Legal terms

Terms of Service

Last updated: June 3, 2026
Use Internal research and analysis are allowed.
Limits No resale, redistribution, scraping, or key sharing.
Data Source facts are served with point-in-time provenance.
Billing Metered runs consume credits when compute is used.

The service

Priorish provides point-in-time knowledge search via a web application and REST API. You query historical financial data with temporal constraints, and we return what was knowable at the timestamp you specify.

By accessing or using Priorish — including the Demo tier, which requires no account — you agree to these Terms, including the acceptable-use and data usage rights below.

Accounts

You need an account to use the Free, Pro, or Institutional tiers. The Demo tier is available without an account. You're responsible for keeping your API key secure. If you suspect your key is compromised, revoke it immediately in the dashboard.

Acceptable use

Use Priorish for lawful purposes. Don't:

  • Attempt to circumvent rate limits or authentication, including using multiple accounts or rotating keys to exceed tier limits.
  • Resell, redistribute, or sublicense API access or query results without an Institutional agreement.
  • Use automated tools to scrape, mirror, or bulk-download the database.
  • Share API keys with third parties or proxy access to unauthorized users.
  • Interfere with the service's operation or other users' access.

Data usage rights

Query results from Priorish are licensed for your internal use only. You may use Priorish data in your own reports, analysis, models, and research. You may not:

  • Redistribute, resell, or sublicense raw query results or bulk data obtained through the API.
  • Build a competing product or service using data obtained from Priorish.
  • Systematically cache or store query results to circumvent rate limits or avoid future queries.

Institutional customers may distribute Priorish data within their organization per their Institutional agreement.

Intellectual property

Source data from government databases (SEC EDGAR, Federal Reserve, FRED/ALFRED) is in the public domain. However, Priorish's enrichment layer — including entity resolution, temporal knowledge graph, derived facts, provenance chains, and cross-document analysis — is our proprietary intellectual property. The Priorish service, API, SDK, and documentation are also our intellectual property.

You retain ownership of any original analysis, reports, or derivative works you create using Priorish data, provided you comply with the data usage restrictions above.

Data accuracy

Priorish aggregates data from public sources (SEC EDGAR, Federal Reserve, FRED/ALFRED). We store and serve this data as-is, with source and provenance metadata. We do not guarantee the accuracy of source data. Point-in-time queries reflect captured source material at the specified cutoff, including errors in the original publication.

Priorish is an information tool, not financial advice. Do not make investment decisions based solely on Priorish output.

Temporal methodology and scope

Plain English: Priorish applies a historical cutoff to returned evidence. Ranking and extraction are described separately below so strict backtest users know where extra controls may be needed.

Priorish's point-in-time methodology bounds returned evidence at an as_of timestamp. Evidence-type-specific predicates apply known_at, retraction or supersession boundaries, and source-episode visibility where applicable. These predicates, together with source provenance, are the primary temporal integrity mechanism of the system.

There are two areas where strict temporal purity is not guaranteed and users running rigorous backtests should be aware of:

  • Retrieval ranking. Semantic similarity uses dense embedding models whose training data post-dates much of the historical corpus. This can introduce look-ahead bias in the ranking of retrieved documents (a 2024-trained model encodes associations — e.g., corporate acquisitions, ticker changes, macroeconomic regimes — that were not known at historical timestamps). The returned evidence is PIT-bounded by as_of; the order in which it is ranked may be influenced by post-cutoff world knowledge. Exact-term retrieval (BM25 / tsvector) is not affected.
  • LLM extraction. Entity and relationship extraction runs through a current-generation LLM. We scope prompts to facts explicitly stated in the source document, but the model's pretraining includes information about outcomes that occurred after the document was published. This can marginally affect which extracted facts are surfaced or how entities are disambiguated. Raw document text and structured source data are unaffected.

For most research, analysis, and LLM-grounding use cases these effects are small and the content-level temporal filter is the property that matters. For strict retrieval-strategy backtests or algorithmic trading research where ranking must also be temporally pure, treat Priorish results as a content source and apply your own era-appropriate ranking layer.

Institutional: if you require end-to-end temporal purity — era-appropriate embedding models trained on pre-cutoff corpora, deterministic or vintage-model extraction, strict-mode retrieval (BM25 / tsvector only), or custom models trained to your cutoff — we can build and operate this for you under an Institutional agreement. Email sales@priori.sh.

Pricing and payment

  • Demo: Free, no account required. Shared community pool, structured search only.
  • Free: Free with account. 200 queries/day, 5 NL queries/hour, 1 API key.
  • Founding Builder ($29/mo, limited launch tier): 5,000 queries/day, 60 NL queries/hour, 5 API keys, commercial-use rights, batch search, document API, MCP tool access, and bulk replay export.
  • Pro ($49/mo): The same API features, plus 18 monthly Deep Research credits. Higher Pro tiers include larger monthly Deep Research credit allotments, replay-export rows, and rate limits. Available plans and their allotments are shown at checkout.
  • Institutional: Custom pricing, unlimited queries, bulk export, full provenance packs, dedicated support with SLA.

Pro subscriptions are billed monthly via our payment processor (Stripe, through Clerk). You can cancel anytime. Cancellation takes effect at the end of the current billing period. On downgrade, access to Pro features is revoked at the end of the billing period.

Deep Research and export metering. Deep Research runs and replay-slice exports are metered. Each tier includes a monthly allotment (Deep Research credits and export rows) that resets at the start of each billing period and does not roll over. Once an allotment is spent, that surface is paused until the next billing period or you upgrade to a higher tier. A Deep Research run draws credits from your allotment in proportion to its actual compute cost; an export draws from your allotment by its exact row count, both shown before the operation runs.

Refunds. Subscription fees are generally non-refundable. Deep Research credits and export rows that have already been consumed are never refundable, because each draws real third-party compute at the time of use. Monthly tier allotments are not redeemable for cash and expire at the end of each billing period.

API keys

API keys are issued per account and are subject to the rate limits of your tier. Keys are non-transferable. Sharing keys or proxying access through your account is a violation of these terms. We reserve the right to revoke keys that violate these terms.

Availability

We aim for high availability but at this time don't provide SLAs. We'll notify registered users of planned maintenance via email when possible.

Liability

Priorish is provided "as is" without warranty. We are not liable for decisions made using data from the service, losses from service interruptions, or inaccuracies in source data.

Termination

You can close your account at any time. We can terminate accounts that violate these terms. On termination, your API keys are revoked and account data is deleted per our privacy policy.

Changes

We may update these terms. Material changes will be announced via email to registered users at least 14 days before taking effect.

Trademarks

Gemini is a trademark of Google LLC. PRIORI.SH is not affiliated with, endorsed by, or sponsored by Google. Where model outputs are shown (for example, on our homepage), they are generated via Google's Gemini API and are labeled for illustration. All other product and company names are the trademarks of their respective owners; nominative references do not imply endorsement.

Contact

Questions? Email legal@priori.sh. Billing or account support: support@priori.sh · +1 916 678 0768.

PRIORI.SH
Home Demo API Docs Terms Privacy