Skip to content
SUNYA
Sunya / Energy AI research

Repsol / Multi-agent workflow platform for employee productivity

AI-031 · Source-linked project profile. Read the scope, reported result and what remains unknown.

Open the interactive projectHow to read this evidence

What the source says

Concrete supervised agent workflows with measured cohort/scale; operational productivity outcomes remain unquantified.

Company
Repsol
Department
IT, data & enterprise transformation
Record type
company_adoption
Disclosed stage
Pilot. Repsol’s 2026 report describes a multi-agent employee-workflow platform piloted from 2025 for over four months and updated to 34 agents across three workflows. This remains explicitly pilot-stage despite the breadth of agents/workflows.
Source date
2026-02-19
Source-date precision
day
Source provenance
B · Primary publication
Review disposition
Access or verification limitation; inspect profile and source history

Reported result

Agents in multi-agent workflows: 34 across 3 workflows (Company-reported)

What this does not establish

Agent counts vary by disclosure date; not a contradiction. Annual report window within research range.

Analyst inference: Concrete supervised agent workflows with measured cohort/scale; operational productivity outcomes remain unquantified.

Original sources

Disclosure history

Accenture and NVIDIA to scale AI agents

Repsol expanded its Digital Program by deploying AI agent systems with Accenture's AI Refinery and the NVIDIA AI platform, targeting process optimization, planning, forecasting, maintenance and customer personalization. It also plans to explore NVIDIA Omniverse digital twins.

Multi-agent platform: 22 agents supervised by 50 staff

Repsol launched a multi-agent platform built with Accenture in which AI agents self-organize and coordinate tasks within workflows. The pilot ran three teams with 22 agents in the 700-person Digitalization and Services department, supervised by 50+ employees for over four months.

Multi-agent platform

Piloting a platform that enables teams to construct coordinated AI agents.

The pilot demonstrates a defined organizational setting. Its 22 agents are distinct from the 22 scaled GenAI use cases in AI-032. No platform-specific financial return is disclosed.

A source grade describes provenance, not independent verification. This is a research profile, not a measure of net return. Report a correction.