — SAML/OIDC sign-in, SCIM provisioning, groups → roles. SSO · SCIM
• Stream your audit trail to a SIEM — every action to Splunk/Datadog, searchable and alertable. audit · SIEM
• Ship EU AI Act-compliant agents — AI-disclosure on every channel + Annex IV technical-doc export. compliance · EU AI Act
• Put your Google Drive or SharePoint to work — connect Google/Microsoft 365, search and summarize docs, write back. Google Drive API · SharePoint
• Air-gapped, self-hosted agents — local models, data residency, nothing leaves your network. self-host · compliance

Platform & Integration

• Give Claude Code your own API — import a schema, publish an MCP gateway, call it from Claude. MCP · Claude Code
• Modernize a legacy SOAP service — wrap a WSDL as modern MCP tools, untouched. SOAP · MCP
• Turn an npm package into tools — point almyty at stripe, get typed tools. SDK tools
• Give an agent a machine to work on — expose any machine — laptop, build server, VPC box. runner · on-prem