Getting serious traction with AI

If you’re a leader at a mid-sized B2B company, you likely have two conflicting messages in your head about AI.

On one hand it seems like magic. BUT.. On the other hand it maybe feels vague, risky, and hard to connect to real results.

The most successful teams I have worked with over the past year are resolving that tension by setting a clear definition of success, sequencing a few visible wins, and getting their data in a place where folks can ask questions in plain English and get answers. They’re not trying to do everything. They’re finding the highest leverage places to start, creating a small coalition of champions who make the work real, and begin stacking proof that shifts their culture/mindset.

This week’s newsletter takes my learnings over the last couple years of working with companies and gives you a simple path to make progress in about 90 days while getting ready for bigger changes in the years ahead.

1. Where to start? Define success.

Begin with the business plan, not the tool list. Put your goals at the center then ask where a digital teammate could move those numbers every week. Look at things like revenue, EBITDA/margin, cash flow, conversion. Also things like customer experience, engagement, acquisition, retention, and churn.

If the company needs more qualified pipeline, maybe aim at shortening the time to first touch and increasing meeting conversion. If margin pressure is the issue, maybe target forecast error and cost to serve so folks spend less time wrestling spreadsheets and more time making better decisions. If customer trust is fragile, focus on follow through, on time delivery, and the quality of the handoffs that shape the experience.

The point is to anchor AI to outcomes the company already cares about so the work is strategic, not a side project. When you connect AI efforts to the metrics that matter, it’ll be much easier to get tangible evidence that pulls skeptics along.

2. Build a pipeline of use cases & prioritize by effort & impact.

Once you’ve gotten clear on what success looks like, start creating a pipeline of use cases. Translate your business priorities and success metrics into measurable bets/tests/experiments/use cases. Build a simple value tree that links each outcome to the operational levers you control and then to the workflows you can automate or augment/accelerate with AI.

Once you’ve done this, then start plotting them across these 3 horizons so you can balance near-term efficiency/productivity with longer term differentiation and potential disruption.

1. Horizon 1: The 1st horizon focuses on efficiency. This is the work that reduces time & effort in the next 6-months or so. This could include things like meeting notes that write themselves and route to your CRM with the right tags. Or talking to your data and removing the need for ad hoc or custom reports. Or turning messy spreadsheets into clear recommendations.

2. Horizon 2: The 2nd horizon focused on differentiation. These are things that make your brand and company stronger and better positioned in a noisy market. This could look like a hyper-personalized experience for customers that helps them choose the right product/service based on their unique needs.

3. Horizon 3: The 3rd horizon focused on disrupting your industry. This looks like building new capabilities with AI that can completely change how you operate and upends your industry or blows your competitors out of the water. This could be something like real-time changes to your production line based on detecting issues earlier with computer vision. Or maybe it’s using “digital twins” (i.e., a digital copy of your physical space) that let you test processes without stopping production. Or maybe bringing completely novel products to market with faster research cycles enabled by AI.

Once you’ve got collected a bunch of ideas along these 3 horizons, I like to map them by effort and impact and focus only on the highest impact, lowest effort items for the 1st wave of AI bets. Document the rest so you have a roadmap instead of a random list of use cases.

Make sure to also nail down the definition of success before you start executing. Align on the outcome, owner, baseline & target metrics, how long the test/experiment will go, and whatever adoption threshold is acceptable before rolling out. Decide in advance what you will kill if numbers don’t move and what you’ll do if you start to see real impact.

3. Change is hard. Select champions to help shift mindsets.

The hardest part of AI is not the tech, it’s the humans. Getting people to buy-in and actually change their behavior and do something different than what they’ve always done can be a huge pain.

There are a lot of ways to approach this, but one big thing I suggest is identify a couple champions in the org who have some level of influence, are AI curious, have a bias for action, and are willing to invest the time/energy to test some things out.

Give them a simple charter:

1. Own enablement in their functional area, department, or team (i.e., they’re on point to help enable/teach/train others). They can do this by hosting show-and-tells so teammates can see live examples in their context to reduce fear and replace theory with evidence.

2. Run at least 1 test that maps to the north star.

3. Share monthly updates on what worked and what didn’t.

I’ve seen every one from C-Suite execs to mid-level managers to frontline workers become the best champions and advocates after they experience how much better/easier their workday gets with a few AI tools. This excitement creates a magnetic pull for the skeptics to want to try it out too.

4. Establish governance that doesn’t kill momentum.

Momentum matters, and so does safety. The best pattern I’ve seen is to set enterprise guardrails while still making it easy to try tools for real work. Obviously, keep sensitive data inside an enterprise grade LLM (large language model, aka AI). Set boundaries so models only draw from your approved knowledge bases when that’s important.

For steps where accuracy is critical, use deterministic modes and require citations. For ideation and drafting, allow more probabilistic modes that help people explore options.

When it comes to non-sensitive data and activities, put some basic guardrails in place for your champions to test new tools quickly while protecting your IP.

Some quick examples/thought starters in no particular order:

• Publish a short and plain consent script for meeting recording and event capture.

• Make sure you have policies that route captured data to approved systems instead of personal devices.

• Name an IT partner who can help shorten the cycles when something breaks or needs approval.

5. Win the basics. Some initial pilots you can run this quarter.

A lot of leaders feel overwhelmed with trying to prioritize all of the use cases and ideas and don’t know where to get started to start building momentum. If that’s you, here are a few easy ones to try out in horizon 1:

1. Automate meeting notes & actions: I feel like I say this daily, but amazingly so many teams aren’t using this yet and it’s an absolute slam dunk. Simply turn on meeting transcription and summarization so your organization stops losing decisions/knowledge in inboxes or personal notes. Then pair that with your CRM or project management tool/s so follow ups and tasks are created without extra effort. My team uses Fathom and Otter, which automatically join calls, generate searchable highlights, and summarize notes and actions. It saves so much time so we can all spend more time with customers instead of writing notes that few people will ever read. If you’re already on Microsoft, use Copilot. If you’re on Google, use Gemini.

2. Augment business intelligence & reporting: Make your data accessible to AI so your team can ask questions in plain English and get answers without waiting on an analyst. Some quick examples include Amazon QuickSight or ThoughtSpot that tighten decisions and reduce the temptation to build another spreadsheet. If you’ve got a ChatGPT Business plan, hook up connectors to things like Google Drive, OneDrive/Sharepoint, HubSpot, etc. and ask it to analyze the data and generate reports.

3. AI-enabled networking at trade shows, conferences, & events: Replace the post event spreadsheet shuffle and needing to manually type contact details in your CRM with tools like Backtrack. You can use it to scan badges, record quick context, and auto create contacts, notes, and tasks while the conversation is still fresh, then tee up immediate outreach so prospects hear from you within hours. It helps you keep track of folks and monitor time to first touch, and meetings booked without nagging people for updates that never seem to arrive.

Wrapping up

If AI still feels abstract, that’s a signal to narrow the scope and choose an initial use case that proves immediate value for your team.

Define a north star that connects the metrics that matter. Build a small pipeline of use cases across the 3 horizons and choose the highest impact, lowest effort ones to start. Name champions who will carry the torch and show their work. Set guardrails that protect the business without stalling progress. Then run a few simple tests/experiments that make everyday work lighter and faster, tell the story with your data, and invite others to join.

The goal is to build a smarter, calmer operating system for your company that helps people do their best work and helps customers feel the difference.

Onward & upward,
Drew

P.s. If we haven’t met yet, hello! I’m Drew Burdick, Founder and Managing Partner at StealthX. We work with brands to design & build great customer experiences that win. I share ideas weekly through this newsletter & over on the Building Great Experiences podcast. Have a question? Feel free to contact us, I’d love to hear from you.