Merlewood Row

AI agents and the customer service they promise. Reviews and practical guides for teams in FCA-regulated UK firms.
Last updated:

Sources

Numbered references distinguish public sources from information supplied by a company. Follow the cited source and its check date when quoting a claim.

  1. A Nexus engineer builds the first procedure with the team and shows it running in its own tools. agent.nexus, checked 8 November 2026.
  2. Blog: the agent does not ship unless a CIO can sign off on data flow, access model and audit trail. agent.nexus/blog/nexus-raises-4-3m-seed-shipping-cue, checked 8 November 2026.
  3. Max is described as: "Max works the way your firm works, researching and analyzing any deal to produce client-ready spreadsheets, slides, and reports.". hebbia.com/product, checked 8 November 2026.
  4. Max runs analysis over documents with traceability: "Reason over any volume of documents or companies with precise analysis and complete traceability back to every finding.". hebbia.com/product, checked 8 November 2026.
  5. Integrations page: "Bring market feeds and your firm's internal data into your Hebbia workflow.". hebbia.com/integrations, checked 8 November 2026.
  6. Admins set data policies in the Power Platform admin center.. learn.microsoft.com/en-us/microsoft-copilot-studio/security-and-governance, checked 8 November 2026.
  7. Admins can turn off agent publishing for the tenant.. learn.microsoft.com/en-us/microsoft-copilot-studio/security-and-governance, checked 8 November 2026.
  8. Release pipelines and approval gates support Dev-Test-Prod promotion.. microsoft.com/en-us/copilot/products/copilot-studio, checked 8 November 2026.
  9. Every agent deployed has a measurable business KPI from day one. agent.nexus/blog/nexus-raises-4-3m-seed-shipping-cue, checked 8 November 2026.
  10. Comparison page: agents escalate with full context when uncertain. agent.nexus/compare/enterprise-ai-platforms/nexus-vs-glean, checked 8 November 2026.
  11. Same comparison row: full audit trails, decision traceability, role-based access. agent.nexus/compare/enterprise-ai-platforms/nexus-vs-glean, checked 8 November 2026.
  12. Information supplied by Nexus; public-page verification unavailable. recorded 8 November 2026.
  13. Audit visibility: "See exactly who accessed your data, when they accessed it, and what they did". hebbia.com/security, checked 8 November 2026.
  14. No direct data access by Hebbia staff: "Hebbia doesn't have direct access to your data. Production access is restricted through just-in-time access management.". hebbia.com/security, checked 8 November 2026.
  15. Agents can be created by describing them in plain language, then tested before publishing.. learn.microsoft.com/en-us/microsoft-copilot-studio/fundamentals-what-is-copilot-studio, checked 8 November 2026.
  16. Monitoring, evaluations and administration are listed as post-build tools.. learn.microsoft.com/en-us/microsoft-copilot-studio/fundamentals-what-is-copilot-studio, checked 8 November 2026.
  17. Customer Conversation harness can escalate to a human representative.. learn.microsoft.com/en-us/microsoft-copilot-studio/fundamentals-what-is-copilot-studio, checked 8 November 2026.
  18. Admins have full visibility into maker audit logs in Microsoft Purview.. learn.microsoft.com/en-us/microsoft-copilot-studio/security-and-governance, checked 8 November 2026.
  19. Admins can receive alerts on agent activity through Microsoft Sentinel.. learn.microsoft.com/en-us/microsoft-copilot-studio/security-and-governance, checked 8 November 2026.
  20. Agent-user authentication limits knowledge answers to content that user can access.. learn.microsoft.com/en-us/microsoft-copilot-studio/knowledge-copilot-studio, checked 8 November 2026.
  21. Enterprise editions are administered with IAM, Workforce Identity Federation and source ACLs.. docs.cloud.google.com/gemini/enterprise/docs/editions, checked 8 November 2026.
  22. Gemini Enterprise app and Gemini Enterprise Agent Platform must be bought separately.. cloud.google.com/gemini-enterprise/faq, checked 8 November 2026.
  23. Each agent gets a unique identity.. docs.cloud.google.com/gemini-enterprise-agent-platform/govern, checked 8 November 2026.
  24. Agent tool-use rules are written in plain language.. docs.cloud.google.com/gemini-enterprise-agent-platform/govern, checked 8 November 2026.
  25. Enterprise editions support Access Transparency and Cloud Audit Logs.. docs.cloud.google.com/gemini/enterprise/docs/editions, checked 8 November 2026.
  26. Employees can create their own custom agents with a no-code workbench.. cloud.google.com/gemini-enterprise/faq, checked 8 November 2026.
  27. Source-system permissions enforced on every read and write. glean.com/platform/security, checked 8 November 2026.
  28. Human approval before agent actions execute. glean.com/platform/connectors, checked 8 November 2026.
  29. Admin-managed agent credentials. docs.glean.com, checked 8 November 2026.
  30. Always-on audit trail, events can stream to SIEM. glean.com/platform/security, checked 8 November 2026.
  31. Agent governance policies. glean.com/platform/security, checked 8 November 2026.
  32. Agent based or per-user policies. glean.com/platform/security, checked 8 November 2026.
  33. WRITER is positioned as an AI platform for marketing and revenue teams.. writer.com/company/about, checked 8 November 2026.
  34. Enterprise Brain is described as the universal context layer that grounds agents.. writer.com, checked 8 November 2026.
  35. The vendor says it provides an embedded team of transformation leads and deployment engineers.. writer.com, checked 8 November 2026.
  36. Guardrails used with the API, SDK and Agent Builder apply only to external provider models for now.. dev.writer.com/home/guardrails, checked 8 November 2026.
  37. Event log observability is not yet available for Agent Builder v2 agents.. dev.writer.com/home/event-logs, checked 8 November 2026.
  38. Guardrails, identity validation, granular permissions and approval flows shape how agents interact with connected systems.. writer.com/trust, checked 8 November 2026.
  39. Activity traces, source citation and decision logs are offered to explain answers.. writer.com/trust, checked 8 November 2026.
  40. Role-based permissions and least-privilege access are offered.. writer.com/trust, checked 8 November 2026.
  41. Policies are enforced outside the LLM (vendor claim). kore.ai/ai-agent-platform, checked 8 November 2026.
  42. StateTrace captures the full execution path across handoffs, state, tool calls and context (vendor claim). kore.ai, checked 8 November 2026.
  43. Forward Deployed Engineers are part of the offer (vendor claim). kore.ai, checked 8 November 2026.
  44. Trust Center lists an on-premises deployment (vendor claim). trust.kore.ai, checked 8 November 2026.
  45. Trust Center lists a private cloud on a dedicated VPC, single tenant (vendor claim). trust.kore.ai, checked 8 November 2026.
  46. Autoloop is the optimization engine that improves agents against goals from first build through production (vendor claim). kore.ai, checked 8 November 2026.
  47. Each agent run is traced across reasoning, tools, guardrails, handoffs and outcomes (vendor claim). kore.ai/ai-agent-platform, checked 8 November 2026.
  48. Search returns permissioned answers from every system the customer runs (vendor claim). kore.ai/employee-experience, checked 8 November 2026.
  49. Homepage offers a proof on the customer's own use case before commitment (vendor claim). kore.ai, checked 8 November 2026.