Flex AI
Company Context
FlexAI is a Paris-based infrastructure startup that builds software to simplify access to GPU compute for AI teams. The company exited stealth in April 2024 with $30M in seed funding, founded by former Nvidia engineers Brijesh Tripathi (CEO) and Dali Kilani (CTO). FlexAI's platform abstracts the complexity of GPU orchestration, allowing ML engineers to train and deploy models without dealing with hardware management.
The GTM build happened during a turbulent period of organizational transition and a product pivot from hardware/software to 100% software. This reset created an opportunity to rebuild the go-to-market strategy from scratch, focusing on building systems ready to scale once product stability improved.
The Challenge
The product is difficult to sell into a highly restricted market. The target is extremely specific: startups and scale-ups with teams actively training their own AI models. This ultra-narrow addressable market demanded a surgical targeting approach.
- Specific ICP: Only companies training custom models (not just API wrappers)
- Credit Barriers: Early-stage startups often run on free cloud credits
- Technical Compatibility: Requirement for custom images and specific infrastructure stacks
Market Intelligence & Segmentation
The first step was defining the Total Addressable Market (TAM). We mapped every company globally with at least one AI/ML engineer, NLP engineer, computer vision specialist, MLOps professional, or data scientist—roughly 80,000 companies.
Tier 1: Growing Compute Needs
Early-stage startups ($2-5M ARR) with small ML teams (1-10 people) in non-sensitive, fault-tolerant markets like media, sports, education, and software.
Tier 2: Production-Ready (Non-Critical)
Companies with $2-20M ARR in markets like manufacturing, retail, and R&D, running production inference and training workloads outside of critical use cases.
Tier 3: Production-Ready (Critical Use Cases)
Scale-ups and enterprises with $20M+ ARR and larger ML teams (10+ people) operating in highly regulated industries like healthcare, financial services, defense, and aerospace.
Campaign Architecture
Signal-Based Targeting
- Job Posting Engine: Automated webhooks capturing 300+ daily AI/ML job postings worldwide to identify hiring intent.
- GitHub Monitoring: Tracking activity on key ML repositories (HuggingFace, DeepSpeed, PEFT) to find engineers actively fine-tuning models.
- Hugging Face Scraping: Identifying organizations regularly publishing new models as a signal of technical maturity.
- Social Listening: Extracting intent from LinkedIn engagement on competitor posts and industry pain points.
Multi-Channel Orchestration
Built a repeatable framework using a modern GTM stack: Clay, Cargo, n8n, and Lemlist.
- Event-specific outreach for major AI conferences
- Lookalike expansion based on existing SQLs
- AI SDR integration via 11x.ai to scale capacity
Key Outcomes
- 80,000: companies mapped globally
- 25,000: qualified targets identified
- 80%: lead coverage achieved
- 300+: daily signals processed
