Arabesque AI (AutoCIO)

AutoCIO is one of the few AI products built around the portfolio itself. Portfolio managers upload holdings, set benchmarks, constraints and sustainability rules, and receive optimised rebalances with risk metrics, attribution and per-stock explainability. AutoCIO 3.0 shipped in April 2026, alongside a listing in Google's Gemini Enterprise agent gallery, with API, Python SDK and MCP access. Unusually, list prices are published, from $25,000 a year.

The corporate footprint has narrowed and the evidence is mostly the group's own. Arabesque AI Ltd filed micro-entity accounts for 2025 after full accounts in prior years; its immediate parent received, and had discontinued, two strike-off notices in 2024; and no funding has been disclosed since DWS took 24.9% in 2020. The live record rests on group funds and a site example in which AI enhancement trailed the base portfolio over one and three years. We did not locate security certifications or a no-training commitment.

We initiate at 5.85, Emerging, with a Stable outlook. Product Capability (6.58) and Workflow Fit (6.50) lead and Commercial Value (6.00) benefits from price transparency; Enterprise Readiness (4.36) and Data & Content (5.50) are the gaps.

Strengths

• Portfolio-native design. Construction, optimisation, rebalancing and attribution in one tool for portfolio managers.
• Explainability. Daily per-stock view of the weight given to each model input, plus factor attribution.
• Holdings ingestion. Portfolios uploaded from CSV, Excel or PDF, with custom universes and client ESG data.
• Published prices. Three tiers from $25,000 to $450,000 a year.
• Assistant access. API, Python SDK and MCP, with a Gemini Enterprise listing (April 2026).
• Regulated heritage. Group asset management arms authorised by the FCA and licensed by MAS.

Cautions

• Corporate scale. Arabesque AI Ltd filed micro-entity accounts for 2025; headcount last disclosed as 30+ in 2024.
• Capital. No funding disclosed since 2020; DWS's current holding not confirmed.
• Evidence. Live results come from group funds and company examples, with mixed AI-enhanced results.
• Assurance. No SOC 2, ISO 27001 or no-training commitment located.
• Methodology. Point-in-time and backtest handling not documented.
• Integration. No OMS/EMS integrations named.
• Leadership. No chief executive of Arabesque AI named since 2023.