Agency-to-Product Playbook: How Delos Secured €2.5M from 20VC by Building an AI Workspace Without Prompting

No Prompting Required: The Strategic Evolution That Won Delos €2.5M from Harry Stebbings
Himanshu Maggu
11 Jan 2022

Listen Now

Youtube // Spotify // Apple 


Founder’s Background and Intro

"Everyone is using ChatGPT even if it's not allowed," says Thibaut De La Grand'rive, gesturing emphatically during our interview. "The problem is that less than 10% of people in companies know how to prompt." This insight sits at the heart of why Thibaut and his brother Pierre founded Delos—not to create yet another AI model, but to solve the actual adoption crisis plaguing enterprise AI.

The brothers launched Delos literally 48 hours after GPT-3.5's release, but with an unconventional approach. While other founders raced to build products, they deliberately started as a high-end AI agency, securing contracts with Allianz across five European countries. This strategic move allowed them to get paid to research AI adoption challenges while building their core orchestration technology. "We never wanted to be an agency," Thibaut admits, "but we didn't want to take the wrong direction." This patience paid off. Their platform now orchestrates capabilities from Claude, Mistral, ChatGPT, and Google's models into a unified workspace where—crucially—no one needs to learn prompting. This focus on eliminating, rather than improving, the user-AI interface attracted investment from Harry Stebbings' 20VC, French billionaire Xavier Niel, and Patrick Pichette (ex-Google CFO). With their team of elite engineers from France's top technical schools, they're now preparing for US expansion and a major fundraise in 2026, positioning Delos not as an AI company, but as the enterprise "cockpit" that makes AI finally usable at scale.

See why top VCs are betting on Delos' vision for frictionless AI adoption in our founder interview → Watch the full interview 👇

What you’ll learn:

  1. Why starting as an agency provided the perfect foundation for building an AI product company

  2. How orchestrating multiple AI models creates more value than building new ones

  3. Strategies for attracting and retaining elite AI engineering talent in a competitive market

  4. The importance of product stickiness in a rapidly evolving AI landscape with hundreds of similar tools

  5. Why removing prompting is the key to enterprise AI adoption at scale

Some Takeaways:

  1. Agency-to-Product Strategy: "When there is a new sector opening up, it's very difficult to understand how people are going to use it, what they expect from you... So we had the opportunity quite quickly to sign a very good contract with Allianz... We were being paid for this research."

  2. Focus on Orchestration: "The concept is very simple. You don't have to choose any LLMs anymore. We are orchestrating the best out of all the best LLMs available in the markets... first we take the best out of them all and then we introduce a lot more technologies to make them way powerful, way stronger, way cheaper."

  3. Build Sticky Products: "Build a product that is going to stick to every people's life. If you are doing Cursor or Lovable, the problem is that the engineers can switch from one application to another in a day. Everyday people take a lot of time to adopt one product."

  4. Avoid Vertical AI Plays: "The meeting notes applications... it was a very good idea one year and half ago. Now we have 250 of them only in Europe. So it's getting very difficult to stay competitive in one sector."

  5. Engineer Retention Strategy: "You have to pay them well... The second thing is you have to create an identity. We impose Prisential, you are not allowed to do teleworking in our company... taking beers after every week when we did the good work, when we have a new client, everyone is celebrating."

Where to find Speaker:

• LinkedIn: https://www.linkedin.com/in/thibaut-de-la-grand-rive/

• Website: https://delos.so

In this episode, we cover:

00:00 - Introduction to Thibaut and Delos platform overview

00:59 - Founding story: Starting 2 days after GPT-3.5 release

03:27 - David's distributed team problem and platform demo request

05:34 - Live demo: Newsletter creation and AI orchestration workflow

11:37 - Using custom prompts and agent templates in Delos

12:52 - Website localization and international expansion plans

13:28 - From agency to product: The strategic validation journey

16:18 - Investor relationships and fundraising philosophy

19:12 - Meeting Harry Stebbings and the 20VC investment story

20:22 - War for AI talent: Competing with OpenAI's Paris office

23:36 - Future fundraising plans and capital allocation strategy

27:51 - Competitive threats: Will OpenAI copy their approach?

31:15 - US expansion strategy and investor considerations

33:55 - Pricing strategy and roadmap for agentic AI features

38:50 - Go-to-market focus: Why B2B enterprises are the target

41:47 - Key lessons: Avoiding vertical AI traps and building sticky products

45:40 - Future of jobs: How AI is reshaping the developer market

47:27 - Closing thoughts and product feedback

For inquiries about sponsoring the podcast, email david@thehomabase.ai

Referenced:

• Delos.so: Delos Platform

• Delos Technical Overview: Post-RAG Technology

• €2.5M Funding Round: French Tech Journal Coverage

• Alma Voice Interface: Product Announcement


Find Case Studies of all other episodes here.

Continue reading this post for free

You've reached the end of free content. Subscribe to continue reading!
🎉 Thank you! You signed up to Homebase newsletter.
Oops! Something went wrong while submitting the form.