研究内容为英文. 本页的研究内容（标题、简报、来源）以英文撰写，未经机器翻译。只有界面已翻译。

新想法Education

LearnHarness-powered hosted tutor

A hosted, no-API-key web version of the open-source LearnHarness tutor (measure → plan → teach → practice, with a personal knowledge graph), in Portuguese and Spanish.

2周前 · 评分于 2026年10月25日 · ID HBV-261025-14

暂定标题： LearnHarness-powered hosted tutor (working name) — no API key, Portuguese and Spanish, for students learning hard subjects

摘要

简报假设估算

Benchmark: amosblomqvist/learnharness — total 1,290 stars (GitHub API, fetched 2026-10-25 09:16 BRT; pending browser confirm); created 2026-10-16 (age 9 days); 24h (24 Oct UTC) +906; 7d +1,035 (147.9/day) vs prior 7d +3 = +34400% (accelerating); 3d +1,019 vs +11 = +9164% (accelerating); since creation 143.3/day; peak day 2026-10-24 (+906), 0 days since peak; 25 Oct partial +259; license AGPL-3.0; forks 215.…

阅读其余部分

Method: daily new stars from OSS Insight analyze-recent-stars (UTC days) and total from the GitHub API (gh api repos/amosblomqvist/learnharness), fetched 2026-10-25 09:16 BRT in the rerun; total appended to /workspace/hbv_star_snapshots.csv. 24h = 24 Oct UTC; 7d = 18–24 Oct vs 11–17 Oct; 3d = 22–24 Oct vs 19–21 Oct; label: >+10% accelerating, <−10% decelerating. No browser: total is 'API value, pending browser confirm'; nothing estimated. Existence check: Checked 25 Oct 2026: GitHub search "learnharness" = 12 repos, all forks/clones with 0–1 stars (wrapper count: 0 real products); maintainer site learnharness.ai offers desktop downloads; README: "If we ever charge for anything, it will be for convenience — managed keys, hosting". Analogous tutors: ChatGPT Study Mode, Gemini Guided Learning, Khanmigo, NotebookLM. X: 51 posts/7d (Up from zero 0% Mon→Sat (full UTC days: Mon 19 0 → Sat 24 35); peak Sat 24 35; Sat 24 is 0% below peak; window 2026-10-18 12:18–12:30 UTC → 2026-10-25 ~12:15 UTC (X API 7-day limit, pulled 25 Oct 2026 ~09:14–09:17 BRT in the rerun; Sun 18 and Sun 25 are partial UTC days; full days Mon 19 → Sat 24)). Revenue hypothesis: ~$500–$4K/month at month 6 (HYPOTHESIS).

描述假设模型

A hosted, no-API-key web version of the open-source LearnHarness tutor (measure → plan → teach → practice, with a personal knowledge graph), in Portuguese and Spanish. Who: University and pre-university students in Brazil and LatAm (ENEM, vestibular, engineering calculus) and self-learners who cannot set up API keys.…

阅读其余部分

Pain: LearnHarness is a desktop app that needs your own provider API keys; most students cannot or will not set that up, and it is English-first. Monetization: Freemium with monthly study-minute plans (managed model costs); school/cursinho self-serve seats (hypothesis). Build: ~2–3 agent-weeks (host the AGPL app, managed keys, PT/ES prompts, billing); stack: Fork of LearnHarness (AGPL-3.0, source published) on Vercel; Supabase; OpenRouter managed keys with per-user caps; Stripe/Pix. GTM: SEO/GEO per topic ("entender derivada passo a passo"), shareable knowledge-graph images, student Discord/WhatsApp groups.

目标客户

University and pre-university students in Brazil and LatAm (ENEM, vestibular, engineering calculus) and self-learners who cannot set up API keys.

痛点

LearnHarness is a desktop app that needs your own provider API keys; most students cannot or will not set that up, and it is English-first.

为何是这个分数

分数较低的原因： The maintainer is likely to launch the same hosted, managed-key version.

护城河假设

Weak: AGPL means competitors can copy. Local-language curriculum maps could help. Status: Hypothesis.

需求信号

X 上的帖子（7 天）

信号趋势

Up from zero 0% Mon→Sat (full UTC days: Mon 19 0 → Sat 24 35); peak Sat 24 35; Sat 24 is 0% below peak

网络需求已验证

GitHub wave (Verified OSS Insight daily stars). Note: the broader query "learn harness" is polluted by "harness engineering" courses and was not used.

市场与竞争对手

TAM

Not sized this run — Scope: Global excl. China. Method: No third-party report fetched. (Basis: n/a)

SAM假设

Not sized — Scope: Target languages/markets above. Method: HBV assumption. (Basis: Hypothesis)

SOM假设

~$6K–$48K ARR (hypothesis) — Scope: Month-6 run-rate. Method: From revenue hypothesis ×12. (Basis: Hypothesis)

竞争对手

ChatGPT Study Mode, Gemini Guided Learning, Khanmigo, NotebookLM, Brainly, Me Salva! (BR), Descomplica (BR).

商业模式

变现方式假设模型

Freemium with monthly study-minute plans (managed model costs); school/cursinho self-serve seats (hypothesis).

收入假设（每月）假设

~$500–$4K/month at month 6 (HYPOTHESIS).

待完成的任务

When I do not understand a hard topic, I want to be taught step by step from what I already know, so I can pass my exam.

HXC

Júlia, 18, vestibulanda in Recife (illustrative) · Struggles with physics and calculus. Buying signal: Uses ChatGPT for homework; no API key.

构建与 GTM

技术栈

Fork of LearnHarness (AGPL-3.0, source published) on Vercel; Supabase; OpenRouter managed keys with per-user caps; Stripe/Pix

GTM 渠道

SEO/GEO per topic ("entender derivada passo a passo"), shareable knowledge-graph images, student Discord/WhatsApp groups

风险

主要风险

The maintainer is likely to launch the same hosted, managed-key version.

全部风险已验证

Value: Supported — +1,035 stars in 7d, accelerating (Verified) | Usability: High — Maintainer hosting | Feasibility: Attention — AGPL | Business Viability: High — Crowded tutors

下一步验证

Do not build now. Watch whether stars keep accelerating for 7 more days and whether the maintainer launches hosting.

来源

github.com
learnharness.ai
api.ossinsight.io (analyze-recent-stars)
X API counts (window above)

ideanewgithub-wrappergithub-neweducationtutoringb2cbrazillatam

想在评分当天收到下一个想法吗？

分数、来源、风险和下一步，通过邮件发送。

ProductBots.ai

喜欢这个想法吗？用 ProductBots.ai 把它变成计划。

ProductBots.ai 是一个 Agentic Product OS：由智能体与你一起完成产品工作，从发现到计划。下载这个想法的简报，作为起点。

打开 ProductBots.ai

简报文件为英文。

预览简报

# Briefing: LearnHarness-powered hosted tutor (working name) — no API key, Portuguese and Spanish, for students learning hard subjects

Source: Startup Ideas HQ (HB Ventures), found 2026-10-25. ProductBots Opportunity Score: 58/100, evidence confidence 56%.
Evidence labels (Verified, Self-reported, Hypothesis) are kept as written by the researcher.

## Goal for ProductBots.ai
Turn this idea into a validated MVP plan and a first build. Start from the next validation step; do not assume unlabelled numbers are verified.

## What it is
A hosted, no-API-key web version of the open-source LearnHarness tutor (measure → plan → teach → practice, with a personal knowledge graph), in Portuguese and Spanish.

Benchmark: amosblomqvist/learnharness — total 1,290 stars (GitHub API, fetched 2026-10-25 09:16 BRT; pending browser confirm); created 2026-10-16 (age 9 days); 24h (24 Oct UTC) +906; 7d +1,035 (147.9/day) vs prior 7d +3 = +34400% (accelerating); 3d +1,019 vs +11 = +9164% (accelerating); since creation 143.3/day; peak day 2026-10-24 (+906), 0 days since peak; 25 Oct partial +259; license AGPL-3.0; forks 215. Method: daily new stars from OSS Insight analyze-recent-stars (UTC days) and total from the GitHub API (gh api repos/amosblomqvist/learnharness), fetched 2026-10-25 09:16 BRT in the rerun; total appended to /workspace/hbv_star_snapshots.csv. 24h = 24 Oct UTC; 7d = 18–24 Oct vs 11–17 Oct; 3d = 22–24 Oct vs 19–21 Oct; label: >+10% accelerating, <−10% decelerating. No browser: total is 'API value, pending browser confirm'; nothing estimated. Existence check: Checked 25 Oct 2026: GitHub search "learnharness" = 12 repos, all forks/clones with 0–1 stars (wrapper count: 0 real products); maintainer site learnharness.ai offers desktop downloads; README: "If we ever charge for anything, it will be for convenience — managed keys, hosting". Analogous tutors: ChatGPT Study Mode, Gemini Guided Learning, Khanmigo, NotebookLM. X: 51 posts/7d (Up from zero 0% Mon→Sat (full UTC days: Mon 19 0 → Sat 24 35); peak Sat 24 35; Sat 24 is 0% below peak; window 2026-10-18 12:18–12:30 UTC → 2026-10-25 ~12:15 UTC (X API 7-day limit, pulled 25 Oct 2026 ~09:14–09:17 BRT in the rerun; Sun 18 and Sun 25 are partial UTC days; full days Mon 19 → Sat 24)). Revenue hypothesis: ~$500–$4K/month at month 6 (HYPOTHESIS).

## Customer and pain
Target customer: University and pre-university students in Brazil and LatAm (ENEM, vestibular, engineering calculus) and self-learners who cannot set up API keys.
Pain: LearnHarness is a desktop app that needs your own provider API keys; most students cannot or will not set that up, and it is English-first.

## Market and competitors
TAM: Not sized this run — Scope: Global excl. China. Method: No third-party report fetched. (Basis: n/a)
SAM: Not sized — Scope: Target languages/markets above. Method: HBV assumption. (Basis: Hypothesis)
SOM: ~$6K–$48K ARR (hypothesis) — Scope: Month-6 run-rate. Method: From revenue hypothesis ×12. (Basis: Hypothesis)
Competitors: ChatGPT Study Mode, Gemini Guided Learning, Khanmigo, NotebookLM, Brainly, Me Salva! (BR), Descomplica (BR).
Geography: Brazil, Mexico, Colombia, Argentina; global English later

## Business model
Monetization: Freemium with monthly study-minute plans (managed model costs); school/cursinho self-serve seats (hypothesis).
Revenue hypothesis (monthly): ~$500–$4K/month at month 6 (HYPOTHESIS).

## Build and go-to-market
Build effort (agent-weeks): ~2–3 agent-weeks (host the AGPL app, managed keys, PT/ES prompts, billing)
Time to first revenue: ~4–6 weeks
Tech stack: Fork of LearnHarness (AGPL-3.0, source published) on Vercel; Supabase; OpenRouter managed keys with per-user caps; Stripe/Pix
GTM channels: SEO/GEO per topic ("entender derivada passo a passo"), shareable knowledge-graph images, student Discord/WhatsApp groups

## Risks and next step
Main risk: The maintainer is likely to launch the same hosted, managed-key version.
Next validation action: Do not build now. Watch whether stars keep accelerating for 7 more days and whether the maintainer launches hosting.

## Sources
https://github.com/amosblomqvist/learnharness; https://learnharness.ai; https://api.ossinsight.io (analyze-recent-stars); X API counts (window above)

新想法Education· 2周前
PaidToStudyAbroad-powered scholarship finder
A searchable, multilingual finder of fully funded study-abroad routes (government scholarships, tuition-free countries, paid PhDs, paid apprenticeships) built on the official-source list from the PaidToStudyAbroad repo.
60/100
可信度 57%
分数较低的原因： Crowded scholarship-directory market and unclear content license.
打开 ProductBots.ai →
新想法Education· 3周前
Anatomy-powered How It Works library
A free, SEO-first library of interactive "how it works" machines (watch movements, jet engines, the heart, seasons, Pix payments) built with the open-source Anatomy skill, in Portuguese, Spanish and English, with embeddable figures for teachers and paid custom figures on request.
60/100
可信度 40%
分数较低的原因： Generation cost and quality: good figures take from minutes to many hours of agent time and the skill admits Claude does not always see visual faults, so the library may grow slowly and need human review; the MIT lice…
打开 ProductBots.ai →

这是研究，不是投资建议。数字保留其发布时的标签。

摘要

简报○假设≈估算

描述○假设≈模型

目标客户

痛点

为何是这个分数

护城河○假设

需求信号

X 上的帖子（7 天）

信号趋势

网络需求✓已验证

市场与竞争对手

TAM

SAM○假设

SOM○假设