From AI Interviews to Intelligence OS: Our Journey Building Propane

We quit our jobs in January 2025 with a mission: building the complete OS for customer intelligence for all teams.

What does that mean? An operating system. Like the OS on your computer handles storage, processing, memory, and visualization without you thinking about it, we hold customer intelligence the same way. You don't think about how you collect, organize, and connect data. You don't think about how you prompt, reason across data, or eliminate hallucination. You don't think about how you act on insights. We provide the whole stack. Collecting data. Organizing it. Reasoning through it and acting on it. So every team can focus on their actual job - building, selling, marketing, servicing customers.

2025 was the year we took the first steps toward it.

We started with AI interview agents.

We've been building product companies for 10-15 years. We've lived the same problem. Product doesn't know what Sales learned. Sales doesn't see what CX is hearing. Marketing guesses at messaging. Product already validated. Everyone rebuilds customer intelligence from scratch.

We were done with it.

But we didn't start with the whole OS. We were a team of three. We started with the first piece: AI interview agents.

Why start here?
This was our opportunity to scale customer intelligence data collection, not build yet another data aggregator and dashboard. We saw tremendous momentum around first-generation agents built for market research - scripted questions, classical survey logic. But we saw something different. Something product companies could get better value from.

When you're engaging actual customers in product use cases, you need context. You need autonomy. You need deployment into the customer journey itself.

We built agents with context about your company and users. You give them instructions for the outcome you want. They deploy anywhere - embedded in product, email, link, voice, text, 50+ languages. You get signals and reporting on the back end.

We launched with customers in April. Closed our pre-seed round in June. Spent the summer through September running product-market fit discovery and experimentation.

Then we kept shipping. Hard.

Contextual engine so agents know your offerings and features. SDKs for full configurability. Link and email distribution. Global incentives in 200+ countries with no marginal cost. Multi-modal voice and chat. Automatic reporting. Centralized repository for every conversation. APIs for full integration.

Case:
‍A web hosting and website building company with 2 million users, saw 30K in savings and 10x interview volume within just one month. They are now rolling it out across multiple product teams and brands. Intelligence was embedded into the customer journey rather than being separate research projects.

Our vision matched the market

While we were shipping, we kept talking to customers and prospects. The same questions kept coming up:

"How do we get everyone working from this intelligence, not just Product?"

The insight: the agent was one feature in a bigger stack.

Modern companies run differently now. Everyone needs customer intelligence. But the data is everywhere. CRM records. Support tickets. Sales calls. Product usage. Reviews. Feedback. Interview transcripts.

What we saw: Go-to-market teams building GTM platforms. User research teams are building research repositories. CX is building customer data platforms. Product building roadmap and strategy analytics platforms. Everyone is solving this in pieces.

We're becoming the horizontal stack for all teams. Not just one department.

Why now

The market is moving faster than the tooling can keep up. Companies need to make decisions in weeks, not quarters. But their intelligence infrastructure still requires specialists, separate tools for each team, and weeks of manual work to connect dots across systems.

Meanwhile, the AI agent explosion shows people want intelligent systems that work for them. But building those systems requires infrastructure most teams don't have - data pipelines, context management, and prompt engineering expertise.

The gap between what teams need and what they can build themselves is widening. That's the opportunity.

What this means for your team

Your Product team talks to 40 customers. Takes months. They write a killer insight doc. It dies in Notion.

Your Sales team keeps talking to customers. They close deals. They lose deals. All the qualification calls sit in Gong.

Your CX team handles tickets. Talks to customers. They see patterns emerge.

Yet no one has the whole picture. Stop doing research alone. Share it across teams.

With an operating system for customer intelligence:Product gets closed-lost deal intelligence automatically. Sales sees churn signals across support and usage. CX spots product gaps from interview data. Marketing validates messaging from actual customer conversations. Design sees what users need from every source.

One connected data tissue. Real-time access. No rebuilding. No siloes.

Here's what we're shipping in Q1

We're building and iterating with pilot customers now. Here's what you'll get access to:

Connected data foundation

We index everything. CRM records, support tickets, sales calls, product usage, conversations, reviews, feedback, and user interviews (AI or manual). We build deep, connected profiles on people and companies. Not isolated data points. A connected tissue where everything relates. A support ticket connects to usage data, interview responses, and the company profile with real-time access.

Intelligence layer with full business context

We inject your internal context. Your features, competitors, pricing, and service model. The system understands what customer data means for your business. You manage what context matters. We handle agent design, prompting, context windows, and token optimization. Zero technical overhead.

Spaces, not dashboards

We think about this as Perplexity thinks about spaces. Not prompt libraries. Not skills. A destination. A place to see things and dive deeper with questions and follow-up.

Every team designs its own space. Product focuses on specific customer segments. Sales needs enterprise deal intelligence. Marketing tracks messaging validation. Each team has different strategies, different focuses. They get access to what they need.

TL;DR summaries. Deep dives. All available. Making growth a team sport on shared intelligence.

Full integration and data activation

We support Intercom, Attio, HubSpot, and more. 100+ integrations coming soon.Our AI agents? One data source in the stack.

But here's where it changes the game. We activate the data.

Alert teams when signals appear. Trigger agents to rewrite website copy. Enroll users automatically into churn prevention workflows. Update systems. Ping Slack channels. Run scheduled research. Intelligence embedded where teams work.
‍

What does that mean for your teams and company?

We handle the infrastructure you don't want to build:

Building connectors to all your tools. Workflow integrations. Data warehouses and query interfaces. Different tools for each team. All of that. Turnkey. Highly personalized. Highly contextualized.

Not just a static database or document repository. A system that actually understands your business and handles all the heavy lifting for you.

From single-player complexity to multiplayer infrastructure

People are doing great work with MCPs, running agents on demand, and building analysis workflows in Cursor. This requires real infrastructure expertise - prompt design, context window management, and token optimization.

It's powerful work. But it's not scalable for most teams. Product managers, sales reps, designers, marketers - they need to focus on their actual jobs, not becoming AI infrastructure experts.

We're building for everyone else. The teams that want the power of AI-driven intelligence without the complexity of building and maintaining it themselves. That's what infrastructure does - it makes the complex accessible.

Pricing built for multiplayer

For the last two years, AI has been driven by token pricing, context window limits, data volumes, and seat gatekeeping. If we genuinely want to provide multiplayer infrastructure for all teams, we need to dramatically change pricing.

Simple, transparent pricing based on customer profiles we manage and index. Not tokens. Not data volumes. Not seats. Not features. Slack pioneered this early. Multiplayer software is available to everyone, regardless of team size. Simple and fair pricing. Everyone gets access.

But there's more. Everyone should care about their data, security, and compliance. So enterprise features need to be baked in. SOC 2. GDPR. Single sign-on. MFA. No premium tiers.

Enterprise-grade software. Highly available. Low cost. Baked in.

Ready to move customer intelligence from single-player to team sport?

You already have your CRM, support systems, and product analytics in place. We're opening up 10 spots to join the early beta in January. After that, you're on the waiting list.

We're working hands-on with founders and building momentum. Shaping what this category looks like. Research tools are siloed today. Nobody owns the entire data tissue across departments. Nobody creates the same availability for all teams.

If you're serious about building a customer intelligence infrastructure rather than treating it as a specialist job, book a call this week.
‍

Book a call with founders | Join the waitlist
‍Best,
Team Propane

‍