Welcome

AI Now Next

5 min

Recommendation: choose one AI-assisted workflow to improve.

  • You already have basic AI exposure.
  • Main deck stays PM-friendly; deeper layers support Developers.
  • Focus on judgment and workflows, not tool tourism.

Website mode

How to use the companion site

The main slide is the live track. The right rail gives PM takeaway, developer extension and risk check for each block.

AI now and next

What changed in AI by June 2026

10 min

Recommendation: track workflow consequences, not model announcements.

  • Ask what teams can safely delegate now.
  • Watch where human judgment becomes more important.
  • Source: BCG, AI at Work: Why Strategy Matters More Than Tools.

Trend filter

Signal over noise

A trend matters when it changes a recurring task, lowers coordination cost, improves quality control or shifts client expectations.

Prompting, verification, responsibility

Using AI well

8 min

Recommendation: treat prompting as work design with a review loop.

  • Define the job, assumptions and stop rule before asking.
  • Verify outputs against factual, client and code constraints.
  • Sources: MIT Sloan on “AI gravity”; HR Dive on AI overdependence.

Developer extension

Prompt pattern

Role, task, context, constraints, output format, verification criteria and stop rule. Short is fine when the contract is clear.

Agentic workflows

AI agents in practice

12 min

Recommendation: use agents for bounded, checkable workflows.

  • Give tools, limits and a visible success test.
  • Start with reversible work before client-critical delivery.
  • Good fits: research, tested code changes, document transformation.

Risk layer

Where agents break

They drift when context is stale, success criteria are implicit, tools are too broad or no one checks the final artifact.

More examples

Agent examples

Good examples: research synthesis, tested code changes, document transformation and repeated admin workflows.

Consulting industry

Real consulting use cases

13 min

Recommendation: choose repeated friction before impressive demos.

  • Require owner, input, output and review path.
  • Start with briefs, stakeholder drafts and testable plans.
  • Source: BCG on strategy and work redesign over tools.

Client risk check

Use-case test

A good AI use case has named owner, bounded input, expected output, review path and measurable benefit.

Market awareness

Stay informed without chasing every trend

7 min

Recommendation: build a weekly signal routine instead of chasing every trend.

  • Scan one capability shift and one delivery implication.
  • Test one small workflow before changing team practice.
  • Ask: what should we try, stop or standardize?

Next steps

Wrap-up Q&A

30 min

Recommendation: make the next step practical, local and reviewable.

  • Choose one repeated work pattern.
  • Add AI only where the review path is clear.
  • Keep the human judgment loop explicit.