# Run enterprise AI agents in the company's own cloud. > The Twistag Agent Platform runs a company's AI agents, workflows and connectors inside its own cloud, with costs, approvals and audit records built in. One agent platform to build, run and approve agent work on the company's own systems. - AWS - Microsoft Azure - Google Cloud - On-premises ## Every AI agent the company runs, governed inside its own cloud. ### Agents Agents take on routine work across the company’s systems and hand every decision that matters to the right person. ### Workflows Long processes keep moving on their own and stop only where a person has to decide. ### Company brain Every answer comes from the company’s own documents and records, with the source attached and access kept by role. ### Evals Each new version proves itself on real cases before it reaches the business. ### Observability Leadership sees what every agent did, what it cost and how long it took. ### Governance The company decides what each agent may do and who signs off, and every decision is recorded. ### Live dashboards Each department gets a live view of its numbers on request, in the platform or in the AI assistant it already uses. ## The company's cloud, the company's keys, the company's record. The platform runs where the company already runs its systems. Data and models stay inside, every run is logged in the company's own store, and each use case carries its EU AI Act risk tier. Three ways to run enterprise AI agents Compared on Twistag Agent Platform Hyperscaler agent studio SaaS agent vendor Where it runs The company's own cloud account or data centre A managed service on one hyperscaler The vendor's cloud Keys and data Held by the company Held by the company, inside that hyperscaler's services Held by the vendor, under contract Model choice Any model that passes the company's evaluation Mostly that hyperscaler's catalogue The vendor's choice Audit trail Every run, kept in the company's own database Logs in the hyperscaler's tools Vendor logs, exported on request EU AI Act file Risk tier and oversight recorded per use case Built by the company's own team Varies by vendor Who installs it A Twistag forward deployed pod The company's team or an integrator The vendor's onboarding team Leaving Code, configuration and data stay with the company Tied to that hyperscaler's services Depends on the contract Every agent, workflow, policy and connector belongs to the company, with its data, tests and audit records. There is no hosted Twistag tenant to move away from. - AWS - Microsoft Azure - Google Cloud - On-premises ## Every run counted, every euro accounted for. Leadership sees how much work the agents do, what it costs and what needs attention, by agent and by department. ### Costs Spend by department and by agent, against the budget each team was given. ### Runs Every run the agents make, with what it did, how long it took and what it cost. ### Alerts Problems reach the person who owns the agent, early enough to act. ## What leadership asks first. Answers on where the platform runs, what it costs and who owns what. What is an enterprise AI agent platform? Software that runs a company's AI agents in one governed place. It connects agents to systems such as ERP, CRM and ticketing, sends decisions to named approvers, measures cost and quality, and keeps a record of every run. The Twistag Agent Platform does this inside the company's own cloud or data centre. Does it run in the company's own cloud? Yes. The platform installs in the company's own AWS, Azure or Google Cloud account, or in its data centre. Agents, workflows, logs and documents stay inside that boundary, under the company's keys and single sign-on. Models run through the company's existing cloud contracts or on its own hardware. How is it different from Copilot Studio or Agentforce? Copilot Studio and Agentforce are built around their vendors' own products and clouds. The Twistag Agent Platform runs in the company's own account, connects to SAP, Salesforce, Zendesk and SharePoint alike, and works with any model that passes the company's evaluation. Code, configuration and the full audit trail stay with the company. Does it help with the EU AI Act? It keeps the evidence. Each use case carries a recorded risk tier, a named person responsible for oversight and a log of every run and approval, which is the material the EU AI Act asks deployers to keep. Twistag maps each use case during the install. The final classification stays with the company's own counsel. What does it cost? The AI Readiness Sprint starts from €5,000. A department install is a fixed-scope project of about twelve weeks. From the second department, the platform runs on a yearly licence, priced per department for mid-market companies and per tier for enterprise and portfolio groups. Model usage is billed through the company's own cloud account. Who operates it after the install? The company's own agent operators, trained by the Twistag pod during the install. They watch runs, review exceptions, adjust policies and approve new versions once they pass evaluation. Twistag stays on call under a support agreement and brings the pod back for the next department. Can a consultancy or agency deploy it under its own brand? Yes. Consultancies, systems integrators and agencies can deploy the platform for their clients under their own name, colours and domain. Twistag supplies the platform, the install playbooks and engineering support, and stays in the background unless the partner chooses otherwise. Who owns the agents and the data? The company. Agents, workflows, prompts, evaluation sets, configuration and data belong to the company and live in its own account. If the company stops working with Twistag, all of it stays where it is, and the team keeps full access to the code. ## See the platform running on the company's own work. A thirty-minute demo on one workflow, in the company's cloud, with the people who would run it. --- Source: https://twistag.com/platform