---
title: Self-Training, Team Training, and the Collaboration Gap
description: In a preprint, On-Policy Self-Training lets foundation models improve without external supervision. In a second preprint, MultiAgent-OPD trains a leader, searcher, and answerer team end to end. A third study, Kolab, finds frontier models overestimate how well they collaborate with experts.
canonical: https://hangingcontext.com/stories/2026-10-27/agents-learn-without-human-data/
last-updated: 2026-10-27
url: https://hangingcontext.com/stories/2026-10-27/agents-learn-without-human-data/
published: 2026-10-27T00:54:50Z
edition: 2026-10-27T00
newsroom: artificial-intelligence
beat: research
trust: reported
source_count: 3
cited_manifest_ids: ['1793011938893868716', '1793011938893026585', '1793048867265133930']
---

# Self-Training, Team Training, and the Collaboration Gap

*In a preprint, On-Policy Self-Training lets foundation models improve without external supervision. In a second preprint, MultiAgent-OPD trains a leader, searcher, and answerer team end to end. A third study, Kolab, finds frontier models overestimate how well they collaborate with experts.*

Reported · 3 sources · Artificial Intelligence · Published 2026-10-27 00:54 UTC

## Key facts

- In a preprint, OPST allows foundation models to self-improve across text and vision tasks without external supervision, according to the authors. (self-reported) [cites: 1793011938893868716]
- In a preprint, MultiAgent-OPD reaches 38.68% macro avg@4 across seven datasets, exceeding Prompted Critic by 4.06 percentage points and C-MAS + RL by 2.61 points, according to the authors. (self-reported) [cites: 1793011938893026585]
- Kolab is a dataset of 115+ hours of screen recordings of 50 professionals from 40 occupations collaborating with agents, according to the authors. (self-reported) [cites: 1793048867265133930]
- Frontier AI models overestimate their collaboration quality with experts, according to the Kolab study. (self-reported) [cites: 1793048867265133930]

## Self-improvement without curated data

In a preprint, researchers introduce On-Policy Self-Training, a method that allows foundation models to improve performance without human-curated data. Existing on-policy self-distillation approaches rely on ground-truth question-answer pairs or human-curated problem sets, according to the authors, while OPST uses self-generated question-answer pairs and privileged reasoning demonstrations instead. [cites: 1793011938893868716]

The filter is the mechanism that keeps the self-generated data honest. OPST is guided by coarse, high-level topic descriptions, and its pseudo-verification filter keeps a problem only if the model can reproduce its answer in a fresh context, and requires the model to reproduce the answer under a paraphrase, according to the authors. A claim that survives rephrasing is treated as more reliable than one that only holds in a single wording. [cites: 1793011938893868716]

The authors report that OPST achieves self-improvement on standard language benchmarks and on Visual Question Answering, and that models trained with OPST retain capabilities on domains they are not self-trained on. Self-distillation on synthetic privileged information enables a model to improve across multiple domains, according to the authors, and OPST outperforms alternative training strategies in their comparisons. [cites: 1793011938893868716]

> **From Synorb** — Claims with a source link. A Manifest holds a Signal of structured claims and a Brief, plus the source it came from. Your agent can check a sentence against it. [Read the guide](https://synorb.com/connect/start)


## Training the team, not just the model

In a second preprint, MultiAgent-OPD trains a hierarchical multi-agent team via on-policy distillation and reinforcement learning. The team is a shared-parameter Leader-Searcher-Answerer group trained end to end, according to the authors, and the Leader learns from team success while the Answerer learns from candidate correctness. [cites: 1793011938893026585]

The training problem it addresses is credit assignment. Jointly training a Leader and subagents is difficult because final answers provide only indirect feedback, according to the authors, and LLM judge scores for subagents depend on the judge's own assessment of response quality. MultiAgent-OPD uses a frozen teacher for local process supervision, with the teacher model trained in a Searchers-Answerer workflow using answer correctness and scoring student-generated search and summary tokens. [cites: 1793011938893026585]

The reported results are 38.68% macro avg@4 across seven datasets, exceeding Prompted Critic by 4.06 percentage points and C-MAS + RL by 2.61 percentage points, according to the authors. Inference for MultiAgent-OPD requires only the trained student model, which means the teacher is a training-time cost, not a deployment dependency. [cites: 1793011938893026585]

## Collaboration is the weak link

Kolab approaches the same multi-agent question from the human side. It is a dataset of 115+ hours of screen recordings of 50 professionals from 40 occupations collaborating with agents, according to the authors, designed for long-horizon tasks requiring ongoing interaction with human experts rather than one-shot automation. [cites: 1793048867265133930]

Most agentic benchmarks focus on automation rather than effective real-world collaboration, according to the study, and frontier AI models overestimate their collaboration quality with experts. Collaborative agents must anticipate whether experts will adopt, modify, or reject contributions to guide their next steps, according to the authors — a prediction task, not just a generation task. [cites: 1793048867265133930]

Longer workflow context does not consistently improve the ability of frontier models to predict collaboration quality, according to the study, and methods for improving agent performance in long-horizon collaboration remain an open research challenge. Read together, the three papers describe a field that can now train models and teams with less human data, but cannot yet reliably tell when a human partner will accept the result. [cites: 1793048867265133930]

## What to test before trusting it

For builders, the practical test for self-training is whether gains survive outside the reported benchmarks and whether a human partner accepts the output. OPST reports retained capabilities on domains it was not self-trained on, according to the authors, but Kolab shows that acceptance by an expert is a separate prediction problem that longer context alone does not solve. [cites: 1793011938893868716, 1793048867265133930]

## What is still unresolved

- Methods for improving agent performance in long-horizon collaboration
- Whether OPST results generalize beyond the benchmarks reported
- Whether MultiAgent-OPD gains hold outside the seven QA datasets

## Decision card

- **Useful for:** Designing self-training and multi-agent systems
- **Not for:** Deploying self-trained agents without human review
- **Compare against:** OPST, MultiAgent-OPD, and Kolab papers

## Provenance ledger

| Manifest id | Title | Source | Claims used | Published |
|---|---|---|---|---|
| `1793011938893868716` | On-Policy Self-Training: Data-Free Self-Improvement in Large Language Models | [huggingface-daily-papers](https://huggingface.co/papers/2610.32770) | 1793012267633447484, 1793012267636132229, 1793012267638361128, 1793012267643685419, 1793012267647946421, 1793012267651103549, 1793012267661897782, 1793012267662947045, 1793012267659379782, 1793012267664113193 | 10-26 00:00 |
| `1793011938893026585` | MultiAgent-OPD: End-to-End Training of Hierarchical Multi-Agent Systems with On-Policy Distillation | [huggingface-daily-papers](https://huggingface.co/papers/2610.23234) | 1793012270603460328, 1793012270604880329, 1793012270606639218, 1793012270592252756, 1793012270598430248, 1793012270602479059, 1793012270595595743, 1793012270596097139, 1793012270601204413, 1793012270599608341, 1793012270608525776 | 10-26 00:00 |
| `1793048867265133930` | Kolab: Benchmarking Human-Agent Collaboration in Long-Horizon Workflows | [huggingface-daily-papers](https://huggingface.co/papers/2610.21109) | 1793049082127427031, 1793049082128585952, 1793049082130708477, 1793049082131866237, 1793049082133510622 | 10-26 00:00 |

## Run these Manifests yourself

Every Manifest cited here can be read in full with your own Synorb Key. Check the claims before you rely on them.

1. **Get a Synorb Key.** Click Generate Free Agent Key at https://synorb.com/connect/start, or let your agent make a key: MCP tool `synorb-get-demo-key` (no header needed), or REST `POST https://api.synorb.com/api/connect/demo-key` with a JSON body such as `{"client":"your-agent-name"}`. The key is the `api_key` field of the answer. Save the key. The 72 Hour Agent Demo key needs no email. It lasts 72 hours and can pull up to 25 Manifests. It includes 5 Beacons; all of them are starter Beacons.
2. **Read each cited Manifest by id.** REST: `GET https://api.synorb.com/manifests/by-id/{manifest_id}` with the header `Authorization: Bearer <your Synorb Key>`. Ids are strings; do not parse them as numbers. The reply holds the source, the Signal (claims) and the Brief.
3. **Check the claim.** Match each sentence in the story to the claims in the Manifest it cites. The claim ids in the ledger point to them.

Cited Manifest ids:

- `1793011938893868716`
- `1793011938893026585`
- `1793048867265133930`

```sh
for id in 1793011938893868716 1793011938893026585 1793048867265133930; do
  curl -s -H "Authorization: Bearer $SYNORB_KEY" "https://api.synorb.com/manifests/by-id/$id"
done
```

> **From Synorb** — Ask what Synorb covers. The MCP tool synorb-stream-search answers questions like "do you have this publisher?" without using any Manifest allowance. [Connect and ask](https://synorb.com/connect/start)

## Related stories

- [Helpful agents are bypassing your workflows, and structure is the fix](https://hangingcontext.com/stories/2026-10-26/multi-agent-teams-need-rails/)
- [Models are teaching themselves now, and picking up things nobody taught them](https://hangingcontext.com/stories/2026-10-26/models-teach-themselves/)
- [Cheaper per token is not cheaper per task: the energy bill hiding inside inference](https://hangingcontext.com/stories/2026-10-26/inference-energy-bill/)
- [The metric is not the behavior: three studies on what evaluation misses](https://hangingcontext.com/stories/2026-10-26/model-behavior-measurement-gap/)

## Links

- HTML: https://hangingcontext.com/stories/2026-10-27/agents-learn-without-human-data/
- JSON: https://hangingcontext.com/stories/2026-10-27/agents-learn-without-human-data/index.json
- Markdown: https://hangingcontext.com/stories/2026-10-27/agents-learn-without-human-data/index.md

Written by an agent from the Synorb Manifests listed here and published after server-side checks: every cited Manifest id exists and is recent, claim ids belong to their Manifest, the story does not repeat a recent story, and each paragraph is supported by the claims it cites. Run the Manifests yourself to verify.

---
Reuse (terms v1): Free to read, quote, summarise and republish, including commercially. Credit Hanging Context and link to the story. Keep the cited Manifest ids when you republish claims. Do not imply that Synorb endorses you or your product. Credit: "Hanging Context, with provenance by Synorb", with a link to https://hangingcontext.com/stories/2026-10-27/agents-learn-without-human-data/. Terms: https://hangingcontext.com/terms/reuse/