Markprompt – AI for customer support

Hi YC!

We’re Marie and Michael from Markprompt.

We are building AI infrastructure to power customer support at scale. We are live with companies like AngelList, Mindbody, and Plotly.

Problem

• Scaling customer support with headcount is expensive: Human resources account for 90%+ of customer support expenses.
• Customer support agents burn out: Support tasks incur large amounts of manual, repetitive, and ungratifying work that humans don't need to be doing, causing burnout, turnover, and reduced quality.
• Customer support is more than a chatbot on a website: In an API-driven world, customer support tools need to evolve with the company’s tech stack. Existing AI-powered solutions are not delivering on the promise, resulting in sub-par customer experience and low NPS and CSAT scores.
• Company resources are siloed across systems: As companies grow, so does their expert knowledge. It is hard to access at scale, and keeping employees up-to-date is nearly impossible.

Solution

• Comprehensive data indexing: Markprompt addresses the root of the information access problem by indexing all available knowledge within a company, setting the foundation for a robust system of record to feed the AI. This includes:
  • Salesforce Knowledge databases
  • Salesforce and Zendesk tickets
  • Notion wikis
  • GitHub repositories, discussions, and issues
  • Company websites, docs, guides, blogs
• An API-first approach: Markprompt is built from the ground up with an API-first approach. Every feature, from indexing and processing of data to generating drafts, triaging tickets, creating chat replies, or triggering automations, can be achieved via our APIs. Markprompt also provides SDKs (such as React hooks and components) and no-code tools to enable integrations at all touchpoints in the customer journey.
• Deep integrations: Markprompt works well within existing support platforms, such as Zendesk and Salesforce. And with our APIs, it is possible to build custom integrations to accommodate any setup.

Here is an example of a bespoke ticket deflection and triaging system built with our React libraries:

How we help

• Efficient ticket deflection: Deploy AI-driven chatbots and ticket deflection systems that efficiently handle routine queries, allowing support teams to focus on more complex cases.
• Accelerated case resolution: Use automated drafts to streamline case resolution, ensuring faster response times based on the up-to-date expert knowledge.
• Actionable AI agents: Enable AI agents to perform user actions, such as processing refunds or canceling a subscription, with high accuracy.

Who we are

We are life partners, and passionate about design and AI. Michael previously founded Snips (acq. by Sonos), a privacy-respecting voice AI platform that now powers all voice interactions on Sonos speakers. Marie and Michael previously founded Motif, a collaborative docs platform based on Markdown and React. While building Motif, we realized that the tasks of maintaining company expert knowledge up to date and providing stellar customer support are intimately connected, and both can in large part be automated. This led us to build Markprompt.

We are backed by founders and operators at Algolia, Microsoft, Snowflake, Stripe and Vercel.

Our asks

• Introductions: We're looking for tech-savvy companies processing 1000+ customer support tickets per month. Please let us know if you have connections to these companies, especially folks leading Eng or Customer Support. Please introduce us via founders@markprompt.com.
• Book a demo: here.