Joe Tannorella

Joe Tannorella Hello, I'm Joe.

I'm an ex Product VP turned AI entrepreneur and SaaS Founder.

When I'm not building my own products, I coach teams on using AI to accelerate growth, from GTM strategy to product development.

Some fun facts about me

Some of the tooling I've built with AI recently

Since being all-in on AI since 2023, I've built a lot of tooling to help me, my teams, and my clients use AI to solve real problems.

Piping prior podcast guests in, getting fully enriched leads out

  • Problem

    Prior podcast guests are a great source of qualified leads, but a source of recent guests with contact details does not exist.

  • Solution

    An automated pipeline that extracts guest name from episode descriptions, then Googles for the domain and LinkedIn profile to understand the individual and discover their contact information, along with relevancy scoring and more.

  • Outcome

    I now am able to find valid email address for 87% of podcast episodes with a guest in the business genre at scale (1000s per day), leading to a 6-figure pipeline within month 1 of use. We launched and monetised a new product thanks to this experiment.

Piping prior podcast guests in, getting fully enriched leads out image 1
Piping prior podcast guests in, getting fully enriched leads out image 2

A full social posting schedule with high quality, data-led content using APIs

  • Problem

    Our sister business wanted to be posting to LinkedIn and other social media platforms with high quality content, but did not have the time or resources to do so.

  • Solution

    We collaborated on what the perfect content strategy would look like (podcast guest breakdowns requiring millions of tokens of research and writing), and then built a system that would automatically generate and post the content to LinkedIn and other social media platforms on autopilot.

  • Outcome

    Their social schedule is completely fully months in advance, and they're getting incredible engagement and leads generated from the content (which is better researched and written than most human content!). AI-generated images are added to further improve the quality of the content.

A full social posting schedule with high quality, data-led content using APIs image 1

Customer Support with MCP servers

  • Problem

    Customer support requests are distracting and require a lot of manual research in order to understand and resolve.

  • Solution

    First I ensure that all core customer actions are available in Postgres audit tables, and then have connected Postgres MCP to Claude to be able to interrogate this accurately and easily via Claude Projects and Claude Code.

  • Outcome

    Higher quality and faster support resolution without needing to hire additional staff by finding root cause issues with ease, even for deeply technical or nuanced user experiences. We've saved thousands of hours of manual research and support time.

Enterprise MCP Server

  • Problem

    As customers move to the elusive chat-based textarea interface, they are expecting "answer engines" and not simply traditional search UX. One of our enterprise prospects wanted to implement Pod Engine entirely through MCP and not touch any API interfaces.

  • Solution

    I built an MCP server on top of our podcast APIs, allowing customers to ask questions of our database directly within their AI client of choice. Included user-level permissions and feature toggles for MCP outputs, including tool selection toggle, profanity filtering, and more.

  • Outcome

    This uncovered new use-cases that we did not anticipate being useful (production research), and also acted as a significant USP in enterprise API discussion talks with prospects (also a surprise).

Enterprise MCP Server image 1

6,500 AI-generated videos for website accessibility

  • Problem

    A client wanted to improve their website for accessibility purposes, and specifically wanted short (30s) videos to show a day in the life of thousands of different jobs.

  • Solution

    A pipeline that takes job title, audience age, and formalities into account, and generates 30 second fully scripted, high-quality videos that are on-brand.

  • Outcome

    The client is able to assess the proof-of-concept we ran to understand the feasibility of running this and further scale across all of the roles they have in their database.

LLM Insight & Evaluation at Scale

  • Problem

    Customers needed to know confidently if podcasts have guests or not, and various other types of discrete and open data points in search.

  • Solution

    We built an evaluation framework and human-verified dataset to run models and prompts against to be able to measure the trade offs between speed, accuracy, and cost.

  • Outcome

    We have one of the richest, most up-to-date datasets in all of podcasting. We process 50k episodes each day, extracting insights in a cost-effective, robust, and high-accuracy manner.

Continuously experimenting

Work with me

Let's build something together

I help teams and founders leverage AI to scale their businesses, improve their go-to-market, and build better products. Whether you need hands-on expertise or strategic guidance, I'm here to help.

Day Rate

Intensive sessions for specific challenges, workshops, or rapid prototyping.

Retainer

Ongoing advisory and hands-on support for continuous growth and implementation.

Get in touch