Variance
AI agents that work AML and fraud alerts from L1 to L3 review, leaving a timestamped, cited evidence trail regulators can read.
Variance is aimed at one job: getting an analyst-class evidence file out of an alert without a human doing the registry, sanctions and adverse-media legwork. The HK shell-company walkthrough on its site is the clearest demonstration — registry check, Mandarin name transliteration, OFAC alias overlap, ICIJ link, verdict with six cited artifacts. If your team runs high alert volume and the bottleneck is evidence gathering rather than detection, that workflow is worth a pilot. The 90%-of-evidence and 10x-cycle figures are vendor claims, so treat them as hypotheses with a defined success metric. Compare against Caseware, Lucinity and Unit21 for the investigative layer, and against your existing
Verified 5d ago · liveness 76/100 · cite: rightaichoice.com/tools/variance
- AML compliance teams at Fortune 500 and regulated financial institutions
- Fraud investigation units working high volumes of transaction alerts
- Banks and fintechs under AML supervision that need an auditable evidence trail
- Teams wanting to close L1 false positives faster by replicating manual investigative steps
- Small businesses with no AML or sanctions obligations
- Buyers looking for a general-purpose AI assistant rather than a compliance workflow tool
- Teams with very low case volumes where manual review is already cheap
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip Variance if your investigation bottleneck is detection quality rather than evidence gathering, or if you have no AML, sanctions or fraud supervision obligations that require a cited audit trail.
Variance sells to supervised financial institutions rather than to individuals or small teams, so the comparison set is enterprise AML platforms — Caseware, Lucinity, Unit21, NICE Actimize — not $20/mo AI subscriptions. Pricing is not published; budget conversations happen with sales and depend on alert volume, review levels covered and how many investigators the agents support. Expect the spend to be justified against analyst headcount, so model the pilot against hours saved per alert rather
In short
Variance — AI agents that work AML and fraud alerts from L1 to L3 review, leaving a timestamped, cited evidence trail regulators can read. Best for AML compliance teams at Fortune 500 and regulated financial institutions, Fraud investigation units working high volumes of transaction alerts, Banks and fintechs under AML supervision that need an auditable evidence trail. Contact Sales pricing.
What's new in Variance
Checked 5 days agoAcross the latest 1 update: 1 news mention.
What people actually say about Variance — is it worth it?
We scanned public community sources for Variance on Aug 12, 2026 and could not establish that the discussion we found is about this tool rather than something else sharing its name. Our own analysis of that scan says the posts were off-subject. Rather than publish a sentiment score built on the wrong subject, we publish nothing here and re-run the scan.
Viability Score
How well maintained and how widely used is Variance? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: October 2026
How we score →Key Features
- AI agents automate evidence collection for AML and fraud alerts
- Investigations run across L1, L2 and L3 review levels
- Ingests alerts from existing transaction monitoring systems
- Searches company registries such as the HK Companies Registry
- Cross-references sanctions lists including the OFAC SDN List
- Transliterates director names from Mandarin filings
- Extracts director names from scanned and handwritten documents
- Hunts aliases, alternate spellings and matching photos on the open web
- Links individuals to adverse media databases such as ICIJ Leaks
- Generates an audit-ready evidence trail with cited sources and timestamps
- Produces an agent verdict with evidence summary and artifact count
- Covers Know Your Customer, Know Your Business, Transaction Monitoring and CDD
- Turns unstructured evidence (documents, images, websites) into structured insight
- Ad-hoc feature generation for investigative analytics on ClickHouse
- Interpretability for AI flags via sparse autoencoders and confidence signals
About Variance
Variance is an AI agent platform for anti-money laundering and fraud investigations at regulated institutions. It sits on top of the transaction monitoring and alerting you already run rather than replacing it: agents take an alert and work the case the way an analyst would. The company's own walkthrough follows a $48k wire to a Hong Kong shell company, confirms the shell was registered three weeks earlier with no website or employees via the HK Companies Registry, pulls a Mandarin director name out of a PDF filing and transliterates it, checks the OFAC SDN List for a partial alias overlap, finds a cached LinkedIn with a matching photo and alternate spelling, and links that person to a 2023 ICIJ investigation on trade-based laundering. The agent closes with a verdict and six artifacts across four data sources, each step timestamped. Variance claims agents collect roughly 90% of the evidence on a case and cut investigative cycles by 10x; both numbers are the vendor's, not independently verified. Coverage spans L1 through L3 review levels with use-case pages for Know Your Customer, Know Your Business, Transaction Monitoring and Customer Due Diligence. The platform also converts unstructured evidence — scanned documents, images, handwritten forms and websites — into structured output. The company rebranded from Intrinsic to Variance in March 2025 and announced a $21M Series A in March 2026, so the roadmap is funded rather than speculative.
Behind the Verdict
Variance positions itself deliberately as an investigative layer, not a detection engine. That framing matters: it means you keep your current transaction monitoring stack and add agents on top, which lowers the switching cost considerably compared to ripping out a monitoring vendor. The clearest evidence of product thinking is the $48k-wire walkthrough on the homepage, which names real data sources — the HK Companies Registry, the OFAC SDN List, ICIJ Leaks — and shows the specific hard steps most tools skip: transliterating a Mandarin director name out of a PDF filing, catching a partial rather than exact SDN alias match, and corroborating an identity through a cached LinkedIn photo. That last sequence is the kind of thing that normally eats thirty minutes of an analyst's morning. The auditability story is the other half. Every step in the walkthrough carries a timestamp, and the output is framed as a complete evidence trail with cited sources, which is what exam and audit teams ask for. The team's published engineering — ad-hoc feature generation on ClickHouse, sparse autoencoder interpretation of AI flags, and work on how model upgrades affect confidence signals — suggests interpretability is treated as a first-class problem rather than a marketing line. ClickHouse as the analytics substrate is a reasonable choice for high-volume alert data. Where it doesn't fit: the product is purpose-built for AML and fraud at supervised institutions, so it won't serve general AI use cases and it's overkill for a business with no sanctions or AML obligations. Small volumes where manual review is already cheap don't need it. The company rebranded from Intrinsic in March 2025 and closed a $21M Series A in March 2026, making it early-stage relative to incumbent case-management vendors — onboarding and support depth are a fair thing to probe in a pilot. Pricing is contact-only, and the site carries no public documentation hub beyond a /docs path that currently serves the marketing page, so plan for a hands-on evaluation with the vendor rather than self-serve discovery. Buyers should also be clear-eyed that the headline 90% and 10x numbers come from the vendor.
Researching Variance? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas Variance actually fits — and what changes day-one when you adopt it.
An unusual $48k wire to a newly registered Hong Kong shell company fires the monitoring system. Instead of opening the registry, sanctions list and search tabs manually, the analyst hands the alert to a Variance agent, which checks the HK Companies Registry, transliterates the Mandarin director name from the filing, runs the OFAC SDN List and surfaces a partial alias overlap.
Outcome: The analyst receives a verdict with six cited artifacts across four data sources and either clears the false positive or escalates with the evidence file already assembled.
An escalated case needs an identity resolved across jurisdictions. The agent pulls the cached LinkedIn with a matching photo and alternate spelling, then links the same person to a 2023 ICIJ investigation into trade-based laundering.
Outcome: The investigator gets an identity linkage chain with sources and timestamps instead of a folder of screenshots, and the case file is ready for the audit trail.
An examiner asks how a specific alert was dispositioned six months earlier. The officer pulls the case and every agent step carries a timestamp, cited source and artifact count.
Outcome: The exam response is reconstructable from the evidence trail without re-interviewing the analyst who worked the alert.
Use Cases
- Automate evidence collection for suspicious transaction alerts
- Cross-reference beneficiary names against global sanctions lists
- Extract and verify director information from foreign company filings
- Link aliases across open web, social media and adverse media databases
- Generate audit-ready case files with cited sources and timestamps
- Streamline L1-L3 review processes in AML compliance teams
- Close false alerts by replicating an existing manual investigative workflow
- Convert scanned, handwritten and web evidence into structured case data
Limitations
- Purpose-built for AML and fraud investigation at regulated institutions, so it will not serve general AI or non-compliance workflows.
- Coverage of the investigative surface is documented around KYC, KYB, transaction monitoring and CDD; other compliance functions are not evidenced on the site.
- The headline 90%-of-evidence and 10x-cycle figures are vendor claims without published methodology.
- The site's marketing pages repeatedly surface in place of documentation, so there is little public material to review before a demo.
- As a company that rebranded from Intrinsic in March 2025 and closed a Series A in March 2026, support and onboarding depth are less proven than at established case-management vendors.
as of 2026-10-03
Verification history
We have re-verified Variance 7 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-checked, vendor evidence unchanged
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
Showing the 6 most recent of 7 verification passes.
Free to cite with attribution — this page re-verifies continuously.
Where the pricing makes sense
The company stage and team size where Variance's pricing actually pencils out — and where peers do it cheaper.
Variance sells to supervised financial institutions rather than to individuals or small teams, so the comparison set is enterprise AML platforms — Caseware, Lucinity, Unit21, NICE Actimize — not $20/mo AI subscriptions. Pricing is not published; budget conversations happen with sales and depend on alert volume, review levels covered and how many investigators the agents support. Expect the spend to be justified against analyst headcount, so model the pilot against hours saved per alert rather
Setup time & first value
How long it actually takes to get something useful out of Variance — broken out by persona, not the marketing-page minute.
For a regulated institution already running transaction monitoring: expect a vendor-led onboarding covering alert ingestion from your existing system plus configuration of the registries, sanctions lists and adverse-media sources you want the agents to consult. Because there is no public documentation hub to self-onboard from, plan for hands-on sessions with Variance before first value. Analysts
Switching to or from Variance
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From manual analyst research: agents replicate the registry, sanctions and open-web steps your team already performs by hand on each alert
- →From a legacy case-management tool: keep your existing transaction monitoring and add Variance as the investigative evidence layer on top
- →From general-purpose AI assistants: replace ad-hoc chat-based research with cited, timestamped artifacts tied to the case file
- ↗To an incumbent AML suite (Caseware, Lucinity, Unit21): consolidate the investigative layer into a platform you already license
- ↗To manual review: retained if alert volume drops enough that analyst research is cheaper than an agent platform
Integrations
Resources & Guides
Tutorials & Learning
YouTube returned 6 videos for “Variance”, and we withheld 6: 6 could not be judged, because “Variance” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about Variance.
Official links
Tools that pair well with Variance
Common stack mates teams adopt alongside Variance, with the specific reason each pairing earns its keep.
Norm.ai
Agentic law platform that encodes legal rules into verifiable AI agents for regulated institutions.
Socialprofiler
Socialprofiler runs an AI social media background check on a person and returns a readable report from public Facebook, TikTok, Instagram and X activity.
Sixtyfour
Sixtyfour is an identity intelligence platform where one identifier—name, email, phone, or wallet—becomes a fully cited people and entity
Featured Head-to-Head Comparisons
Variance vs Bitsgap
Choose Variance if you are a regulated financial institution that needs AI to automate AML/fraud investigations with full audit trails and explainable outputs. Choose Bitsgap if you are a crypto trader looking for automated trading bots across multiple exchanges with backtesting and a demo mode.
Variance vs Presto Voice
If you're in financial compliance, Variance is the clear choice: it's purpose-built for AML investigations with audit trails and automated evidence collection. For QSR drive-thru operations, Presto Voice delivers proven revenue lift and high automation rates. Both are contact-sales enterprise products, but their domains are entirely separate—choose based on your industry.
Variance vs Truleo
If you're in law enforcement needing to connect siloed data (RMS, jail calls, BWC) and reduce report writing time, Truleo is the clear choice. For AML/fraud teams in regulated finance, Variance's audit-ready AI agents with explainable evidence trails are unbeatable. Your domain dictates the winner.
Alternatives to Variance
View all
Norm.ai
Agentic law platform that encodes legal rules into verifiable AI agents for regulated institutions.
Socialprofiler
Socialprofiler runs an AI social media background check on a person and returns a readable report from public Facebook, TikTok, Instagram and X activity.
Frequently Asked Questions
Used Variance? Help shape our editorial sentiment research.