The New Rules Of AI Innovation

If you cannot innovate with emerging technologies, your business is losing its right to exist.

C-level leaders know they must innovate to succeed with AI, but most businesses aren’t built to innovate. A growing number of executive leaders are realizing that they must transform their business from agile or fast followers to innovators. CIOs to CEOs are interested in hearing what that means and how to make it happen. Speakers say the word ‘innovation’ less because no one’s showing up just to hear the word surrounded by aspirational statements.

At the same time, I’m hearing business leaders explain how to innovate much more often. Innovation can’t be a buzzword anymore because executives are expected to manage innovation cycles and generate returns from the process. At Salesforce’s Agentforce World Tour NYC event, Indeed’s CIO explained an unseen barrier that prevents most businesses from innovating: “You have to take the fear out of the process.”

Managing The Risk Of Innovation

The uncomfortable truth about innovation and AI is that sometimes they go wrong. Executive leaders to front-line workers are afraid of failure, and it’s impossible to take that out of the equation. LLMs hallucinate, and innovation initiatives that look promising may not meet success immediately.

We delivered an AI agent that supports clients during meetings, so of course, we focused on the meeting workflow. When only a few people adopted the agent, we realized that it wasn’t integrated into the full workflow. Meeting scheduling happens days or weeks before the actual meeting. The agent’s AI worked, but the product was a partial solution.

The agent wasn’t part of the meeting scheduling workflow, so our clients didn’t see the option presented to them when they needed it. Adoption suffered because people only thought of the agent after it was too late. We addressed the issue by giving the agent its own persona, email, and calendar (even though it’s always available). We introduced the agent like a new teammate, and adoption rates have risen significantly in the last quarter.

I have been innovating with data and AI for 13 years. I teach innovation strategy and management, and even wrote a book on it. I still can’t mitigate all the risks and prevent every mistake. The best we can do is rapidly detect issues and deploy an effective solution the first time.

If I had fired the team for missing the meeting scheduling workflow (myself included), no one would want to work on the next innovation initiative. The career risks are just too high. You have to take the fear out of the process, which means accepting and incentivizing new behaviors.

Building A Culture Of Experimentation

How do business leaders keep AI innovation from becoming a massive boondoggle? Risk must be rewarded on both sides of the equation. People taking risks can’t be punished for failure, but the business will quickly stop taking risks if some of them don’t deliver significant value. In other words, business leaders expect to get paid to take risks.

It’s a lot like a treasure hunt. Business leaders must fund the search and dig before anyone is certain of the outcome. There’s enough data to indicate something is worth exploring, but not enough to say with certainty what will be discovered. The best innovators and treasure hunters have failed. It’s part of the process.

The fear is that innovation failure becomes endemic because innovators are taking the wrong kinds of risks. Again, the CIO of Indeed offered a key insight. Businesses must “create a culture of experimenting, learning, and going after value.” Reframing innovation as an experimental workflow is the right approach.

Experiments must be justified. It’s not enough to propose a promising experiment. Innovators must provide evidence showing that there’s something worth investigating further. Business leaders must fund the experiment based on its merits. They need complete transparency into the potential for this to result in learning, not a monetizable artifact.

Last year, Agentforce launched with a $2 per conversation pricing model, but Salesforce found customers quickly wanted their agents to do more than have conversations. So, they announced new pricing tiers: credits priced at $0.10 per action, a new flexible agreement that allows companies to turn in seats for credits, and new licenses that give employees unlimited usage of internal agents. All this is designed to spur experimentation and innovation.

It’s critical for innovation failures to create value in the form of new information, learning, or experience. Every failure must make the business better at innovating and inform decisions about what experiments to fund or how to fix what went wrong. Learning and improvement cycles are part of the innovation process. How do businesses “go after value?”

Keeping Innovation Connected To Value

It’s no longer enough to just sell technology as a vendor. Salesforce is building its platform and events to go all-in on a partnership business model. It’s delivering a vision through technology of what an Agentforce business could be. The opening video at Agentforce World Tour NYC featured Mathew McConaughey asking, “What if...” and laying out a vision of what’s possible with agents.

Salesforce is no longer a SaaS vendor, and it has decided to completely reinvent its business model. Agentforce is an ecosystem. Everyone on the platform is a partner focused on delivering outcomes rather than technology. Even the AI research team is transforming around the new business model.

I spoke with Itai Asseo, Head of Incubation & Brand Strategy, AI Research, at Salesforce, and he explained that Salesforce is embedding AI engineers with their customers to align advanced AI innovation initiatives. Innovation artifacts are built around a customer need, challenge, or pain point. Value and customer outcomes are baked in from the start.

One challenge they tackled was how to validate and fully test AI agents. Part of trusting agents enough to put them in front of your customers is being able to validate that the agent won’t go off the rails. Testing agents that can answer thousands of questions or request variations is completely different from testing apps. The Salesforce AI Research team had to build a new environment for it. They call it CRMArena.

AI companies that lock in on delivering technology don’t see how testing environments must be a part of any solution. Focusing on the outcome and partnering with customers reveals the complete workflow that must be supported, or the outcome never materializes. Innovation disconnected from outcomes is just a shiny object.

AI Innovation Is A Journey, Not A Destination

There is no finish line. Innovation is neither a marathon nor a sprint. It must become part of every business’s DNA for it to thrive and benefit from disruption. Technology is changing faster than ever, so there will be no break in the waves of disruption. Transformation is continuous.

Most executive leaders realize that their business’s technical reliance will only increase from here. Put differently, the business’s reliance on AI will never be this low again. Technology’s role in the business will never be this simple again. That also means the AI opportunity will never be this small again.

We’re in the early innings of AI. What defines an AI innovator today will be commonplace in 3 years. Businesses can’t afford to stop innovating. Standing still, even when it’s at the front of the pack, really is falling behind.