Reconcile faster.
Built for Operations Teams
who can't afford to wait.

Vajra reconciles positions, transactions, and balances across equities, bank debt, cash, derivatives, corporate actions, and private assets - classifying breaks and proposing resolutions - with 95% handled autonomously at onboarding, scaling toward 99% at full maturity.

Request early access See how it works
95-99%
Autonomous resolution
All
Asset classes covered
Real-time & self-improving
Alerts and resolution model

Reconciliation has always been treated as a matching problem. But in practice, matching is not where teams spend their time. The real work begins when things do not match.

Exceptions are where operational cost accumulates, where risk hides, and where human effort is concentrated.

Yet most systems are designed to optimize matching rates, not to understand or resolve the underlying causes of exceptions.

Vajra is built differently.

Instead of focusing on matches, we focus on exceptions. We identify them, understand their root causes, resolve them through intelligent workflows, and ultimately reduce their occurrence over time.

The goal is not better reconciliation. The goal is better risk adjusted decisioning.

Every asset class. Every data format.
All transactions and balances.

Vajra reconciles across any asset class - as long as we can get the data. The examples below reflect common starting points, but coverage expands with each new data source connected.

Bank Debt - Loans & Revolvers
Email / PDF / JSON/XML → API
Parse unstructured agent bank notices, normalize timing mismatches, and transition clients to direct API push as relationships mature.
Cash - Bank Accounts
SFTP / BAI2 / MT940 / API
Auto-match across inconsistent formats, identify intraday vs EOD gaps, and surface missing references at scale.
Equities & Listed Securities
CSV / Excel / XML / Custodian API
Position and trade reconciliation with custodian records. Handles lot mismatches, FX differences, and corporate action timing gaps.
Corporate Actions
Email / PDF / CSV / Admin warehouse
Parse and structure lifecycle events, identify data inconsistencies, and handle manual elections - with admin warehouse access when available.
Derivatives - OTC
CSV / Excel / PDF / Counterparty API
Compare counterparty valuations, explain deltas with enriched context, and surface unstructured commentary from email attachments.
Private Assets - PE & Credit
PDF / Excel / GP Portal / Admin API
Document parsing as a core capability. Handles highly unstructured GP reports and delayed reporting, with structured ingestion when admin access is provided.
Any asset class - if we can get the data
Any format, any source
The six above are starting points, not limits. Vajra's ingestion layer is built to normalize any structured or semi-structured data source. If your counterparty, custodian, or administrator can provide the data, Vajra can reconcile it.
ad> Vajra - AI-Driven Investment Operations Reconciliation

A self-improving resolution engine.

95-99%
Deterministic & autonomous
Adaptive
For novel breaks

Vajra's decision layer replicates how a skilled analyst approaches reconciliation - matching positions and transactions, classifying breaks by type and cause, and proposing ranked resolutions. At onboarding, ~95% of breaks are handled deterministically and autonomously. As Vajra learns the client's patterns, that number moves toward 99% - leaving only true edge cases for human review.

1
Detect & classify
Rule-based validation and historical pattern matching handle the vast majority of breaks automatically - classifying each one, matching it to a known resolution pathway, and logging the outcome auditably. Coverage increases as the system learns the client's specific environment.
2
Recommend & validate
Novel breaks are evaluated using probabilistic reasoning. A ranked resolution proposal is surfaced for human review before any external action is taken.
3
Learn & improve
Outcomes - explicit or inferred from state changes - feed back into the decision layer. Each reconciled break improves future classification and proposal accuracy.

The right alert,
at the right time.

Every break is scored by a composite model across three signals. The score determines whether a notification fires instantly, joins the next batch, or waits for EOD review.

01
Materiality
Break size as a percentage of fund NAV sets the baseline severity tier - critical at ≥1%, elevated at 0.1–1%, routine below that.
02
Break type
Trade breaks shift live risk exposure and escalate automatically. Non-trade breaks default to scheduled batch delivery.
03
Timing
Proximity to settlement cutoffs and NAV deadlines boosts urgency. A break at 3:45 PM is not the same as one at 9:30 AM.

Founded inside the system,
built to change it.

Vajra reflects 15+ years of experience leading and transforming investment operations at scale.

GK
Gayathri Kalyansundar
Founder

Gayathri has spent 15+ years inside investment operations at Goldman Sachs, Citco, CIBC, and Citadel, not observing it but running it. She has built and led corporate actions, FX settlement, and middle office teams through periods of rapid scale, regulatory change, and post-crisis recovery.

The pattern across her career is consistent. She takes broken or underperforming processes, understands them deeply, and fixes them at the root. At Goldman Sachs, this meant reducing exceptions by 98% in six months and delivering over $2B in capital savings through a large-scale client account integration. At Citco, this included heading the Toronto Middle Office and successfully managing some of the firm's most complex clients, driving strong outcomes in risk and operational efficiency. At CIBC, she led operational risk programs, including board reporting and enterprise-wide risk initiatives, strengthening risk visibility and decision-making. At Citadel, she led data-driven workflow redesigns that reduced reliance on manual exception handling.

Having worked across a wide span of Operations teams, she developed a firsthand understanding of the challenges that drive operational risk, inefficiency, and manual effort across financial institutions. She repeatedly saw talented teams spending countless hours investigating exceptions, chasing breaks, and piecing together information across fragmented systems.

Those experiences shaped a simple belief: AI should help people make better decisions. The future of Operations is not replacing human expertise, it is augmenting it. Humans bring judgment, context, and experience. AI brings speed, scale, and pattern recognition.

Gayathri founded Vajra to combine the strengths of both, helping Operations teams identify issues faster, understand root causes sooner, and focus their time where human expertise creates the most value.

LinkedIn

Ready to cut your
break queue in half?

We're onboarding select investment operations teams now. Tell us about your environment.