News
eBook
Designing an AI-Native Contact Center
Key Takeaways
- Poetiq raised $45.8M from leading venture capital firms.
- Funds will accelerate self-improving expert AI agent development.
- Business leaders may see more powerful, cost-efficient AI integrations ahead.
A six-person startup founded by former Google DeepMind researchers just secured $45.8 million to prove reasoning overlays can outperform raw LLM scaling.
Poetiq announced on Jan. 29, 2026, that it raised the seed funding needed to accelerate development of self-improving AI agents. The round was co-led by Surface and FYRFLY, with participation from Y Combinator, 468 Capital, Operator Collective, NeuronVC and HICO.
According to company officials, the capital will fund a system that automatically creates expert agents capable of outperforming their underlying language models. The platform integrates with major frontier models including ChatGPT, Claude and Gemini, aiming to reduce the data, time and cost required for advanced problem solving.
Table of Contents
- Inside Poetiq’s Reasoning Layer
- Benchmark Breakthroughs Signal Early Traction
- Why LLM Strategy Is Moving Beyond Pretraining
- Poetiq at a Glance
Inside Poetiq’s Reasoning Layer
According to Poetiq, its platform delivers several benefits for enterprise AI teams:
| Feature | What It Enables |
|---|---|
| Recursive self-improvement | System aims to improve with every iteration |
| Model-agnostic integration | Works with ChatGPT, Claude, Gemini and other LLMs |
| Expert agent creation | Automatically builds specialized agents for tasks |
| Cost optimization | Targets reduced data, time and cost for problem solving |
Benchmark Breakthroughs Signal Early Traction
"LLMs are impressive databases that encode a vast amount of humanity's collective knowledge. They are simply not the best tools for deep reasoning... For ARC-AGI 1 and 2, we used recursive self-improvement to produce specialized agents in a matter of hours."
- Shumeet Baluja
Co-CEO, Poetiq
Poetiq offers a model-agnostic intelligence layer that enhances large language models through an overlay reasoning system.
The platform's key capabilities include learned test-time reasoning that achieved state-of-the-art results on the ARC-AGI-2 Semi-Private Test Set, model-agnostic deployment, automatic system creation, cost optimization reducing per-problem costs by 60% and self-improvement capabilities that learn from each solved task.
In December 2025, Poetiq surpassed the ARC-AGI-2 benchmark with 54% accuracy while cutting per-problem costs by more than half.
Founded in June 2025 by Shumeet Baluja and Ian Fischer, the company later pushed accuracy to 75% on the public evaluation set using GPT-5.2 X-High at under $8 per problem. A Puck profile noted the team achieved its benchmark results using roughly $40,000 in compute.
Why LLM Strategy Is Moving Beyond Pretraining
Major providers are pivoting from pretraining to reasoning models that work through complex problems rather than simply scaling compute and data.
OpenAI, Anthropic, Google and DeepSeek now offer reasoning models designed for math and coding tasks. Enterprise testing shows strong interest in these systems and the agentic capabilities they enable.
Fine-tuning with proprietary data is becoming less necessary as model capabilities improve. Experts recommend enterprises implement continuous AI purple teaming, third-party validations and autocorrection layers.
Poetiq at a Glance
Poetiq targets enterprises, AI product teams and research organizations seeking to improve the reasoning performance of large language models without retraining.
Webinar
Feb
25
Content Strategy Leaders Live: Scaling for Speed, Complexity and AI in High Tech
A candid roundtable on how high-tech leaders are rethinking content at scale.
Register
Webinar
Feb
26
Do More with Less: Modernizing the Cloud Contact Center for 2026
Learn how to leverage cloud platforms without adding a single hire to personalize every customer interaction.
Register
Webinar
On demand
Cut the Noise: Deploying AI That Actually Moves the Needle
Learn how to turn AI experimentation into concrete revenue operations.
Watch Now
Webinar
On demand
Ditch the Desk Phones: How Modern Teams Drive AI-First Communications
Find out how one team finally pulled the plug on a legacy phone system. And built something smarter.
Watch Now
Webinar
On demand
Rebrand. Migrate. Optimize. How to Do It All (Without Slowing Down)
Cresta leveled up site speed, design flexibility and marketer sanity (in record time). Find out how.
Watch Now
Webinar
On demand
From Manual to Magical: How AI Transforms CX Teams
Learn how to replace manual support processes with automation that actually delivers.
Watch Now
The company was founded in 2025 and offers a model-agnostic intelligence layer that enhances reasoning through agent-based, iterative problem solving and self-auditing.