When Gunnar and Magnus from Amplify Innovation approached me about their Systematic Innovation Management program, they had a classic learning and development challenge: comprehensive, valuable content that learners needed time to process, and often had follow-up questions about. Some would get stuck on specific concepts or implementation questions as they were going through the program.
Amplify Innovation's program teaches how to proactively and systematically lead and organize innovation efforts and build crucial innovation capabilities. During the program, learners would hit barriers: "How does this apply to my industry?", "What if my organization doesn't have this resource?", "Can you give me an example for a service business?". Amplify's team couldn't be available 24/7, yet the knowledge was essentially there - captured in the extensive video scripts, tools, and the wealth of examples accumulated over the years by the team. It was spread over multiple unstructured documents. We needed to find a way to provide the answers at the highest standards, rooted in the company's knowledge and in their unique voice.
This became the perfect test case for creating customer value through AI: Could we scale expertise without diluting quality?
Starting Point: The Friction
We mapped the learner journey: enroll, consume foundational content, attempt to apply frameworks, get stuck on specifics, email or message the instructors with questions, lose momentum or make assumptions that lead them in the wrong direction.
The friction wasn't content quality - it was accessibility of guidance at the moment of need. Amplify Innovation's team needed to focus on creating new value through other projects and engagements, and could not always be availalble for supporting the learners.
The AI Capability Match
After some research and tool comparisons, we identified that Voiceflow's conversational AI with knowledge base integration could help us with the following:
Understand questions in natural language from learners about innovation frameworks and concepts related to the program
Reference the comprehensive program content to provide accurate answers
Explain concepts with examples tailored to learner context
Available immediately, 24/7, in any language the learner preferred
Could scale at an affordable cost, and required a relatively low effort for monitoring and maintenance
We named it "Ainno" a learning companion that represents Amplify Innovation's knowledge and experience.
Starting Narrow: Two-Way Door Approach
We started with one specific use case: answering questions about program content. At the first phase we avoided any implementation guidance, or requests for content generation, just "help me understand this framework and content better." We built the initial version in a few short weeks. The knowledge base included program materials, but the prompts were carefully designed to:
Allow users to have follow-up questions on the initial querries
Provide reference to tools, concepts, examples and elaboration tied back tot he program content
Identify and acknowledge Ainno's specific focus when questions were outside its scope
Ask clarifying questions to ensure it understood context
Early Results and Evolution
Within the first month of usage, patterns emerged. Learners asked questions we anticipated - definitions, clarifications, explanations of concepts and frameworks, step-by-step breakdowns. But they also asked questions we didn't: implementation scenarios specific to their organizations, help generating first drafts of innovation briefs or strategies, examples from industries not covered in course materials, and more.
With time, we added more capabilities, including AI agents that could:
Help learners translate frameworks into their specific organizational context, with the agent asking clarifying questions about constraints, resources, and stakeholders.
Assist with creating first drafts of innovation artifacts (problem statements, opportunity briefs, stakeholder maps) based on Amplify Innovation's proprietary knowledge and best practices.
Ainno has already engaged in hundreds of conversations since it's introduction, and learner feedback is overwhelmingly positive. It also helped Amplify Innovation global partners around with supporting the program in other languages and contexts.
Key Learnings and Principles
- AI Amplifies Expertise, Doesn't Replace It
Ainno made the program and instructors knowledge more valuable, not less. Learners went through the program with better-formed questions, having already worked through basic understanding with Ainno. Amplify Innovation's team could spend time on nuanced guidance and complex scenarios rather than explaining foundational concepts for the twentieth time.
The principle: Design AI to handle the questions that have clear, documented answers so humans can focus on ambiguity, judgment, and relationship.
- Proprietary Knowledge is Your Moat
Generic AI tools can answer generic questions. Ainno's value comes from being trained on Amplify Innovation's specific methodologies, terminology, and examples. A learner could ask ChatGPT about innovation frameworks and get reasonable answers. But they can't get answers grounded in the specific approach they're learning, using consistent language, referencing the materials they're working with.
This created an added benefit: Ainno reinforces the program's unique value proposition. It's not just an assistant - it's proof that Amplify Innovation's methodology is sophisticated enough to require specialized guidance.
The principle: AI built on your proprietary knowledge becomes a competitive differentiator, not just a convenience feature.
- Measurement Drives Improvement
We tracked three categories of metrics:
Usage: How often do learners use Ainno? At what points in their journey? What questions do they ask?
Quality: Do learners consider responses as helpful? Do they ask follow-up questions (suggesting the first answer was inadequate)? How often is Ainno unbale to answer based on it's knowledge and capabilities?
Outcomes: Do learners who use Ainno complete more program modules? Apply frameworks more successfully? Rate the overall program higher?
The principle: Measure both process (is AI working as designed?) and outcome (does it actually improve the customer experience you care about?).
- Language Flexibility Unlocked Unexpected Value
We built Ainno in English because the program content was in English. But the large language models used by Voiceflow were able to handle translation capabilities, and a learner could ask questions in Spanish, Swedish, or Mandarin and receive answers in their language - even though the source material remained English.
This wasn't planned as a feature when we started out. We discovered it when a partner asked a question in Mandarin and got a coherent response. Now, Amplify Innovation could better serve international learners who were strong enough in English to consume course content or if it was delivered by a local partner, but could then query and explore complex concepts in their native language.
The principle: Build core functionality first, but stay alert for emergent value your customers discover that you didn't anticipate.
From Prototype to Strategic Asset
Ainno started as a two-month experiment to answer a focused question: Can AI help learners when human advisors aren't available? The answer was yes, but the real value emerged from iterating based on how learners actually used it.
This is the pattern across successful implementation of AI for customer experience:
Start with a specific friction or opportunity
Build a narrow solution
Test with real users
Learn where they push boundaries
Expand deliberately based on validated need
Gunnar and Magnus didn't need to believe AI would transform their entire program. They just needed to believe it might help with one important but focused problem. That belief, plus willingness to iterate based on evidence, turned a modest chatbot into a learning companion that enhanced their program's value proposition.
That's the second AI business value bucket in action: Creating new customer value by using AI to solve real problems, not by adding AI because everyone else is.