Autonomous Research Tools: Market Map Digest — The Price Ladders Go Public
TL;DR
The two biggest open pricing questions from last cycle both resolved: Perplexity published its full enterprise seat ladder ($40–$325/seat/month) and Lumivero's NVivo AI Cloud shipped with the first published individual prices for an audit-grade vertical research agent (perplexity-deep-research-consumer-agent, defensible-research-agent-verticals
). Meanwhile Google is about to gut its free tier, pushing the whole market up a paid ladder. The map now has public price points at nearly every layer — and the audit-grade premium is real, measured, and an order of magnitude above consumer.
The audit-grade vertical: "show your work" prices at $770/year
The vertical layer of the market just got its first verifiable willingness-to-pay number, and it's not close to consumer pricing. Lumivero's shop page now lists individual NVivo plans from $520 to $962.50/year, with the AI tier at $770/year — roughly 2.6–4x the $20/month consumer research tier (defensible-research-agent-verticals). The premium is explicitly sold on auditability, not model quality:
"Every AI-generated theme, code, or insight links back to source data through a citation-linked coding trail. Every AI write action requires human approval before it becomes part of your project, and a full audit trail is available so AI-assisted coding can be reviewed the same way manual coding is." — NVivo AI Cloud | Agentic AI qualitative research software
This resolves the open thread from last cycle on the NVivo launch: the Oct 6 individual launch happened on schedule, and the product page confirms "NVivo AI Cloud is available today for individual license subscriptions," with teams and institutions waitlisted until the new year (defensible-research-agent-verticals). That individual-first, institution-waitlisted motion is the mirror image of OpenEvidence's institution-first EHR embed — and for the market map, it proves the citation-linked evidence chain plus human approval gates is a product surface that commands vertical pricing, adoptable by a horizontal orchestrator without owning a corpus.
What to watch: Institutional NVivo pricing when the waitlist opens — if it lands well above $770/seat, the audit-grade premium compounds at the org tier.
Perplexity's enterprise ladder: the destination layer's prices are all public
The consumer deep-research incumbent's enterprise play is no longer opaque. Perplexity's own FAQ now publishes the seat list prices that were missing at the Computer for Enterprise launch:
"Enterprise Pro: Each seat is $40 per month or $400 per year (a discount of approximately 16.7% versus monthly subscriptions)" "Enterprise Max: Each seat is $325 per month or $3,250 per year" — Enterprise Pricing and Billing: Frequently Asked Questions — Perplexity Help Center
The full ladder runs free → Pro $20/mo → Enterprise Pro $40/seat → Enterprise Max $325/seat, with Max carrying 15,000 credits/month per seat versus Pro's 500 — a taste tier, not a workload (perplexity-deep-research-consumer-agent). Two structural details matter for the map: no self-serve trials at either enterprise tier (a sales-assisted motion at the top), and the seat-plus-shared-credit-pool metering that makes agent usage a budgeted opex line — the same billing shape as Sakana's per-run pricing and Parallel's per-request CPM. The agent-work meter, not the chat subscription, is becoming the industry's unit of billing.
What to watch: Whether the ~$1.00-per-100-credits overage rate reported by third parties gets confirmed on Perplexity's own pages — that number is what makes the credit pool actually comparable across vendors (perplexity-deep-research-consumer-agent).
Google narrows the free lane: one model, two days out
The free-incumbent layer of the map is about to thin out. Multiple outlets report that from October 9, free Gemini accounts get exactly one model — Flash-Lite — with AI Plus ($4.99/mo) losing Pro, AI Pro ($19.99/mo) gaining Deep Think, and AI Ultra holding at $99.99/mo ([watch] Gemini free tier cut to Flash-Lite-only takes effect (Oct 9)). The practical consequence: free-tier Deep Research quality and limits likely degrade immediately, forcing the price-sensitive segment of the research market up the $4.99 → $19.99 → $99.99 ladder. For positioning purposes, this matters because Google has been the marginal-cost pricing anchor in the category — if the free tier becomes a demo rather than a tool, the pressure that held consumer pricing down eases for everyone above it.
What to watch: Whether Deep Research on the free tier is degraded in practice or just gated — the first week of user reports after Oct 9 will tell.
What surprised us
- The audit premium is measured, not asserted. $770/year individual versus $20/month consumer is the first hard number showing that "defensible chain of evidence" — not speed, not model quality — carries vertical willingness-to-pay (defensible-research-agent-verticals
). Hey, Lefty can copy the pattern without owning a corpus.
- Perplexity's local-first hedge is capped, and the cap is the opportunity. Portable Computer runs 27B-class local models with cloud-only memory — meaning "frontier capability + local control" remains unoccupied even by the vendor hedging toward sovereignty (perplexity-deep-research-consumer-agent
).
- The enterprise tiers hide their own funnels. No self-serve trials at either Perplexity enterprise tier means the consumer funnel stops at Pro — a clean signal of where the sales-assisted motion begins, and a gap a self-serve-friendly challenger can exploit (perplexity-deep-research-consumer-agent
Open threads worth a vote
- [watch] Gemini free tier cut to Flash-Lite-only takes effect (Oct 9) — whether free-tier Deep Research actually degrades, and how fast users climb the paid ladder; steer follow-up on the topic page.