INFINITUM CAPITAL LIMITED

Assessment

Entity resolution at the registry layer is HIGH confidence. An authoritative registry returned a single, internally consistent record matching the subject name and the user-provided jurisdiction (GB). Entity resolution at the open-source layer is LOW confidence — see Sections 5 and 7.

Evidence

Key Findings

• The company has been on the register for just over ten years as at the report date and is active. Longevity of registration is a mild positive indicator but says nothing about trading activity.
• The registered office is a care-of address at Goldwyns London LLP, a third-party firm. This is lawful and routine for smaller UK companies. It is not an adverse finding; it does mean the entity's operational premises, staffing and place of effective management are unevidenced.
• No SIC code or business description was retrieved. The word "Capital" in the legal name is not evidence of investment or financial-services activity and must not be treated as such.

Risk Implication

Under MLR 2017 reg. 28(2)–(3) the firm must identify the customer and verify its identity from a reliable, independent source, and must understand the nature of its business. The first limb is satisfied by the Companies House record. The second is not. Without a SIC code, filed accounts or a documented business description, no expected-activity profile can be constructed, which in turn undermines any transaction-monitoring calibration. The absence of an FCA Register check (Section 6) compounds this: it is currently unknown whether the entity conducts, or purports to conduct, regulated activity.

Assessment

Control is partially mapped. One officer is confirmed by the registry; one controller is declared by the company but not independently verified. Ownership percentages, the nature-of-control statement and any corporate chain above the PSC are absent from the evidence.

Evidence

No corporate PSC, parent undertaking, trust, nominee arrangement or intermediate holding entity appears in the evidence. Absence of such an entry in the retrieved extract is not proof that none exists — the full PSC and filing history was not obtained.

Key Findings

• The 25% threshold. Under the UK PSC regime (Companies Act 2006 Part 21A, as amended), a person is registrable where they hold, directly or indirectly, more than 25% of shares or more than 25% of voting rights, hold the right to appoint or remove a majority of the board, or otherwise exercise significant influence or control. The evidence records that Shruti Luthra is a PSC but does not record which of these conditions applies or the applicable percentage band. That detail is material and must be obtained.
• DECLARED ≠ CONFIRMED. Companies House does not verify PSC filings substantively. Treating a PSC entry as verified beneficial ownership is a recognised control failure.
• The relationship (if any) between the director and the PSC is not established. No evidence of family, commercial or contractual linkage was retrieved, and none should be inferred.

Risk Implication

• FATF Recommendation 24 requires adequate, accurate and up-to-date beneficial ownership information; EU 5AMLD Article 30 and MLR 2017 reg. 28(4) require the firm to take reasonable measures of its own. Registry reliance alone does not discharge this.
• Because the PSC is unverified, the sanctions conclusion in Section 3 and the PEP gap in Section 4 both inherit that uncertainty: an unverified natural person cannot be reliably screened. This is the principal reason the Transparency Risk sub-score of 8/100 (Section 11) should be regarded as optimistic on the evidence actually supplied.
• No opacity indicators of the classic type (offshore holding layers, bearer arrangements, nominee directors) were identified — but equally, no chain was mapped, so their absence is unproven.

Assessment

Screen status: RAN — genuine nil result. The sanctions screen executed and returned no findings. This is a meaningful clean signal, distinct from a source that failed or was skipped.

Evidence

Key Findings

• No designation, no exact match, and no possible match was returned for the entity name screened.
• The screen covered OFAC (US), UK OFSI and UN consolidated lists. EU consolidated list, and country-specific regimes beyond those three, are not named in the ledger and cannot be assumed to have been covered.
• The evidence does not confirm that the individuals identified in Section 2 were screened as natural persons. Entity-level screening does not satisfy the requirement to screen controllers.

Risk Implication

UK financial sanctions obligations (Sanctions and Anti-Money Laundering Act 2018; OFSI reporting obligations) apply on a strict-liability basis to dealings with designated persons. The nil result materially supports the low Sanctions sub-score of 3/100 applied at the heaviest weight in the model (34.1% — see Section 11), and is the single largest contributor to the composite remaining low.

Two limitations must be recorded on file: the list vintage/screening timestamp is not evidenced, and individual-level screening of the director and declared PSC is not evidenced. Both should be closed and documented before the file is signed off, so that the nil result is auditable rather than merely asserted.

Assessment

Screen status: NOT RUN

No PEP screening provider is configured. No PEP, RCA (relative or close associate) or HIO (head of international organisation) determination has been made for this entity or for any individual connected to it.

Evidence

Key Findings

• PEP status is UNDETERMINED, not negative. Nothing in this report may be read as stating that the entity or its controllers are not PEPs.
• No inference of non-PEP status has been drawn from the absence of political keywords in the Google search results. Role-keyword matching in general web search is not a PEP screen and has not been used as one here.
• This gap is explicitly excluded from the composite score (Section 11), where the PEP factor is shown as UNDETERMINED with no weight.

Risk Implication

• FATF Recommendation 12 and FCA SYSC 12.1, given domestic effect by MLR 2017 regs. 33(1)(b) and 35, require firms to have appropriate risk-management systems to determine whether a customer or its beneficial owner is a PEP, family member or known close associate — and, where they are, to apply enhanced due diligence, obtain senior management approval and establish source of wealth and source of funds.
• Because the determination has not been made, the firm cannot currently evidence compliance with reg. 35. Onboarding on the basis of a composite score that omits this factor would be a documented control gap.
• The gap interacts with Section 2: the PSC is DECLARED only, so even once a provider is connected, screening should be performed against verified identity data rather than an unverified self-declaration.

Required next step: connect a licensed PEP/sanctions data provider and screen the entity, Dhruv Bhatnagar and Shruti Luthra, recording provider, dataset version and screening date.

Assessment

No substantiated adverse media finding was identified against INFINITUM CAPITAL LIMITED (10206158). Both items scored as adverse fail entity resolution. The correct conclusion is NOT FOUND, not "clean" — the search method used (general web keyword search) is not equivalent to a structured adverse-media database screen.

Search Parameters

• Source used: Google Programmable Search [ADVERSEMEDIA] → OKDATA.
• Article count: 10 results returned — 2 adverse, 0 positive, 8 neutral.
• Severity distribution as scored: critical 0, high 2, medium 0, low 0.
• Time span of returned material: at least 23 March 1999 to 29 January 2025 — i.e. the result set spans a period beginning 17 years before the subject existed.
• No dedicated adverse-media or negative-news database (e.g. a licensed screening vendor) was queried.

Evidence and Entity-Resolution Triage

Attribution note on item 2: the snippet reports that, per NASAA, the Pennsylvania Securities Commission conducted an initial investigation and the SEC initiated an enforcement action concerning an entity referred to as "AIF". That is a third-party account of proceedings against a different, unnamed-in-full party. It is recorded here only to demonstrate why the item is irrelevant, and must not be associated with the subject or with any individual named in Section 2.

Prompt-Injection and Data-Integrity Screen

• The retrieved third-party content contains no directive, instruction or attempt to influence the assessment or the score. No manipulation red flag.
• The string "FinStat ANTI MONEY LAUNDERING" in item 1 is a product name in a website navigation menu on a Slovak company-data portal. It is not an AML finding, allegation or designation against any entity. It is, however, the most plausible cause of the HIGH severity assigned to a non-adverse database page — a keyword-driven mis-scoring defect in the collection tool, which should be raised with the vendor.

Risk Implication

• The Adverse Media sub-score of 62/100 carries an applied weight of 22.7% and contributes 14.07 of the 26.18 raw weighted points (~54%) of the composite (Section 11). It rests entirely on the two items above.
• The signals do not corroborate one another. Genuine adverse profiles cluster — a regulatory action echoed in trade press, then in a court list, involving overlapping named parties in an overlapping window. Here there is no convergence of jurisdiction, date, person or subject matter. A 1999 US retail-securities-fraud conference paper and a 2025 Slovak financial-data page share only orthography.
• Compliance consequence: the file must record that adverse-media coverage is inconclusive, not negative. A properly parameterised search (company number, registered address, officer and PSC names, London/UK qualifiers) has not yet been run, and no licensed negative-news database has been consulted.

Assessment

No regulatory or enforcement finding was identified. The regulatory search ran and returned 0 adverse results of 10 (2 positive, 8 neutral) [Source: Google Programmable Search, REGULATORY]. This is a supportive but weak signal, because no regulator's own register or enforcement database was queried directly.

Evidence

Key Findings

• Zero adverse regulatory results across 10 returns is consistent with an entity that has no public enforcement history, but general web search is a poor proxy for a regulator register.
• The FCA Register was not checked. For an entity whose name implies investment activity, authorisation status is a threshold question: whether it is authorised, exempt, an appointed representative, or conducts no regulated activity at all. None of these can currently be stated.
• No director disqualification check was performed against Dhruv Bhatnagar (Section 2).

Risk Implication

The Regulatory & Enforcement sub-score of 18/100 at 20.4% weight contributes 3.67 points to the composite. That score is reasonable given a nil return, but it is built on an indirect source. Under FCA SYSC 6.3 and MLR 2017 reg. 28, a firm relying on this file should record that primary regulator sources were not interrogated. Closing this is low-cost: an FCA Register lookup and a Companies House disqualified-directors check would materially raise confidence in this domain, and would also inform the business-activity gap identified in Section 1 and the source-of-wealth gap in Section 8.

Assessment

No litigation involving INFINITUM CAPITAL LIMITED (10206158) was identified. All three items scored as adverse are misclassifications or string artefacts. Court records were not searched directly — the only source was general web search.

Evidence and Triage

Required litigation attributes — jurisdiction, court, case type, parties, status, outcome — are not applicable to any of the above, because none of the items is a proceeding involving the subject.

Pattern Screen

• Repeat litigation: none identified.
• Class actions: none identified.
• Insolvency proceedings: none identified; registry status is active, which is inconsistent with an open winding-up or dissolution process but does not exclude other insolvency events not captured in the retrieved extract.
• Two of the three sources are Facebook posts. Social-media posts are not reliable evidence of legal proceedings and should never sustain a litigation risk rating.

Risk Implication

• The Litigation sub-score of 41/100 at 11.4% weight contributes 4.67 points to the composite on the basis of the three items above. On the evidence, that contribution is not supported by any identified proceeding against the subject.
• No court registry was searched. Neither the English & Welsh courts (including the Business and Property Courts / Companies Court), the Insolvency Service register, nor any foreign court database appears in the source ledger. The file must therefore state that litigation exposure is UNVERIFIED, not absent.
• Cross-reference: the Ayar Labs item is the only evidence in the entire set that touches on business activity. If entity resolution confirmed it as the subject, it would be relevant to Section 8 (source of wealth, venture-investment model) and Section 9 (US exposure). It is currently unresolved and must not be relied upon.

Assessment

Screen status: NOT RUN (no dedicated ESG data source)

No ESG, sustainability, human-rights, labour or supply-chain database was consulted. The only inputs capable of surfacing ESG-related conduct issues were the four general Google Programmable Search domains, which returned no ESG-relevant material concerning the subject.

Evidence

Key Findings

• ESG risk is UNDETERMINED, not low. No conclusion of good ESG standing is supported.
• Applicability of the UK Modern Slavery Act 2015 s.54 (annual turnover threshold) cannot be assessed because turnover is unknown (Section 8). The same limitation applies to any CSRD/TCFD-style reporting expectation.
• The two rent-regulation Facebook items in Section 7 concern housing-cost disputes involving unrelated parties in the US and Trinidad & Tobago. They are not ESG findings against the subject and are not treated as such.

Risk Implication

ESG & Conduct is not a weighted factor in the scoring model (Section 11), so this gap does not affect the composite. For institutions with ESG-linked onboarding policies, reputational-risk appetite statements or supply-chain obligations, the file is currently silent on this dimension and cannot support an ESG attestation. If the entity is confirmed to be an investment vehicle (see the unresolved Ayar Labs reference, Section 7), portfolio-level ESG exposure would become the relevant unit of analysis and is entirely unexamined.

Assessment

No financial data for the subject entity was obtained. This is a material due-diligence deficiency rather than a neutral outcome. Financial Profile is not a weighted factor in the scoring model (Section 11) and therefore this gap does not depress the composite — a limitation that must be understood when reading the 26/100 figure.

Evidence

Key Findings

• The financial search returned zero adverse items and three positive items, with all severity counts at zero. No insolvency, default, distress or fraud signal was returned. The evidence does not state which entity the three positive items concern, so they cannot be attributed to the subject.
• Source of wealth and source of funds are entirely unevidenced. No shareholder capital, revenue stream, client base or funding history has been established.
• The Slovak entity in Section 5 is a financial-results database page; its figures relate to a different legal person and must not be imported into this profile.

Red-Flag Screen

Risk Implication

MLR 2017 reg. 28(2) requires the firm to understand the nature of the customer's business and, where relevant, its ownership and control structure; reg. 33 requires source-of-funds enquiry in higher-risk cases. Neither can be evidenced here. For a UK company incorporated in 2016, filed accounts should exist and are obtainable at negligible cost. Obtaining the latest accounts and confirmation statement is the highest-value, lowest-effort remediation step in this file, and it also addresses the business-activity gap flagged in Sections 1 and 6.

Assessment

The only evidenced geography is the United Kingdom: incorporation in England & Wales and a registered office in the City of London [Source: Companies House]. Operating jurisdictions, customer geography and counterparty geography are not established.

Evidence

Key Findings

• The 35/100 jurisdiction score is model-internal. No FATF list, EU high-risk list or TI CPI dataset was queried in this run, so the reference data underpinning that figure cannot be independently verified from this evidence set. It is reported verbatim in Section 11 but should not be represented to a regulator as the output of a dataset lookup.
• Incidental foreign geography in the evidence belongs to other entities, not the subject: Slovakia (finstat.sk item, Section 5), the United States (NASAA 1999 paper; Ayar Labs Series C; New York rent-regulation posts) and Trinidad & Tobago (BreakingNewsTnT post). None of these establishes any exposure of the subject.
• The only potential non-UK nexus is the unresolved Ayar Labs investor reference (US semiconductor sector, 2022). If confirmed, it would indicate US cross-border investment activity. It is not confirmed.

Risk Implication

The UK is a well-regulated jurisdiction with a mature AML framework, which supports a low-to-moderate jurisdictional rating. However, UK incorporation is also a known vector for shell-company misuse, and the mitigants against that risk — verified beneficial ownership (Section 2), evidenced business activity (Section 1), filed accounts (Section 8) — are precisely the items missing from this file. Jurisdiction risk should therefore be re-assessed once operating geography and counterparty exposure are documented, particularly if any non-UK nexus is confirmed.

Composite Result

Composite: 26/100 — model band "Medium-Low". Coverage: PARTIAL.

Weights below are the applied (renormalised) weights supplied by the model, redistributed across the domains that actually ran. They sum to 100%.

Scoring Table

Calculation

Composite Score = Σ(Factor Score × Applied Weight)

``` Sanctions & Watchlists 3 × 0.341 = 1.023 Adverse Media 62 × 0.227 = 14.074 Regulatory & Enforcement 18 × 0.204 = 3.672 Politically Exposed Persons UNDETERMINED × 0 = 0.000 Litigation 41 × 0.114 = 4.674 Jurisdiction Risk 35 × 0.068 = 2.380 Transparency Risk 8 × 0.045 = 0.360

Composite = 26.183 ≈ 26/100 ```

What Actually Drives the Score

• Adverse Media alone accounts for 14.07 of 26.18 raw weighted points — approximately 54% of the composite. Adverse Media plus Litigation account for 18.75 points, or ~72%.
• Both of those domains rest exclusively on items that fail entity resolution against Companies House 10206158, as triaged item-by-item in Sections 5 and 7.
• The single largest downward driver is the Sanctions factor: a genuine nil result scored at 3/100 carrying the heaviest applied weight (34.1%).

Sensitivity — Illustrative Only

The scores above are reported verbatim and are not modified. For committee information only: if Adverse Media and Litigation were each re-scored to a nominal 5/100 to reflect zero entity-resolved adverse findings, holding all other inputs constant, the composite would fall to approximately 9/100 (1.023 + 1.135 + 3.672 + 0.570 + 2.380 + 0.360 = 9.14). This is presented solely to quantify how much of the current score is attributable to unresolved items; it is not an alternative rating and must not be substituted for the model output.

Two Errors That Do Not Cancel

1. 1.Upward distortion on the domains that were searched (adverse media, litigation), caused by name-collision false positives.
2. 2.Downward incompleteness on a domain that was not searched (PEP), plus unweighted gaps in beneficial-ownership verification and financial profile.

These operate in opposite directions but are not offsetting: they must be remediated separately. Consequently, the reported risk level is MEDIUM rather than LOW, notwithstanding the numeric band. The composite is a partial measure and should not be presented internally or externally as a completed screen.

Recommendation

STANDARD CDD — CONDITIONAL. Additional information required before onboarding sign-off.

• Model output: Standard CDD; EDD required: NO; monitoring frequency: ANNUAL.
• Analyst position: Standard CDD is the correct baseline. It cannot be signed off as complete until the PEP screen has actually run and the declared PSC has been independently verified. This is a conditional approval, not an escalation to EDD, and not a recommendation to decline.
• EDD is not currently indicated by the evidence. No sanctions match, no substantiated adverse media, no identified enforcement matter and no identified litigation. EDD would become mandatory if PEP screening returns a positive result (MLR 2017 reg. 35) or if any of the trigger events below occurs.
• Rejection is not supported. No evidence justifies refusal; the deficiencies are informational gaps that are readily closable.

Rationale

Outstanding Information Gaps

1. 1.PEP/RCA/HIO status of the entity, Dhruv Bhatnagar and Shruti Luthra — wholly unassessed.
2. 2.Independent verification of Shruti Luthra as PSC, including nature and percentage of control against the 25% threshold, and the chain to any ultimate natural persons above her.
3. 3.SIC code / business description; latest statutory accounts; confirmation statement; officer and filing history.
4. 4.FCA authorisation status and any director-disqualification record.
5. 5.Operating jurisdictions, expected counterparties and expected transaction profile.
6. 6.Resolution of whether the "Infinitum Capital" named as an Ayar Labs Series C investor is the subject entity.
7. 7.Sanctions screening metadata: list vintage, matching threshold, screening date, and whether individuals were screened.

Required Actions

1. 1.Connect a licensed PEP/sanctions provider and screen the entity and both named individuals. Record provider, dataset version and screening date. Do not close the file before this is done.
2. 2.Verify the PSC declaration independently — identity evidence for Shruti Luthra, nature-of-control statement, ownership percentage band, and any layer above her.
3. 3.Pull the full Companies House record — SIC code, latest accounts, confirmation statement, officer history, mortgage/charges register.
4. 4.Check the FCA Register and the Companies House disqualified-directors register.
5. 5.Re-run adverse media and litigation with entity-resolved parameters (company number, registered address, officer/PSC names, UK qualifiers), and search a court/insolvency source directly rather than relying on general web search.
6. 6.Log the false-positive triage in Sections 5 and 7 on the customer file, so the same name-collision hits are suppressed at next review rather than re-escalated.
7. 7.Raise the collection-tool defect (HIGH severity assigned to a non-adverse Slovak database page, apparently on the navigation string "FinStat ANTI MONEY LAUNDERING") with the screening vendor.

Monitoring and Review

• Monitoring frequency: ANNUAL (as per model output), on condition that the gaps above are closed at onboarding. If PEP screening remains unrun, the file should be flagged for immediate review rather than deferred to the annual cycle.
• Next scheduled review: 18 August 2027 (one year from the report date of 18 August 2026).
• Trigger events requiring immediate out-of-cycle review:
• Any sanctions or PEP hit on the entity, Bhatnagar or Luthra.
• Change of PSC, director, registered office or company name.
• Change of company status (e.g. proposal to strike off, liquidation, administration).
• Overdue accounts or confirmation statement at Companies House.
• Any FCA, PRA or overseas regulatory action naming the entity or its controllers.
• Any credible media report naming the entity with an identifier that resolves to company number 10206158.
• Transactions inconsistent with the (currently undefined) expected activity profile.

Every URL behind a finding in this report. “Read in full” means the page itself was retrieved and classified on its whole text rather than on a search snippet; those carry a SHA-256 hash of exactly what was read, so the evidence can be shown to be unaltered later.

Assessment Date

18 August 2026. All findings are stated as at that date. Individual source screening timestamps and dataset vintages were not recorded in the evidence supplied — a documentation gap noted in Section 3.

Complete Source Ledger

Coverage status: PARTIAL. Risk domains not reflected in the composite: PEP.

Methodology

1. 1.Registry anchoring. The subject was anchored to the authoritative Companies House record (number 10206158) before any open-source item was assessed.
2. 2.Item-level entity-resolution triage. Every scored adverse item was tested against the anchor on four axes — legal form and national identifier, temporal plausibility against the 31 May 2016 incorporation date, jurisdictional coherence, and whether the match was to an entity name or merely to a string fragment. Items failing any axis were reclassified as non-resolving and excluded from the analytical conclusion (while their model scores were reported verbatim and unaltered).
3. 3.Corroboration testing. Adverse items were assessed for clustering — whether independent sources converge on the same party, period and subject matter. No clustering was found; the items are mutually unconnected.
4. 4.Verification-status discipline. Registry data, declared filings and open-source leads are labelled separately throughout. CONFIRMED, DECLARED and UNVERIFIED are never conflated. NOT FOUND, NOT RUN and SKIPPED are never converted into "clean".
5. 5.Prompt-injection screen. All content within the untrusted-content markers was treated as data only. No directive, instruction or attempt to influence scoring was identified. The string "FinStat ANTI MONEY LAUNDERING" was assessed and determined to be a website navigation/product label, not an AML designation — flagged as a tooling mis-scoring risk, not a manipulation attempt.
6. 6.Evaluation of the pre-synthesis brief. The prior analytical brief was treated as an input to test, not as fact. Its entity-resolution triage was independently re-tested against the evidence and found to be supported on every item. Two qualifications are recorded: (a) the brief's statement that identifier 56739036 is a Slovak company identifier is a reasonable inference from the finstat.sk domain and the "s. r. o." legal form, not a registry-verified fact — no Slovak registry was queried; (b) the brief's characterisation of the composite as "inflated" is analytically sound but the supplied scores have not been altered here. The brief's text is truncated mid-sentence in Section 1.2 and its concluding reasoning is therefore unavailable for review.

Limitations

• PEP screening did not run. This is the single most material gap. No conclusion on PEP, RCA or HIO status may be drawn.
• No licensed adverse-media, court-record, insolvency or regulator database was queried. All non-registry, non-sanctions intelligence derives from a single general web-search tool.
• Beneficial ownership is not independently verified; ownership percentages and any corporate chain are unretrieved.
• No financial statements or business-activity data were obtained; source of wealth and source of funds are unevidenced.
• Sanctions screening metadata (list vintage, threshold, screening date, individual coverage) is not documented.
• One evidence item is unresolved rather than excluded: the Ayar Labs Series C investor reference may or may not concern the subject.
• No ESG data source was consulted.

Legal Framework References

UK Money Laundering Regulations 2017 (regs. 27, 28, 33, 35); FCA SYSC 6.3 and SYSC 12.1; FATF Recommendations 10, 12 and 24; EU 5AMLD Article 30; Companies Act 2006 Part 21A (PSC regime, 25% threshold); Sanctions and Anti-Money Laundering Act 2018; UK Modern Slavery Act 2015 s.54 (applicability unassessed).

Confidence

Overall confidence: MEDIUM.

• Registry layer — HIGH. A single authoritative record with a registration number, incorporation date, company type, status and registered office, consistent with the user-provided GB jurisdiction.
• Sanctions layer — MEDIUM-HIGH. The screen genuinely ran and returned nil, but the metadata needed to make that result auditable was not captured.
• Open-source layer — LOW. Every scored adverse item failed entity resolution; the searches were not entity-parameterised; and no primary regulator, court or negative-news database was used. The correct reading is that these domains are uninformative, not favourable.
• PEP layer — NONE. The screen did not run.

The composite of 26/100 should therefore be presented to the risk committee as a provisional, partial figure, materially inflated by unresolved name-collision hits and materially incomplete on PEP exposure.