GroundSpring

05/29/2026

We got really REAL about AI governance yesterday at SingleStone as I led a fast-paced session for leaders in highly-regulated organizations! And the discussion was fantastic. Here's one highlight:

Our conversation was cross-functional, which made it so rich. In one exchange, a CISO and an engineer candidly shared why security mandates can fall flat, even if they're important for compliance.

For the engineer working in defined Agile sprints, they w**d out anything not absolutely required to get the next working feature shipped. On the other hand, Security folks who are thinking longer-term about harding up their defenses find it hard to convince engineers to add in these 'extras.'

What I've seen that helps bridge the two perspectives to make governance work:

- Executive eadership needs to be aware of these internal cultural friction points and align incentives and priorities accordingly. When AI-related mandates are one-size-fits all, they're quickly discarded as irrelevant. This is why I nearly always start a governance engagement with stakeholder research.

- Cross-functional management should share their perspectives openly. You need a culture that nurtures this kind of communication, but understanding teams' respective workflows and KPIs is a big unlock to finding common ground.

- Demonstrate importance, don't just talk about it. Effective AI governance impacts real people's lives. Shortcomings can lead to true harms. This isn't new. Showcase how governance gaps are creating friction, cost, and distrust in organizations, in the news, and in use cases to bring the downsides to life for executive leadership. Often, it's hard to really understand the importance of something until it's put in concrete terms that hit home. I find that when doing a use case inventory, these important anecdotes crop up.

Big thanks to SingleStone, Alex Otanez, and our group of engaged attendees for bringing this critical content to leaders in regulated spaces. Thanks too to our attendees for making the time for terrific engagement (no small feat at 7:45am!!).

Let's keep these conversations going!

🛣️ I'm planning my next stop for my in-person REAL AI Governance workshop...details to come next week!! 🫢

05/18/2026

Why Ground in order to Spring?

As my new brand launch for 𝗚𝗿𝗼𝘂𝗻𝗱𝗦𝗽𝗿𝗶𝗻𝗴 is underway, I want to share an example that inspired the new framing:

One of my clients pronounced an ambitious mandate. 𝗧𝗵𝗲𝘆 𝘄𝗮𝗻𝘁𝗲𝗱 𝘁𝗵𝗲𝗶𝗿 𝗲𝗺𝗽𝗹𝗼𝘆𝗲𝗲𝘀 𝘁𝗼 𝗶𝗻𝗰𝗿𝗲𝗮𝘀𝗲 𝘁𝗵𝗲𝗶𝗿 𝗡𝗲𝘁 𝗣𝗿𝗼𝗺𝗼𝘁𝗲𝗿 𝗦𝗰𝗼𝗿𝗲 (𝗡𝗣𝗦) 𝟮𝟬% 𝗾𝘂𝗮𝗿𝘁𝗲𝗿 𝗼𝘃𝗲𝗿 𝗾𝘂𝗮𝗿𝘁𝗲𝗿 𝗯𝘆 𝘂𝘀𝗶𝗻𝗴 𝗔𝗜.

Their motives were well-intentioned. They're weren't looking to lay off their folks! They wanted to encourage innovation, increase their appeal to new-hires, and help reduce employees' stress. All good things.

The problem?

There was no collective agreement on the best means to improve NPS with AI across all functional areas. And 20% across the board was quite a big jump for many who have few high-touch but generally satisfied clients.

Evidence of AI strategy being disconnected from ground truth was abundant. For example, developers went about trying to improve NPS by turning around more code quicker, but customer service staff who prided themselves on delivering white glove service for demanding clients feared that introducing AI could lower the score.

By talking to the people on the front lines and how they perceived the role of AI in their workflows, I could convey the reasons why AI adoption was uneven or absent. AI targets that were too abstract or in conflict with other goals and values simply couldn't take off.

𝗕𝘆 𝗴𝗿𝗼𝘂𝗻𝗱𝗶𝗻𝗴 𝗺𝘆 𝗰𝗹𝗶𝗲𝗻𝘁 𝗶𝗻 𝘁𝗵𝗲 𝗼𝗽𝗲𝗿𝗮𝘁𝗶𝗻𝗴 𝗿𝗲𝗮𝗹𝗶𝘁𝗶𝗲𝘀 𝗼𝗳 𝗲𝗮𝗰𝗵 𝗱𝗲𝗽𝗮𝗿𝘁𝗺𝗲𝗻𝘁, 𝘄𝗲 𝗰𝗼𝘂𝗹𝗱 𝗰𝗿𝗲𝗮𝘁𝗲 𝗮 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁𝗶𝗮𝘁𝗲𝗱 𝗔𝗜 𝗴𝗼𝘃𝗲𝗿𝗻𝗮𝗻𝗰𝗲 𝗮𝗻𝗱 𝗮𝗱𝗼𝗽𝘁𝗶𝗼𝗻 𝘀𝘁𝗿𝗮𝘁𝗲𝗴𝘆.

It was a great case for looking-before-you-leap, and one of many reasons why 'GroundSpring' resonates.

𝗜'𝗺 𝗴𝗲𝗻𝘂𝗶𝗻𝗲𝗹𝘆 𝗰𝘂𝗿𝗶𝗼𝘂𝘀: 𝗪𝗵𝗲𝗿𝗲 𝗱𝗼 𝘆𝗼𝘂 𝘀𝗲𝗲 𝗔𝗜 𝗺𝗮𝗻𝗱𝗮𝘁𝗲𝘀 𝗮𝘁 𝗼𝗱𝗱𝘀 𝘄𝗶𝘁𝗵 𝗿𝗲𝗮𝗹𝗶𝘁𝘆 𝗶𝗻 𝘄𝗮𝘆𝘀 𝘁𝗵𝗮𝘁 𝗽𝗿𝗲𝘃𝗲𝗻𝘁 𝗵𝗲𝗹𝗽𝗳𝘂𝗹 𝗔𝗜 𝗮𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻𝘀 𝗳𝗿𝗼𝗺 𝘁𝗮𝗸𝗶𝗻𝗴 𝗵𝗼𝗹𝗱? (𝗜 𝗞𝗡𝗢𝗪 𝘆𝗼𝘂 𝗺𝘂𝘀𝘁 𝗵𝗮𝘃𝗲 𝘀𝗼𝗺𝗲 𝗲𝘅𝗮𝗺𝗽𝗹𝗲𝘀!)

05/14/2026

🥳 The GroundSpring brand birthday party continues with a 𝗙𝗥𝗘𝗘 𝗚𝗜𝗙𝗧 𝗳𝗼𝗿 𝗬𝗢𝗨! 🎁

This 𝗔𝗜 𝗢𝗽𝗽𝗼𝗿𝘁𝘂𝗻𝗶𝘁𝘆 𝗖𝗮𝗻𝘃𝗮𝘀 is a fun little reminder to stay grounded with your AI strategy by putting pen to paper 𝘣𝘦𝘧𝘰𝘳𝘦 you consult an AI tool or begin solutioning.

I included a few responsible AI-specific prompts to consider:
• 𝗗𝗮𝘁𝗮 𝘀𝗼𝘂𝗿𝗰𝗲𝘀: Good AI = Good Data. Think about what sources you're pulling from and whether it's sensitive or not.
• 𝗪𝗵𝗼 𝗶𝘀 𝗮𝗳𝗳𝗲𝗰𝘁𝗲𝗱? 𝗔𝗡𝗗 𝗪𝗵𝗮𝘁 𝗰𝗼𝘂𝗹𝗱 𝗴𝗼 𝘄𝗿𝗼𝗻𝗴?: We don't ask these questions early or seriously enough in our strategy process. Make sure you're not just planning for the Happy Path.
• 𝗞𝗲𝘆 𝗢𝘂𝘁𝗰𝗼𝗺𝗲 𝗮𝗻𝗱 𝗦𝘂𝗰𝗰𝗲𝘀𝘀 𝗖𝗿𝗶𝘁𝗲𝗿𝗶𝗮: Start with ONE targeted outcome and be specific about what good would look like if you succeed in achieving it. Being clear and targeted allow you to learn if you're actually benefitting from AI.
• 𝗪𝗵𝗲𝗿𝗲 𝗰𝗮𝗻 𝗔𝗜 𝗵𝗲𝗹𝗽 𝗮𝗱𝗱𝗿𝗲𝘀𝘀 𝘆𝗼𝘂𝗿 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄 𝗽𝗮𝗶𝗻 𝗽𝗼𝗶𝗻𝘁𝘀?: There's a continuum of how much/little can and should be automated. Worth a think before you dive into handing tasks off to AI.

💡 𝗣𝗿𝗼 𝗧𝗶𝗽: 𝘈𝘤𝘵𝘶𝘢𝘭𝘭𝘺 𝘱𝘳𝘪𝘯𝘵 𝘵𝘩𝘪𝘴 𝘰𝘶𝘵 𝘢𝘯𝘥 𝘸𝘳𝘪𝘵𝘦 𝘰𝘯 𝘪𝘵. Don't ask AI to fill it out first! After getting your thoughts out, you can see where your natural inclinations are before AI steers you down a path. 𝗬𝗼𝘂𝗿 𝗼𝘄𝗻 𝗰𝗼𝗻𝘁𝗲𝘅𝘁𝘂𝗮𝗹 𝗸𝗻𝗼𝘄𝗹𝗲𝗱𝗴𝗲 𝗶𝘀 𝗮𝗻 𝗶𝗺𝗽𝗼𝗿𝘁𝗮𝗻𝘁 𝘀𝗶𝗴𝗻𝗮𝗹.

𝗛𝗮𝘃𝗲 𝗳𝘂𝗻 𝗮𝗻𝗱 𝗲𝗻𝗷𝗼𝘆 𝘁𝗵𝗶𝘀 𝗴𝗶𝗳𝘁 𝗳𝗿𝗼𝗺 𝗚𝗿𝗼𝘂𝗻𝗱𝗦𝗽𝗿𝗶𝗻𝗴! Don't forget to keep it REAL and touch grass! 🌱

🥳 The GroundSpring brand birthday party continues with a 𝗙𝗥𝗘𝗘 𝗚𝗜𝗙𝗧 𝗳𝗼𝗿 𝗬𝗢𝗨! 🎁 This 𝗔𝗜 𝗢𝗽𝗽𝗼𝗿𝘁𝘂𝗻𝗶𝘁𝘆 𝗖𝗮𝗻𝘃𝗮𝘀 is a fun little reminder to stay grounded with your AI strategy by putting pen to paper .....

05/13/2026

🎂 Today marks a milestone: My new brand's birthday!! 🎈

I'd like you all to meet 𝗚𝗿𝗼𝘂𝗻𝗱𝗦𝗽𝗿𝗶𝗻𝗴!

If you listen to me or read my work, you know I talk all the time about grounding your AI strategy in what's real - your organizational goals and values, actual team workflows, and your stakeholders' interests.

Being 𝗴𝗿𝗼𝘂𝗻𝗱𝗲𝗱 and rooted in reality is foundational for counteracting hype to achieve big things and 𝙨𝙥𝙧𝙞𝙣𝙜 ahead toward your vision. (See what I did there? 😏)

🧘‍♀️ I've likened this idea to "root-to-rise" in yoga. It's only be getting firmly settled into your foundations that you can lift and stretch.

The same is true for AI, and the services I bring my clients.

I work with people to understand their context, pull that into their tech stack and governance, and make sure they stay true to what people want and need.

It also perfectly supports my new service line: REAL AI Governance™️.

𝘎𝘳𝘰𝘶𝘯𝘥𝘪𝘯𝘨 𝘵𝘰 𝘴𝘱𝘳𝘪𝘯𝘨!

In the coming days and weeks, you'll see GroundSpring roll out. (Anyone who has rebranded knows it's more than a flip switch!)

As for my brand of 5+ years, Saddle-Stitch Consulting will hang around in the background for a while. The brand came from a time, place, and service line that served people well, but feels a little forced today.

I'll be sharing some fun GroundSpring-y developments in the coming weeks. 𝗦𝘁𝗮𝘆 𝘁𝘂𝗻𝗲𝗱 𝗮𝗻𝗱 𝘁𝗼𝘂𝗰𝗵 𝗴𝗿𝗮𝘀𝘀! 🌱

05/10/2026

There's little I like more than moderating a conversation among AI subject matter experts! Join me on 6/17 for a panel discussion that will inform how you think about the "US Factor" in risk management, including:

- US-China Tensions: The impact on cybersecurity and data sovereignty.

- The AI Arms Race: How intensifying competition reshapes global regulation.

- Big Tech: The invisible hand in European risk exposure.

Thanks to GRC World Forums Ltd and these awesome panelists for what will undoubtably be an important conversation:

🎙️ Dr. Nathalie Moreno (Kennedys)

🎙️ Nicola Cain (Handley Gill Limited)

🎙️ Joel Schwarz (Global Cyber Expert)

Register FREE here: https://webapp.spotme.com/login/grcforums/risk-digital-global-June-2026

04/23/2026

"𝘗𝘳𝘪𝘤𝘪𝘯𝘨 𝘪𝘴 [𝘢𝘯 𝘈𝘐] 𝘱𝘳𝘰𝘥𝘶𝘤𝘵 𝘥𝘦𝘤𝘪𝘴𝘪𝘰𝘯." - Charu Maheshwari.

I learned from many great sessions at last week's Applied Machine Learning Conference, but Charu's presentation, "𝗣𝗿𝗶𝗰𝗶𝗻𝗴 𝗜𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲," stands out for going well beyond the surface for a thoughtful analysis of what AI economics means for product decision-making.

Pricing is a topic more and more AI product folks will be contending with, and, surprising to me, future pricing models may even enable greater customer leverage.

I've known that, unlike SaaS products, genAI-enabled products don't benefit from the same economies of scale. For example, marginal token costs hold steady even as customers consume more of them.

What I didn't think about were the many product implications this fact carries with it, along with some pricing variables I didn't know existed.

𝗛𝗲𝗿𝗲 𝗮𝗿𝗲 𝘀𝗼𝗺𝗲 𝗼𝗳 𝗖𝗵𝗮𝗿𝘂'𝘀 𝗽𝗼𝗶𝗻𝘁𝘀 𝘁𝗵𝗮𝘁 𝘄𝗲𝗿𝗲 𝗲𝗻𝗹𝗶𝗴𝗵𝘁𝗲𝗻𝗶𝗻𝗴:

• 𝗢𝘂𝘁𝗽𝘂𝘁 𝘁𝗼𝗸𝗲𝗻𝘀 𝗰𝗼𝘀𝘁 𝗠𝗔𝗡𝗬 𝗧𝗜𝗠𝗘𝗦 𝗠𝗢𝗥𝗘 𝘁𝗵𝗮𝗻 𝗶𝗻𝗽𝘂𝘁 𝘁𝗼𝗸𝗲𝗻𝘀. Implication: Longer prompts with shorter answers cost less than the reverse.

• The Iron Triangle of AI pricing consists of 𝗜𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲, 𝗦𝗽𝗲𝗲𝗱, 𝗮𝗻𝗱 𝗖𝗼𝘀𝘁. Implication: Product folks should think about the customer experience trade-offs that make sense when factoring in latency.

• 𝗧𝗵𝗲𝗿𝗲 𝗮𝗿𝗲 𝟯 𝘄𝗮𝘆𝘀 𝘁𝗼 𝗺𝗲𝗮𝘀𝘂𝗿𝗲 𝘀𝗽𝗲𝗲𝗱; Not all work equally well for every use case and AI approach: 𝟭. 𝗧𝗧𝗙𝗧 (Time to First Token), 𝟮. 𝗢𝗧𝗣𝗦 (Output Token Per Second), 𝟯. 𝗧𝗧𝗟𝗧 (Time to Last Token)

• 𝗧𝗵𝗲𝗿𝗲 𝗮𝗿𝗲 𝟰 𝗽𝗿𝗶𝗰𝗶𝗻𝗴 𝗺𝗼𝗱𝗲𝗹𝘀 𝘁𝗼 𝗰𝗼𝗻𝘀𝗶𝗱𝗲𝗿: 𝟭. 𝗨𝘀𝗮𝗴𝗲, 𝟮. 𝗛𝘆𝗯𝗿𝗶𝗱 (𝘀𝘂𝗯𝘀𝗰𝗿𝗶𝗽𝘁𝗶𝗼𝗻 + 𝘂𝘀𝗮𝗴𝗲), 𝟯. 𝗖𝗿𝗲𝗱𝗶𝘁-𝗯𝗮𝘀𝗲𝗱, 𝗮𝗻𝗱 𝟰. 𝗢𝘂𝘁𝗰𝗼𝗺𝗲-𝗯𝗮𝘀𝗲𝗱. #4 requires VERY concrete agreement on what a successful outcome is, but this method could give customers power to ensure they're getting the value they demand.

Overall, Charu noted it's 𝗶𝗺𝗽𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝘁𝗼 𝗳𝗶𝗻𝗱 𝗣𝗿𝗶𝗰𝗶𝗻𝗴-𝗠𝗮𝗿𝗸𝗲𝘁-𝗙𝗶𝘁, which requires 𝗼𝗻𝗴𝗼𝗶𝗻𝗴 𝗲𝘅𝗽𝗲𝗿𝗶𝗺𝗲𝗻𝘁𝗮𝘁𝗶𝗼𝗻 𝗮𝗻𝗱 𝗮𝗱𝗷𝘂𝘀𝘁𝗺𝗲𝗻𝘁𝘀.

👏 👏👏 Bravo on a great talk that I'm definitely going to come back to and apply to my strategy work!

04/22/2026

👋 Hello there, Facebook! I haven't seen you in a while. It's good to be back.

Why so quiet? All good reasons, I'm glad to say. Over a four-day period, I delivered two new workshops and caught up with a cohort of AI product managers. It's been a wildly invigorating time. I also needed a good nap...or two.

Now that I've perked up (🙏☕ ), I have lots to share.

I'll start with the event I self-hosted on Monday: ⚖️ 𝗔𝗻 𝗔𝗜 𝗚𝗼𝘃𝗲𝗿𝗻𝗮𝗻𝗰𝗲 𝗟𝗲𝗮𝗱𝗲𝗿𝘀𝗵𝗶𝗽 𝗪𝗼𝗿𝗸𝘀𝗵𝗼𝗽. 💪

Nearly 20 people bought tickets to gather for an in-person, 90-min. session about how we can support each other in making our AI use and deployment responsible and HUMANE. I was fired up to offer this session because I've seen far too many well-intentioned leaders trying to make sense of the chaotic technical, regulatory, and cultural environments we're in thrown into without much guidance. We all want to do the "right thing," but it's hard to know what that is and where to start.

To help, 𝗜 𝘀𝗵𝗮𝗿𝗲𝗱 𝗮 𝗻𝗲𝘄, 𝗰𝗼𝘂𝗻𝘁𝗲𝗿-𝗵𝘆𝗽𝗲, 𝗱𝗲𝘀𝗶𝗴𝗻-𝗼𝗿𝗶𝗲𝗻𝘁𝗲𝗱 𝗮𝗽𝗽𝗿𝗼𝗮𝗰𝗵 𝗜'𝗺 𝘂𝘀𝗶𝗻𝗴 𝘁𝗼 𝗴𝘂𝗶𝗱𝗲 𝗽𝗲𝗼𝗽𝗹𝗲 𝗶𝗻 𝗺𝗮𝗸𝗶𝗻𝗴 𝗔𝗜 𝗴𝗼𝘃𝗲𝗿𝗻𝗮𝗻𝗰𝗲 𝘁𝗵𝗮𝘁 𝘄𝗼𝗿𝗸𝘀 𝗶𝗻 𝗽𝗿𝗮𝗰𝘁𝗶𝗰𝗲. 𝗜 𝗰𝗮𝗹𝗹 𝗶𝘁 𝗴𝗲𝘁𝘁𝗶𝗻𝗴 𝗥𝗘𝗔𝗟 𝘄𝗶𝘁𝗵 𝗔𝗜 𝗴𝗼𝘃𝗲𝗿𝗻𝗮𝗻𝗰𝗲.

𝗥ooted, 𝗘mbedded, 𝗔ctionable, 𝗟egible

Getting REAL means we don't just write policy docs and throw them over the fence. Instead, 𝘄𝗲 𝗱𝗲𝘀𝗶𝗴𝗻 𝗴𝗼𝘃𝗲𝗿𝗻𝗮𝗻𝗰𝗲 𝘁𝗼 𝗺𝗲𝗲𝘁 𝗽𝗲𝗼𝗽𝗹𝗲 𝘄𝗵𝗲𝗿𝗲 𝘁𝗵𝗲𝘆 𝗮𝗿𝗲: We talk and empathize with our stakeholders, build in ways to make it easy to comply, and provide tools, education, and clear guidance so that people can actually do their work in alignment with organizational values and risk profiles.

Any time someone ventures to spend time learning from me, I take their trust seriously. I hope attendees had as terrific a time as me getting to know each other and how we're tackling shared challenges.

𝗜'𝗹𝗹 𝗯𝗲 𝗿𝗲𝘃𝗶𝗲𝘄𝗶𝗻𝗴 𝗳𝗲𝗲𝗱𝗯𝗮𝗰𝗸, 𝗮𝗱𝗷𝘂𝘀𝘁𝗶𝗻𝗴, 𝗮𝗻𝗱 𝘁𝗮𝗸𝗶𝗻𝗴 𝘁𝗵𝗶𝘀 𝗼𝗳𝗳𝗲𝗿𝗶𝗻𝗴 𝗼𝗻 𝘁𝗵𝗲 𝗿𝗼𝗮𝗱! ( 👀 𝗟𝗼𝗼𝗸𝗶𝗻𝗴 𝗮𝘁 𝘆𝗼𝘂, 𝗥𝗶𝗰𝗵𝗺𝗼𝗻𝗱, 𝗥𝗮𝗹𝗲𝗶𝗴𝗵, 𝗮𝗻𝗱 𝗖𝗵𝗮𝗿𝗹𝗼𝘁𝘁𝗲!)

🙌 Kudos to all the leaders out there trying to make the world, their organizations, and teams better. I see you and am here to help!

Stay tuned for a lot more new-news coming up!

02/14/2026

Why 'Human-in-the-Loop' isn't a magic fix.

You'll often hear that a best practice for mitigating agentic AI risk is to put a human in (or on) the loop to ensure that the AI is behaving as it should.

This is a good practice, but you need to consider both how to scale it AND how to design the review step for the human beings to make good decisions.

Be sure to put design effort into supporting those humans in and on the loop to keep your systems running safely.