The idea that digital signage could exist without AI has faded quickly in the past couple of years. The question is no longer if AI should be integrated but where to start. Generative AI, in particular, has made its way into nearly every part of an organization – from back-office operations to marketing and product development.
Jason Cremins, CEO at Signagelive, is seen as one of the industry’s innovation leaders – not just because of Signagelive’s modular, headless CMS platform, but also because of the company’s openness to new technology.
Signagelive has followed an API-first strategy for over a decade and, earlier this year, moved its entire tech stack to AWS.
For Cremins, one thing is clear: as AI becomes more integrated, every software provider will be forced to rethink their system architecture. Without data-rich APIs, AI integration simply isn’t possible. Going forward, AI will be a core part of every modular, service-oriented platform.
From scattered tools to structured integration
The AI landscape is changing fast. In 2024, it was still mostly a chaotic mix of tools and platforms. AI applications are increasingly being integrated into established frameworks like AWS, increasing security and adding guardrails. While new tools are still popping up, open standards like Model Context Protocols (MCPs) and Google’s Agent2Agent (A2A) are helping bring structure and consistency to the space.
Anthropic, the creators of Claude and MCP, call MCP the ‘USB-C port for AI applications’, enabling integration with existing software and solutions. The more API endpoints a platform makes available via MCPs as Tools, the more capabilities can be unlocked through AI – a crucial competence of today’s digital signage platforms.
MCP-connected digital signage software will enable users to use natural language to communicate with their platform and retrieve information and visuals exactly according to their needs at that moment, not as prescribed by a software vendor within a traditional CMS interface.
Semi-automation with AI agents
Cremins sees the next big step as using AI agents to semi-automate workflows. The potential is huge – but it comes with risks. At some point in the process, a human still needs to have the final say, whether through approvals or built-in control mechanisms.
Is the market ready for automated content creation? According to Cremins, the answer is yes – but only with the right controls in place. This requires a clear set of rules to guide the process, such as flagging systems for oversight. While setting up these safeguards takes effort upfront, they can leverage existing frameworks or be built using shared standards.
To wrap up, Jason Cremins shares his six current favorite AI tools:
- Notebook LM – Turns your own files and data into a personalized LLM.
- Gemini – Google’s AI model with versatile application possibilities.
- Zaper Agents and MCP – Ideal for automating internal processes, including Salesforce prospect and customer enrichment with background information provided to sales teams via Slack.
- Claude Desktop – A B2B-friendly alternative to ChatGPT with native MCP support, enabling agentic workflows between.
- Zapier MCP and other MCP-supported applications.
- Snipd – AI Podcast application that automatically captures highlights, creates AI summaries and integrates with my Readwise knowledge store along with Kindle and website highlights.