Investing in Drafted
The following excerpt is from Allied’s original deal memo to investors. To review investment opportunities in full, please consider joining the syndicate at Allied.vc/join
Problem
As artificial intelligence transitions from simple chat interfaces to autonomous, long-horizon agents, developers face a critical technical bottleneck: the human data constraint.
While synthetic data can accelerate model development within known boundaries, recursively training models on synthetic outputs causes model collapse, degrading performance on rare behaviors and critical edge cases. Consequently, high-reasoning human feedback and validated real-world trajectories remain the essential upstream assets for model training and post-deployment evaluation (source: OpenAI).
The current market for human annotation is deeply fragmented:
Low-Cost Crowdsourced Networks: Existing providers rely on anonymous, low-wage gig workers. This leads to severe quality issues, with up to 40% of outputs rejected due to contributors using unauthorized AI to expedite their tasks.
Scarce Expert Networks: Platforms matching specialist talent are prohibitively expensive ($125–$200/hour) and suffer from quality resets with every project, limiting scalability.
Enter Drafted: the default human data and evaluation layer for the next generation of AI systems.
Solution
Drafted Labs is building the default human data and evaluation layer for the next generation of AI systems.
Sourcing exclusively from accredited universities (e.g., USC, UChicago, Georgetown, and UMiami), Drafted deploys a vetted, STEM-heavy student workforce to generate high-reasoning RLHF (Reinforcement Learning from Human Feedback), model evaluations, and complex computer-use trajectories.
By capturing unique behavioral metadata, including keystroke counts, scroll depth, and tab switches, and implementing rigorous quality assurance filters, Drafted delivers a 20% average increase in data quality over legacy contractor baselines.
Why Now?
1. The Synthetic Ceiling & Model Collapse
Recursive training on synthetic data has hit a technical wall. This self-referential training triggers model collapse, erasing coverage of essential edge cases and tail distributions.
To advance capabilities, developers must inject high-signal human ground truth. Because synthetic outputs are derivative, high-quality human data has become the ultimate upstream constraint.
2. The Shift to Agentic Workflows
AI is moving from simple text generation to autonomous, long-horizon agents that control browsers and software environments. Training and validating these multi-step systems (e.g., computer-use trajectories) requires real human demonstration data and precise behavioral telemetry.
3. Enterprise Demand Scaling
While foundational model training is concentrated, agent deployment expands the buyer base to every enterprise using AI. With corporate adoption rising, enterprise agent evaluation is emerging as a permanent, recurring infrastructure spend, projected to reach $13B by 2028 (PWC).
4. Dawn of the Third Wave of Human Data
The market is consolidating around a new quality-and-scale paradigm:
Wave 1 (Quantity):
Offshore, crowdsourced legacy providers suffer from 40% rejection rates due to anonymous annotators using AI to cheat.
Wave 2 (Specialization):
Elite expert-matching networks are prohibitively expensive ($125–$200/hr).
Wave 3 (Compound Scale & Quality):
Pioneered by Drafted. By deploying a verified, cost-efficient university ecosystem, Drafted delivers a 20% average quality lift across core benchmarks at a highly scalable, mid-market cost structure.
As hyperscalers, frontier labs, and enterprise customers look for distinctive data to train the next generation of AI models, we believe Drafted is uniquely positioned to build the leading supply chain for university-native human intelligence. Their university network addresses the critical risk of synthetic model collapse in an agentic training and evaluation market we believe can exceed $30B by 2030.
Certain information contained in this post has been obtained from third-party sources, including from portfolio companies of Allied Venture Partners. While taken from sources believed to be reliable, AlliedVP has not independently verified such information and makes no representations about the enduring accuracy of the information or its appropriateness for a given situation. In addition, this content may include third-party advertisements; AlliedVP has not reviewed such advertisements and does not endorse any advertising content contained therein.
This content is provided for informational purposes only, and should not be relied upon as legal, business, investment, or tax advice. You should consult your own advisers as to those matters. References to any securities or digital assets are for illustrative purposes only, and do not constitute an investment recommendation or offer to provide investment advisory services. Furthermore, this content is not directed at nor intended for use by any investors or prospective investors, and may not under any circumstances be relied upon when making a decision to invest in any fund managed by AlliedVP. (An offering to invest in an AlliedVP fund will be made only by the private placement memorandum, subscription agreement, and other relevant documentation of any such fund and should be read in their entirety.) Any investments or portfolio companies mentioned, referred to, or described are not representative of all investments in vehicles managed by AlliedVP, and there can be no assurance that the investments will be profitable or that other investments made in the future will have similar characteristics or results.
Charts and graphs provided within are for informational purposes solely and should not be relied upon when making any investment decision. Past performance is not indicative of future results. The content speaks only as of the date indicated. Any projections, estimates, forecasts, targets, prospects, and/or opinions expressed in these materials are subject to change without notice and may differ or be contrary to opinions expressed by others.