Poetiq Raises $45.8M for AI Meta-System, Surpasses Top LLMs on Industry Benchmark

Founded by former Google DeepMind scientists, Poetiq emerges from “stealth” after setting a new state-of-the-art (SOTA) on ARC-AGI-2

MOUNTAIN VIEW, Calif., Jan. 29, 2026 /PRNewswire/ — Poetiq, developer of an AI meta-system that makes LLMs work better, announced today that it raised $45.8 million in Seed funding co-led by FYRFLY Venture Partners and Surface Ventures with Y Combinator, 468 Capital, Operator Collective, Hico Ventures, and Neuron Venture Partners participating. The funding news follows Poetiq’s commanding results on ARC-AGI-2, an industry benchmark for machine reasoning and progress towards artificial general intelligence (AGI).

Poetiq can pair with any frontier LLM (OpenAI’s Chat GPT, Anthropic’s Claude, Google’s Gemini, Meta’s Llama, etc.) to make it learn faster and solve harder problems. Clients provide Poetiq with a problem and a few hundred examples instead of the thousands or millions required for fine-tuning or RL post-training. Poetiq’s meta-system generates an agent that specializes in solving that problem and recursively improves the agent to become more accurate and cost-efficient.

Poetiq was founded in June 2025 by co-CEOs Shumeet Baluja, PhD, and Ian Fischer, former AI researchers at Google DeepMind. Baluja was previously the CTO of Jamdat Mobile (IPO 2004) and spent the last 21 years with Google DeepMind, where he founded their mobile practice and started their fundamental computer vision research group. He has contributed to more than 170 patents in neural networks, machine learning, and applications and is one of the originators of YouTube’s copyright system. Fischer joined Google DeepMind through its 2015 acquisition of Apportable, a platform that ported iOS games to Android, where he was co-founder and CTO.

Collaborating at Google DeepMind, Fischer and Baluja noticed that frontier LLMs were struggling to solve most hard (or easy) problems. The current solution—to pre-train and post-train LLMs through reinforcement learning (RL)—takes weeks and is far too expensive for all but the biggest companies.

“LLMs are impressive databases that encode a vast amount of humanity’s collective knowledge,” said Shumeet Baluja, co-CEO of Poetiq. “They are simply not the best tools for deep reasoning. That’s why efforts to improve their problem-solving skills are so slow and expensive. For ARC-AGI 1 and 2, we used recursive self-improvement to produce specialized agents in a matter of hours. It demonstrates how much we can help with problems that have been too hard or too expensive for LLMs alone.”

An MIT study of 300 public AI implementations, published in August 2025, underscores the need for Poetiq. Although enterprises have invested $30 to $40 billion in GenAI, 95% of organizations are “getting zero return,” according to the researchers. Use cases that have struggled to generate an ROI are ideal candidates for Poetiq, as it can improve the reasoning capabilities of any LLM, including proprietary, in-house models.

“That Poetiq managed to top ARC-AGI within six months of launching is remarkable,” said Philipp Stauffer, General Partner at FYRFLY Venture Partners. “Rather than compete against frontier models, their team of six found a way to coax more intelligence from every LLM available. Poetiq will be a must-have for companies trying to make AI work for real-world business applications.”

“Poetiq is one of the rare AI startups that doesn’t need to outcompete frontier models or pick sides,” added Gyan Kapur, co-Managing Partner at Surface Ventures. “It can enhance any combination of LLMs, any native AI platform, and any AI use case. Poetiq can provide better performance at lower costs across diverse use cases by sitting on top of foundation models, and that is a unique position to be in.”

The Abstraction and Reasoning Corpus (ARC-AGI), developed in 2019 by AI researcher François Chollet, is a benchmark that measures an AI’s “human-like generalization” of problem-solving skills. In early December, Poetiq established a SOTA on the ARC-AGI-2 semi-private evaluation set, topping Gemini 3 Deep Think, the previous leader, at half the cost per task; this was done using Poetiq’s system on top of Gemini 3 Pro. Within a few days, OpenAI released GPT-5.2. Poetiq immediately incorporated this model into their system and showed a new SOTA at 75% accuracy (on the public evaluation set), a 16 percentage point improvement on the previous SOTA. OpenAI co-founder and President Greg Brockman tweeted in response to this feat, Poetiq is “exceeding the human baseline on ARC-AGI-2 with gpt-5.2.”

For technical details on Poetiq’s ARC-AGI-2 results, read their blog posts at:

To learn more about the company, visit poetiq.ai

About Poetiq

Poetiq is on the fastest path to AGI with a meta-system that makes frontier LLMs smarter and more cost-effective at solving real-world problems. Through recursive self-improvement, Poetiq generates AI agents to solve business problems that are too difficult or too expensive for LLMs alone. With each problem they solve, Poetiq agents become faster and more accurate at solving related problems. Visit poetiq.ai to learn more.

Press Contact:
press@poetiq.ai