← Briefing history

Two developments this cycle sharpen the map's edges.

Read-only snapshot of Autonomous research competitive landscape

Sep 26, 2026 · 3 findings · closed 1 thread · ran 2m 34s

Autonomous Research Tools: Market Map Digest

TL;DR

Two developments this cycle sharpen the map's edges. SciencePro's pricing question — open since launch — resolved as "no public price anywhere," confirming China's champion is running a sales-led, institution-first game rather than competing for self-serve subscribers. And Firecrawl's Alexandria, launched with a $75M Series B, is building the first priced, brokered supply layer for what research agents read — meaning the data underneath every tool on this map is about to have a visible per-call price.

The supply side of research is becoming a priced layer — and it just got its first funded broker

The data that research agents consume is moving from free externality to auditable, per-call infrastructure. Firecrawl — the crawl/extract backend used by 1.5M+ builders — launched Alexandria on September 22 alongside a $75M Series B led by Smash Capital, positioning it as a brokered marketplace connecting research agents to data providers (compensated-knowledge-supply-agentssearchenginewatch.comfirecrawl.dev). The pricing interface is genuinely new: discovery is free, retrieval is priced per tool, and each catalog listing exposes provider, inputs, expected response, and price.

"The documentation separates discovery from retrieval. Finding tools is free; running one costs the credits listed for that tool. Each listing describes its provider, required inputs, expected response and price." — Search Engine Watch

The Wikimedia Enterprise deal is the template for what actually gets sold when content is free: Firecrawl was handling 2–3 million Wikipedia requests a month by scraping pages with headless browsers, and replaced that with direct API access — paying for structure, maintenance, and guaranteed delivery, not the material itself (compensated-knowledge-supply-agentssearchenginewatch.comfirecrawl.dev). Note the enterprise plumbing appearing from day one: some providers require an organization administrator to accept their terms before access is allowed, which means licensed retrieval is emerging as the clean alternative to contested autonomous crawling. For the map, this creates a new layer between the tools and their sources — an orchestrator like Hey, Lefty that plugs into Alexandria gets per-call costs and provider provenance as first-class primitives rather than legal gray areas.

What to watch: Whether Alexandria's self-service contributor payouts open — provider sign-up is still a waitlist, and until it does, the economics only run one direction (agents pay providers), with contributor revenue splits unpublished (Search Engine Watch).

SciencePro's pricing resolved: China's champion is sales-led, not self-serve

The open question from last cycle — what does SciencePro's Plus tier cost, and is it available outside China — has an answer, and the answer is no public numbers exist. A hunt across the Kyodo/PRNewswire English launch release, scienceone.cn, and scienceone.com.cn confirms the three service models (individual free/Plus, institutional SaaS, on-premises deployment) but publishes no price for any of them (wenge-sciencepro-research-platformenglish.kyodonews.netscienceone.cnwenge.com). The only forward-looking commercial statement is that Wenge will "pursue diversified commercialisation pathways."

"To meet the different needs of individual researchers and institutions, SciencePro offers three service models: individual subscriptions, institutional SaaS and on-premises deployment." — Kyodo/PRNewswire release

This slots SciencePro into the pricing segmentation as the mirror image of transparent self-serve tools: free individual tier plus unpublished paid tiers behind institutional sales (wenge-sciencepro-research-platformenglish.kyodonews.netscienceone.cnwenge.com). International ambition is signaled — English press distribution and English product pages — but not committed, and the distribution mechanism remains the state-coordinated channel: a strategic cooperation agreement with the CAS Talent Exchange and Development Center pushing into university research, plus a tripartite deal for vocational education. The read for the map: SciencePro isn't competing for the individual-subscriber dollar that Perplexity, Elicit, and Undermind fight over. It's capturing institutions through administrative channels, and pricing opacity is a feature of that motion, not an oversight.

What to watch: Any published Plus-tier price or international availability commitment — either would signal a pivot from institutional capture toward competing in the open market.

What surprised us

  • The pricing opacity is the strategy. You'd assume a launch distributed via international newswire would come with at least a headline price. It doesn't — and paired with ministry-adjacent distribution deals, the silence reads as deliberate: when the buyer is an institution negotiating through a state coordinator, published self-serve prices only anchor negotiations downward (wenge-sciencepro-research-platformenglish.kyodonews.netscienceone.cnwenge.com).
  • The supply layer got funded before the payout side exists. Alexandria raised $75M on a marketplace where prospective providers cannot yet calculate what they'd earn — "Alexandria's public launch materials do not publish payout rates or a revenue split for new contributors" (Search Engine Watch). Investors are betting on the demand side (agents needing licensed data) being urgent enough to carry the supply side into existence.
  • Charging for free content has a working precedent. The Wikimedia deal shows providers can monetize structure and delivery guarantees even when the underlying material is freely licensed — which means "but Wikipedia is free" is not an argument against the priced supply layer (compensated-knowledge-supply-agentssearchenginewatch.comfirecrawl.dev).
  • Google is running the incumbent version of the same idea. Search Engine Watch points to Google's "AI contribution pilot" for publishers as a parallel to Alexandria — compensated agent-era data supply is appearing in both startup and incumbent form simultaneously, which strengthens the case that it's a durable stack layer rather than one company's experiment.

Open threads worth a vote

  • The SciencePro pricing thread is resolved — no public price exists for any tier, institutional sales are sales-led, and international availability is signaled but not committed (threads). If you want a new hunt opened, the natural successor is tracking Alexandria's payout-system launch.

Findings from this cycle

Current topic brief

Shown for context; the brief may have changed since this cycle ran.

What is the market for autonomous or AI research tools? There's gemini deep research, google scholar, perplexity. Sacra is another research platform. What do all of these tools do? What are their features? Their value prop? Their core technology? Their data and where does it come from? Who do they sell to? what is the pricing/business model? Help me build a market map to see where Hey, Lefty fits and we should position it.