⚠️ This due diligence report is generated from publicly available information and AI analysis. It should supplement, not replace, human judgment, reference calls, and direct founder interactions.

Deal Screening Card: Taku

Date: 2026-03-23 Analyst: DD Memo Writer Stage: Early Seed Sector: AI Productivity / Agentic Workflows

One-Liner

Taku is an AI-native "Agentic Canvas" and personal OS designed for creators and journalists, utilizing Hierarchical Reasoning Models (HRM) to automate complex research and content workflows.

Scorecard

Dimension	Score	Signal
Team	5/10	🟡
Market	7/10	🟢
Competition	3/10	🔴
Product/Tech	4/10	🟡
Financials	4/10	🟡
Risk	8/10	🔴
Overall	5.1/10	🟡

Scoring: 🟢 >= 7 | 🟡 4-6 | 🔴 <= 3 Risk score inverted: 🟢 <= 3 (low risk) | 🔴 >= 7 (high risk)

Key Strengths (Top 3)

1. Founder Pedigree (Tier B): Austin Zheng has a proven track record of attracting significant capital, previously leading Sapient Intelligence to a $200M valuation.
2. Niche Focus (Tier B): By targeting "creators and journalists," Taku may find an initial wedge in high-intensity information-gathering workflows that generic LLMs struggle with.
3. Public Validation (Tier A): The product was demoed at the "1st Burial of AI Hype Competition" (Jan 2026), suggesting a functional prototype exists that can withstand public scrutiny.

Key Concerns (Top 3)

1. Stability & Commitment (Tier C): The founder exited his previous startup, Sapient Intelligence, less than a year after raising a massive $22M seed round. This is a significant "Guilty until proven innocent" red flag regarding long-term commitment.
2. The "Technical Gap" (Tier B): Taku claims to build "Next-gen OS" and "Hierarchical Reasoning Models," yet there is no identified technical co-founder or CTO. Austin is recognized for strategy/product, not deep-tech architecture.
3. Hyper-Competitive Environment (Tier A): Taku is competing directly against better-funded unicorns (Manus, Flowith) and incumbents with platform advantages (OpenAI, Apple Intelligence), creating a high risk of being "featured" or out-scaled.

Verdict

Recommendation: Deeper Look Reasoning: While the founder's ability to raise capital is undeniable, the early departure from his previous venture and the lack of a technical "heavy hitter" in an OS-level project are major risks. We move to a deeper look solely to investigate the circumstances of the Sapient exit and the actual performance of the HRM tech.

Questions for Founder Meeting

1. The Sapient Exit: Can you explain the specific reasons for leaving Sapient Intelligence so shortly after a $22M raise? What is your current relationship with those investors?
2. Technical Moat: Who is the architect behind the Hierarchical Reasoning Model? How does Taku's "intent-based" coding differ from the autonomous agent capabilities being released by OpenAI/Manus?
3. Platform Risk: How does Taku remain relevant once Apple Intelligence integrates agentic canvas features directly into the MacOS/iPadOS ecosystem?
4. Hiring: Why have you opted not to name a technical co-founder thus far, given the "OS-level" ambition of the project?

Source Hierarchy & Reliability

• Tier A: Jan 2026 Demo at "Burial of AI Hype" (Observable functionality).
• Tier B: Founder claims regarding HRM and "vibe coding" (Unverified technical claims).
• Tier C: Media reports on Sapient Intelligence funding/exit (Filtered narrative).

Most Critical Finding: The founder's abandonment of a high-valuation, well-funded startup (Sapient) within a year creates a "Red Signal" for team stability and fiduciary reliability.