Startup profile for Poetiq: latest funding, latest known valuation, total disclosed funding, investors, status, sources, and why the company is interesting. Part of the Venture Capital Tracker startup directory.

Startup profile · funding coverage

Poetiq funding, valuation and investors

Research startup improving reasoning performance through recursive self-improvement around existing models.

Verified company research · funding, investors and latest developments

Keep track of Poetiq

Save this profile to your VCT watchlist for a quick return.

View watchlist

Funding, valuation & investors

Answer-first snapshot

Latest funding

Seed

$45.8M · January 29, 2026

Latest known valuation

Not publicly disclosed

Total disclosed equity funding

$46M

Excludes debt, grants, acquisitions, secondaries, and IPO proceeds.

Current status

private

seed · United States

Investors in latest funding

Other: Surface Ventures , FYRFLY , Y Combinator , 468 Capital , Operator Collective , NeuronVC , HICO

Sources: latest funding · poetiq.ai Last verified: 2026-11-13

Overview

Poetiq was founded by former Google and DeepMind researchers Shumeet Baluja and Ian Fischer. It develops a recursive meta-system that orchestrates and improves reasoning with existing LLMs, targeting tasks such as ARC-AGI and code. On January 29, 2026 Poetiq announced a $45.8M seed with backers including Surface, FYRFLY, Y Combinator, 468 Capital, Operator Collective, NeuronVC and HICO. The company did not identify a lead investor.

Its benchmark improvements are selective and may involve extra inference-time compute.

Why Poetiq is interesting

A funded alternative to expensive foundation-model pretraining: Poetiq optimizes how external models solve tasks, rather than training its own NCM or diffusion base model.

Product & use cases

A recursive agent and reasoning meta-system that applies optimization and search around foundation models without full-model retraining.

  • Boosting smaller model performance on complex reasoning tests
  • Automatic task-specific agent optimization
  • Routing and combining foundation models for structured problem solving

Key facts

  • $45.8M company-announced seed on January 29, 2026
  • Backers include Surface, FYRFLY, Y Combinator, 468 Capital and Operator Collective
  • No lead investor explicitly disclosed in the company's funding post

Funding history (newest first)

Seed

2026-01-29 $45.8M
  • Surface Ventures (participant)
  • FYRFLY (participant)
  • Y Combinator (participant)
  • 468 Capital (participant)
  • Operator Collective (participant)
  • NeuronVC (participant)
  • HICO (participant)

Source: Read source · poetiq.ai

Competitive landscape

Edge: Can improve target-task reasoning by changing inference-time search and orchestration without owning a large model pretraining stack.

Better task scores can come at increased inference-time computation. Its efficiency claims should be tested on total cost per solved task. The business is complementary to NCM and diffusion models, not directly competing in token generation.

Market / IPO context

Editorial / static context — not a live quote.

FAQs about Poetiq

Practical answers founders, operators, and investors typically search for.

Its published approach uses a recursive meta-system around existing models rather than relying solely on full-model retraining.
Poetiq publicly named participants but did not explicitly disclose a lead investor in its January 2026 announcement.

By Venture Capital Tracker

Last updated:

Editorial note: AI tools assisted with research, structure, or drafting. Venture Capital Tracker retains human editorial responsibility for factual accuracy, relevance, and source quality before publication.