Blog

Directing the symphony of intelligence – Orchestrating agentic AI part 7

Per Ohlqvist Head of Enterprise Design & Security Protection Officer, Solita

Published 29 Dec 2025

Reading time 2 min

Our agentic AI blog series explores what it means to move from deploying AI systems to orchestrating them. Read the last part of the series. Part 7: The final crescendo – The conductor’s playbook (and the orchestra within).

We stand at a moment where every organisation can become a symphony of intelligence if we learn to conduct. Our instruments are ready: agentic AI, knowledge graphs, EA 4.0. The score is drafted: guardrails, ontologies, culture. And importantly, this isn’t a hypothetical score, and it’s already being played in the real world. 

Walmart’s integrated AI ecosystem coordinates countless processes and even compresses product cycle times from months to weeks. Intuit’s new QuickBooks AI agents have 78% of users saying the software makes it easier to run their business, and 68% saying it frees them to spend more time on growth. Johnson & Johnson, meanwhile, reports that an AI-assisted surgery planning tool reduced a week-long procedure planning to just days. Sector by sector, AI orchestration is already happening, and it’s working.

Here is a simplified playbook for the new conductor, which is you:

  1. Start knowledge-first. Don’t just collect data; model your core knowledge. First step: Model your top 20 core business concepts and the relationships between them.
  2. Embed guardrails early. Build the rules of the road before the traffic arrives. First step: Review Sewak’s six guardrails and Löwgren’s seven laws, and ask which one poses the biggest risk to your business today.
  3. Simulate before you scale. Treat your architecture as a living lab, not a static blueprint. First step: Identify one process where you can stress-test multi-agent behaviour in a safe environment (a sandbox or pilot project).
  4. Keep humans in the loop. Don’t automate the conductor. Use humans to frame, guide, and validate key decisions. First step: Map a key workflow and pinpoint the three most critical moments where human judgment should intervene.
  5. Measure the culture. You can’t have a symphony without trust and psychological safety. First step: Ask your team, “On a scale of 1–10, how safe do you feel experimenting and failing with new tech?” and listen carefully to the answers.

So I’ll end with the question I ask my clients: Which guardrail, semantic backbone, or human-in-the-loop practice is still missing at your company, and what’s your first step to fix it?

Read the full blog series by Per Ohlqvist: Directing the symphony of intelligence – Orchestrating agentic AI.

  1. Tech