For operations leaders and workflow owners
Design AI around the work as it actually happens.
Demonstrations usually show the clean path. Operations has to manage every case. I help teams redesign workflows so AI can reduce total work, keep exceptions visible, and give people clear roles, controls, and fallbacks.
Designing an AI-enabled workflow
A workflow is redesigned in stages. Each one depends on what the previous stage found.
Understand the current workflow
Map the real process, including informal workarounds, handoffs, delays, judgment calls, and existing measures.
Find the right point for AI
Identify where AI could improve an outcome, and where conventional automation or no technology change may be more appropriate.
Design human and AI roles
Define what AI proposes or performs, what people review or decide, and who remains accountable for the outcome.
Plan for exceptions and failure
Decide where uncertain cases go, how work is escalated, and what happens when the AI or its provider is unavailable.
Measure the total effect
Evaluate quality, time, cost, workload, service, risk, and adoption across the whole workflow, not only the speed of one task.
Where does the work go when AI cannot handle the case?
Standard cases may account for most of the volume, but exceptions are where workflows become expensive and dependent on experience. If AI handles the easy cases, does it remove work, or send a smaller but much harder queue back to the team?
A successful workflow does not hide exceptions. It makes them visible, routes them deliberately, and gives the team a fallback it can operate.
What I help operations teams put in place
- A clear view of the current workflow
- The baseline, handoffs, exceptions, bottlenecks, ownership, and work that existing documentation may not capture.
- A practical operating design
- Defined roles for people and AI, decision points, review requirements, exception routes, controls, and fallback.
- A path to implementation and adoption
- A bounded first use, requirements, measures, team involvement, training needs, and a way to learn before scaling.
Faster is not always less work
AI may complete one task quickly while creating new checking, correction, escalation, and support work elsewhere.
The relevant measure is whether the complete workflow performs better, not whether one step looks impressive in a demonstration.
Ways to work together
Engagements can start with one workflow workshop or extend into ongoing advisory and hands-on workflow design and delivery.
Workflow workshops
Bring one real workflow. We examine how it operates today, where AI may help, what it would change, and the smallest sensible next step.
Operational advisory
Ongoing support for operations leaders as they prioritize opportunities, redesign workflows, prepare pilots, manage adoption, and evaluate results.
Workflow design and delivery
Hands-on support to define requirements, human and AI responsibilities, exception handling, controls, measures, and implementation with technology teams.
Prefer email? Write to hossein@initialblock.com.
Relevant experience
- More than fifteen years building technology
- Nine years building operational infrastructure for financial firms
- Experience translating real workflows, regulatory needs, and business requirements into working systems
- MSc in Artificial Intelligence from the University of Amsterdam
I have had to stay with systems as organizations grow and workflows change. That is why I examine not only what an AI system can do, but what the team will have to review, correct, support, and operate around it.
Book a free call
Bring one workflow that may be ready for change.
In a free 15-minute call, we can discuss how it works today and whether a workshop, advisory engagement, or hands-on support would be useful.