L E N D I Q

Upload a bank statement. Get an underwriting decision.

The official Python SDK for the LendIQ API.
Turn months of manual underwriting into a single API call.

       

Get API Key  ·  Documentation  ·  API Reference

---

7 lines. That's it.

from lendiq import LendIQClient

client = LendIQClient(api_key="liq_live_...")

deal = client.deals.create(business_name="Acme Trucking LLC")
client.documents.upload(deal.id, "statements/chase_jan.pdf")

result = client.deals.get(deal.id)
print(result.recommendation.decision)  # "approved"

That PDF just went through Gemini-powered extraction, transaction screening, tamper detection, 12-factor health scoring, and a full underwriting recommendation. Your team didn't write a single rule.

Why teams switch to LendIQ

Before LendIQ Manual PDF review (20-40 min per deal) Spreadsheet-based scoring Inconsistent underwriting criteria No tamper detection Weeks to onboard new data sources

After LendIQ Automated analysis (seconds per deal) 12-factor composite health scoring Configurable rulesets with version history PDF integrity + tampering detection One API call to process any bank statement

LendIQ vs. the alternatives

	LendIQ	Ocrolus	LendAPI	Plaid (Asset Reports)	DIY (in-house)
Upload PDF → full underwriting decision	Yes	No	Partial	No	Build it yourself
Bank statement + tax return + P&L support	All three	Statements only	Statements only	No PDFs	Depends on scope
Gemini native PDF extraction	Yes	Template OCR	Template OCR	N/A	Build it yourself
Tamper / fraud detection	Built in	Add-on	No	No	Build it yourself
Health scoring (12 sub-factors)	Built in	No	Basic	No	Build it yourself
Underwriting recommendation engine	Built in	No	Basic	No	Build it yourself
Custom rulesets with version history	Yes	No	No	No	Build it yourself
MCA position detection	Yes	No	No	No	Build it yourself
Real-time SSE streaming	Yes	Polling	Polling	Polling	Build it yourself
Typed SDKs (Python + TypeScript)	Both	Python only	None	Multiple	N/A
Pricing	Per document	Per document	Per document	Per connection	Engineering time

We don't just extract data from bank statements. We understand them — and give you a decision you can act on.

Install

pip install lendiq

Requires Python 3.10+. Only two dependencies: httpx and pydantic.

Sync and async, your choice

Both clients have identical APIs. Pick the one that fits your stack.

# Sync
with LendIQClient(api_key="liq_live_...") as client:
    deals = client.deals.list(status="ready")

# Async
from lendiq import AsyncLendIQClient

async with AsyncLendIQClient(api_key="liq_live_...") as client:
    deals = await client.deals.list(status="ready")

Everything is typed

Every method returns a Pydantic model. Not dict. Not Any. Real models with real autocompletion.

detail = client.deals.get(deal_id)

# IDE knows every field, every nested object, every type
detail.health.health_score                     # float
detail.health.health_grade                     # str
detail.recommendation.decision                 # "approved" | "conditional" | "declined"
detail.recommendation.risk_factors             # list[str]
detail.recommendation.strengths                # list[str]
detail.financials.avg_monthly_deposits         # float
detail.mca.mca_credit_score                    # float | None

# 12 health sub-factors, individually scored and weighted
for name, factor in detail.health.factors.items():
    print(f"{name}: {factor.score}/{factor.max} (weight: {factor.weight})")

All response models are forward-compatible (extra="allow"). New API fields are accepted automatically — your code never breaks on deploy.

Document intelligence, built in

Upload a PDF and get back structured intelligence — no OCR pipeline, no prompt engineering, no infra to manage.

doc = client.documents.get(doc_id)

# Extraction metadata
doc.extraction_method       # "gemini_native_pdf"
doc.extraction_confidence   # 0.94
doc.document_type           # "bank_statement"

# Tamper detection
doc.integrity.tampering_risk_level  # "clean" | "low" | "medium" | "high"

# Financial analysis (bank statements)
doc.analysis.average_daily_balance   # 16215.30
doc.analysis.total_deposits          # 48920.00
doc.analysis.health_score            # 78.5
doc.analysis.health_grade            # "B"

# Identity verification (driver's licenses)
doc.driver_license_analysis.full_name         # "John Smith"
doc.driver_license_analysis.expiration_date   # "2028-06-15"

# Bank verification (voided checks)
doc.voided_check_analysis.routing_number       # "021000021"
doc.voided_check_analysis.is_voided            # True

Instant analysis — no account required

Let prospects try LendIQ before they sign up. The instant endpoint processes a PDF in under 2 seconds with no data persistence.

result = client.instant.analyze("statement.pdf")

print(result.summary.total_deposits)        # 284500.00
print(result.summary.total_mca_positions)   # 3
print(result.summary.avg_revenue_quality)   # 0.82

# Per-file breakdown
for file in result.results:
    print(file.bank_name, file.nsf_count, file.mca_daily_obligation)

Real-time pipeline streaming

Watch documents process in real time. No polling.

for event in client.events.stream(deal_id=deal_id):
    if event.event == "stage":
        print(f"Stage: {event.data}")  # "classifying" → "extracting" → "screening" → "scoring"
    elif event.event == "complete":
        print("Analysis complete")

Auto-pagination

Forget page math. Iterate over thousands of records with a single loop.

for deal in client.deals.list_all(status="ready"):
    print(deal.business_name, deal.health_grade)
    # Typed as DealSummary — full autocompletion

Error handling that doesn't suck

Every error is a typed exception with the context you need to debug. No parsing strings.

from lendiq.exceptions import NotFoundError, RateLimitError, ValidationError

try:
    client.deals.get(deal_id)
except NotFoundError:
    # e.status_code == 404
    pass
except RateLimitError as e:
    # e.retry_after — seconds until you can retry
    pass
except ValidationError as e:
    # e.body — full validation error details
    pass
# Every error has e.request_id for instant support correlation

Webhook verification in one line

from lendiq.webhooks import verify_signature

# HMAC-SHA256 with constant-time comparison. Timing attacks don't apply.
verify_signature(payload=body, signature=headers["X-Webhook-Signature"], secret="whsec_...")

Built for production

Automatic retries	Exponential backoff with jitter. Honors Retry-After. Safe for mutations — POST/PUT/PATCH only retry on connection errors, never on HTTP status codes.
Idempotency	Pass an idempotency key on any write. Retry all you want.
Request tracking	Every request gets a UUID. Every error includes it. client.last_request_id is always available.
Timeouts	Sensible defaults per operation type. TIMEOUT_READ (10s), TIMEOUT_WRITE (30s), TIMEOUT_UPLOAD (120s), TIMEOUT_REPORT (300s).
Forward compatible	All models use extra="allow". API updates never break your code.
PEP 561	py.typed marker included. Full mypy --strict support.

Configuration

client = LendIQClient(
    api_key="liq_live_...",               # required
    base_url="https://iq.lend.works", # default
    timeout=30.0,                         # default (seconds)
    max_retries=2,                        # default
    logger=logging.getLogger("lendiq"), # optional debug logging
)

25 resources. Full API coverage.

Every endpoint in the LendIQ API has a typed method in this SDK.

	Resource	What it does
Core	client.deals	CRUD, decision, evaluate, notes, stats, analytics, batch
	client.documents	Upload, bulk upload, status, reprocess, triage, classify
	client.transactions	List, correct, corrections history
	client.exports	Deal and document CSV/PDF exports
	client.rulesets	Underwriting criteria CRUD, versioned, set default
Intelligence	client.instant	Free-tier instant PDF analysis (sub-2s, no persistence)
	client.lvl	Business validation runs, call queue, SAM entities
	client.sam_profiles	SAM.gov search profiles, watchers, automated triggers
	client.reviews	Document review queue, approve/correct workflow
Real-time	client.events	SSE streams for deals, org events, batch progress
	client.webhooks	Webhook config, test, delivery logs, retry
	client.notifications	In-app notifications and preferences
	client.push	Web push subscriptions
Platform	client.team	Invite, update, deactivate members
	client.keys	API key management
	client.shares	Public deal share links
	client.ingest	Bulk CRM ingest with batch tracking
	client.crm	Provider config, field mapping, bidirectional sync
	client.integrations	Slack, Teams, SMTP notification channels
	client.admin	Health, usage, error logs, DLQ, pipeline settings
	client.usage	Metering and processing time analytics
	client.oauth	Client credentials token exchange

Plus 5 sub-resources on every deal:

client.deals.comments.list(deal_id)        # threaded discussion
client.deals.assignments.create(deal_id)   # assign reviewers
client.deals.doc_requests.create(deal_id)  # request missing docs
client.deals.timeline.list(deal_id)        # full activity history
client.deals.users.search(q="jane")        # find team members

Retry behavior

Method	Rate limit (429)	Server error (5xx)	Connection error
GET / DELETE	Retry	Retry	Retry
POST / PUT / PATCH	—	—	Retry

Exponential backoff with jitter. 0.5 s base, 30 s cap. Honors Retry-After.

We ship fast

This SDK is actively maintained by the LendIQ engineering team. We release weekly, respond to issues within 24 hours, and treat SDK quality with the same rigor as our core platform.

If something isn't right, open an issue. We'll fix it.

License

MIT — use it however you want.

Get your API key and start analyzing statements in minutes.

Built with care by the LendIQ team.