Episode 517: Flip the Script: Start with Your Business Problem, Not the AI Tool
Dynamics Corner podcast with Matt Shrubsole (Red Hawk Technologies) on starting AI adoption from the business problem, not the tool. It is general strategy and covers no specific Business Central features. Topics: a downloadable AI decision matrix, mapping opportunity to cash, data privacy checks when choosing a tool, AI policy and shadow IT, agent guardrails and token costs, fractional CIOs, and value-based pricing for professional services.
- Introduction and context
- Welcome to Dynamics Corner and guest introduction
- Matt Shrubsole background and Red Hawk Technologies
- AI as a tool and personal assistant
- Real-world AI applications and limitations
- Starting the AI journey - finding the right questions
- Flipping the script - business-first approach
- Start with business problems and AI decision matrix
- Mapping opportunity to cash and identifying inefficiencies
- Using AI as research assistant and choosing AI tools
- AI tool selection and data privacy validation
- AI policy governance and shadow IT risks
- Cost implications and prerequisites before agent implementation
- Finding reliable help and fractional expertise
- Scaling AI oversight and getting started with conferences and peers
- Agent reality and setting guardrails
- Cost of running agents and influencer hype
- Client expectations and automation success stories
- Business readiness and value-based pricing impacts
- AI impact on professional services billing models
- AI agents for CVE remediation and automated documentation
- Closing remarks and guest contact information
- Episode conclusion
Overview
Dynamics Corner episode 517 features Matt Shrubsole of Red Hawk Technologies. He argues that AI planning should look like normal business strategic planning: identify costly workflows, compliance failures and transaction-heavy processes first, then decide where AI, an agentic workflow or an integration helps. Mapping the opportunity to cash flow and using a notebook tool to query interview and meeting data are given as ways to find inefficiencies.
The discussion then covers practical risks and effects. These include checking that a tool does not train on your data, setting an AI policy with an owner and pre-onboarding tests, shadow IT, token costs without limits, and clean data as a prerequisite for agents. It closes with agent guardrails, fractional CIOs, examples of CVE remediation and automated documentation, and how AI efficiency pushes professional services toward value-based pricing.
Key points
- Start with the business problem. Identify the costly workflows and operations first, then decide whether AI, an agentic workflow or an integration fits.
- Map the opportunity to cash flow, from lead hand-off to revenue, to find inefficiencies and ways to reduce time to cash. Companies structure this differently.
- Before adopting an AI tool, read its privacy policy and terms to confirm it does not train on your data, and use a business account rather than a personal one.
- AI governance: define a policy for each adopted tool, assign an owner, train people, and test their knowledge before they are onboarded to the tool.
- Set token or cost limits. One example given was $40,000 spent on tokens in a day with no limit in place.
- Agents need clean, accessible data and explicit guardrails. Many multi-agent setups are single-threaded processes with several instruction files, not true parallel agents.
- Professional services firms face pressure to move from labor-based to value-based pricing as AI cuts task time. This requires retooling sales processes and contracts.
Features
| Feature | Status | At |
|---|---|---|
| AI decision matrix | status not stated | 12:59 |
| Business-first AI approach | status not stated | 10:56 |
| Opportunity to cash workflow | status not stated | 13:35 |
| AI as personal assistant | status not stated, demoed | 6:42 |
| Agentic workflow optimization | status not stated | 15:25 |
| AI notebook as research assistant | status not stated | 18:00 |
| AI tool selection guidance for non-technical users | status not stated | 22:31 |
| Data privacy validation for AI tool selection | status not stated | 23:13 |
| AI governance and policy framework | status not stated | 25:09 |
| AI cost monitoring and token limits | status not stated | 27:58 |
| AI agents with guardrails and boundary setting | status not stated | 39:57 |
| Data quality prerequisites for AI agents | status not stated | 30:24 |
| Fractional CIO model for AI governance | status not stated | 35:57 |
| AI-driven quality assurance and testing | status not stated | 45:11 |
| Value-based pricing model adoption | status not stated | 51:09 |
| Agentic workflows for CVE remediation | status not stated | 56:27 |
| Automated technical documentation generation | status not stated | 56:56 |
| AI documentation for AI agent maintenance | status not stated | 58:29 |
Quotes
- 10:56 “I think that it’s very much along the lines of traditional business strategic planning. If for example, take AI completely out of the conversation.”
- 12:21 “i would say flip the script look at your business holistically first and think about”
- 12:47 “But really what you’re doing is you’re talking about the business challenge first to find the opportunities.”
- 12:59 “we actually created an AI decision matrix. It’s really pretty simple. And you know, people can download it on our website”
- 13:35 “Because, you know, you if you’re talking about opportunity to cash, right? Every business is opportunity to cash.”
Presenters (as heard): Chris, Brad, Matt Shrubsole, Matt, Matt Triplett, Max, Brett.