Deliverance AI and Hewlett Packard Enterprise (HPE) have validated an architecture for deploying AI agents in private-Cloud environments, aiming at organisations that need to automate business processes while retaining control over sensitive data.
The solution combines Deliverance Agentic OS with HPE Private Cloud AI, part of the HPE AI Factory with NVIDIA portfolio. It is designed for regulated and business-critical applications, with support for on-premises, private-Cloud and air-gapped deployments.
The validation covers model selection, multi-agent orchestration, human oversight, quality assessment and audit trails. The aim is to allow AI agents to handle complex workflows without giving them unrestricted access to enterprise systems or information.
Deliverance AI said the architecture supports the use of approved models, including NVIDIA NIM and self-hosted options. Its SmartRouter selects models according to task complexity, cost, performance and governance requirements, allowing organisations to balance processing costs against the capabilities required for individual tasks.
Controls include permissions for individual agents, checks on inputs and outputs, protection of personally identifiable information and a global kill switch. Work that fails predefined quality checks can be rerun or referred to a designated human reviewer, while actions and supporting sources are recorded for audit.
The platform can also retrieve information from approved enterprise document collections without moving sensitive data outside the customer’s controlled environment.
Deliverance AI cited results from existing customer deployments:
At a global systems integrator, its agents reportedly complete 80% of an initial request for proposal (RFP) response before human review, saving the bid team around 30 hours a week. Responses are based on previous submissions, with uncertain answers referred to named human specialists.
At a tier-one mobile operator, agents analyse network logs and telemetry to produce a root-cause assessment and draft an engineering report. Deliverance AI said the deployment had reduced the time taken to reach a finding from two to three hours to less than five minutes.
And a clinical-stage biotech company is using the platform to draft sections of regulatory documents from an agreed clinical trial dataset. The company expects to complete around 200 documents by year-end, with claims linked to source evidence and unsupported assertions flagged for human review.
Mick McNeil, Co-Founder and chief executive of Deliverance AI, said enterprises needed agents capable of completing useful work while maintaining the controls expected of production systems.
Robin Braun, Vice-President of AI business development at HPE, said the partnership was intended to help customers move beyond pilot projects to applications in areas such as investigations, RFP responses and compliance reviews.
There’s plenty of other editorial on our sister site, Electronic Specifier! Or you can always join in the conversation by visiting our LinkedIn page.
