Nimble, formally The Data Company Technologies Inc., announced a $47 million Series B on February 24, 2026. Norwest led the round, with participation from Databricks and existing investors. TechCrunch reported that the financing brings Nimble’s total funding to $75 million. The company says it will use the money to develop multi-agent web search and a governed data layer that helps enterprise AI systems retrieve, structure and validate information from the live web.
Nimble is positioning itself as more than a conventional scraping API. Its Web Search Agents are intended to find relevant pages, navigate difficult sites, extract specified fields and return structured records that can feed applications, warehouses and AI agents. Whether that becomes indispensable infrastructure—or another usage-priced web-data service—will depend on accuracy, freshness, compliance and cost at production scale.
What Nimble raised
The round announced on February 24, 2026, has the following disclosed terms:
| Item | Details |
|---|---|
| Round | Series B |
| Amount | $47 million |
| Lead investor | Norwest |
| Strategic participant | Databricks |
| Returning investors named by TechCrunch | Target Global, Square Peg, Hetz Ventures, Slow Ventures, R-Squared Ventures, J-Ventures and InvestInData |
| Total funding | $75 million, according to TechCrunch |
| Stated use of proceeds | Research and development for multi-agent web search and a governed data layer |
TechCrunch reported that Nimble had more than 100 customers when the round was announced, including retailers, hedge funds, banks, consumer-packaged-goods companies, AI-native startups and Fortune 500 and Fortune 10 companies. Those customer and enterprise-mix figures are company claims, not independently audited customer or revenue disclosures. Nimble’s official announcement confirms the Series B and platform expansion.
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallCrashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minute#1 Best Overall
Why AI agents need a web-data layer
Language models can summarize and reason, but their training data is not a live, complete database of product prices, inventory, corporate filings, regulations or news. Enterprise workflows often need information that changes daily or hourly and sits outside internal systems.
- Pricing and availability: consistent product, price, currency, stock and timestamp fields across competing sites.
- Due diligence and financial research: repeatable collection of public company, market and counterparty information.
- Compliance and KYC: current public information rather than an infrequently refreshed internal snapshot.
- Market and brand monitoring: broad collection across sites, languages, regions and media sources.
- AI applications: machine-readable evidence instead of search snippets that an agent cannot reliably reuse.
The web is difficult input: layouts change, pages are rendered in JavaScript, anti-bot systems intervene, content is personalized or geo-specific, and a page designed for a person rarely exposes a stable schema. Nimble’s proposition is to absorb much of the retrieval, browser, parsing and maintenance work required to turn those pages into application data.
What Nimble actually sells
Nimble’s documentation describes several related interfaces rather than one monolithic product. Its API overview is available at docs.nimbleway.com/api-reference/introduction.
Web Search Agents
A user can select a pre-built agent or create a custom one, provide a URL, query, product identifier or other input, and request fields in a desired schema. Nimble says the agent handles retrieval, rendering, anti-bot challenges and extraction logic, then returns structured JSON. The workflow and agent types are documented at the Web Search Agent guide.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Compared with a fixed scraper, an agent is meant to describe the data needed rather than hard-code every selector. That can reduce site-specific engineering, but it does not eliminate errors, maintenance, access restrictions or legal review. “Auto-healing” and similar reliability language remains a first-party product claim.
Search
The Search API searches the web and retrieves content from results, with optional AI-generated summaries. It is suited to research and discovery, but search coverage, ranking, regional results and indexing gaps can affect completeness. Details are in Nimble’s Search documentation.
Extract
Extract retrieves clean HTML and structured data from a specified URL. It is closer to a direct page-retrieval operation than an open-ended research task.
Map and Crawl
Map discovers URLs and site structure. Crawl extracts information from multiple pages across a website. These operations are useful when the source domain is known and the task is repeatable.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Tasks and asynchronous processing
Long-running operations can be tracked through the Tasks API. Production systems should account for asynchronous completion, retries and partial failures rather than assuming every request is an immediate single response.
Agentic Studio
Agentic Studio supports custom-agent creation. “Any website” marketing should be read alongside practical limits such as login requirements, paywalls, consent flows, personalization, session state, rate limits and terms that restrict automated collection.
A concrete workflow: competitor-price monitoring
- A retailer defines the target sites and fields: product name, identifier, price, currency, availability, seller, source URL and retrieval time.
- The team selects a maintained agent or creates a custom agent and supplies product identifiers or search inputs.
- Nimble chooses an appropriate retrieval path, including browser rendering when a page requires JavaScript.
- The agent extracts the requested fields and returns JSON rather than a page intended only for visual reading.
- The retailer stores the record in its warehouse or application, preserving the URL and timestamp for review.
- Rules or analysts investigate exceptions such as a missing price, variant mismatch, currency discrepancy or sudden change.
That workflow illustrates the value of structure: stable field names and typed records are easier to compare and load into systems. Structure does not prove that a price is correct, current, complete or legally reusable.
What “real-time” means
Nimble’s documentation describes real-time web search and extraction, but the phrase should not be interpreted as a guarantee that every page is queried continuously or that data is current to the second. In practice, it generally means retrieval from the live web at request time or close to it, rather than relying only on a periodically refreshed index.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Fix the driver behind crashes, sound loss and screen glitches3Repair Windows errors before they cause bigger problemsFreshness depends on the endpoint, caching, crawl schedules, site availability, response latency and the customer’s own pipeline. A source can change after retrieval, disappear before downstream use or show different content by location, account state or device.
What “verified” or “validated” should mean to a buyer
Nimble uses language such as verified, validated, governed and decision-grade. The available product material establishes that the service processes and structures web results; it does not establish an independently tested accuracy rate or a universal verification standard.
Before relying on those terms, ask:
- Are values checked against multiple sources or only formatted?
- Are data types, units, currencies and field formats validated?
- Are source URLs, page snapshots and retrieval timestamps retained?
- Are conflicts, missing values and inferred values distinguished?
- Are confidence scores, audit logs and correction workflows available?
- Can reviewers inspect the original content behind a field?
A clean JSON response can still contain a stale price, duplicate product, wrong variant, incorrect conversion or value extracted from the wrong page section. For financial, compliance or other consequential decisions, Nimble should be treated as a collection component, with independent controls and human review.
Why Databricks matters
Databricks’ participation links Nimble to the enterprise data-platform market. TechCrunch also reported partnerships involving Databricks, Snowflake, AWS and Microsoft for deployments that combine external web data with internal sources.
The strategic model is straightforward: Nimble retrieves outside information, structures and processes it, and sends it into a warehouse, lake, application or agent workflow. The CEO has described an approach in which customer data can remain in the customer’s existing infrastructure, but retention, isolation, residency and security depend on the specific architecture, product tier and contract. Buyers should verify those terms rather than assume a universal deployment model.
Pricing and economics
Nimble’s public pricing is usage- and endpoint-dependent. The figures below were listed in the supplied pricing material and should be rechecked before a purchase because they can change.
Rank #4
| Public signal | Listed figure and qualification |
|---|---|
| Trial | 5,000 free web pages |
| Agent API | Headline starting price of $3 per 1,000 pages scanned; managed Web Search Agents add 10% according to the pricing page |
| Search API | Headline starting price of $5 per 1,000 search inputs on the public pricing page |
| Extract, Crawl and Map | VX6 $0.90, VX8 $1.30 and VX10 $1.45 per 1,000 URLs on the public pricing page |
| Data Services | Startup $2,500/month, Scale $7,000/month, Professional $15,000/month, all billed annually; Enterprise custom |
The detailed documentation lists different pay-as-you-go figures by driver: community/custom WSA-6 at $0.99, WSA-8 at $1.45 and WSA-10 at $1.60 per 1,000 URLs; maintained WSA-6M at $1.08, WSA-8M at $1.55 and WSA-10M at $1.75; Search at $1 per 1,000 inputs; AI-generated Answer at $4 per 1,000 inputs; and Residential Proxy API at $5.30 per GB. The discrepancy shows why a buyer should model the exact endpoint, driver and managed-service combination rather than quote one universal page price. See the detailed pricing table and the public pricing page.
| Data Services plan | Monthly price | Monthly page credits | Concurrent agents | Storage |
|---|---|---|---|---|
| Startup | $2,500, annual billing | 350,000 | 5 | 7 days |
| Scale | $7,000, annual billing | 1.2 million | 10 | 30 days |
| Professional | $15,000, annual billing | 3 million | 20 | 90 days |
| Enterprise | Custom | Custom | Unlimited listed | Custom |
Real cost can include multiple searches, extraction calls, JavaScript or stealth rendering, AI-generated answers, crawl operations, storage and support. Compare cost per completed business task—not merely cost per page—and include the engineering time saved, avoided browser infrastructure and cost of incorrect data.
Rendering, anti-bot and access trade-offs
Nimble documents three broad driver levels: VX6 for standard non-JavaScript pages, VX8 for JavaScript rendering and VX10 for JavaScript plus stealth capabilities. Web Search Agent equivalents are WSA-6, WSA-8 and WSA-10 for community or custom agents, and WSA-6M, WSA-8M and WSA-10M for Nimble-maintained agents.
- Static pages are generally faster and cheaper.
- Rendering consumes more resources and costs more.
- Stealth handling may improve access but cannot guarantee it.
- Sites can change challenges, fingerprints, login flows, CAPTCHA systems, rate limits and regional policies.
- Technical access is not permission to copy, republish or commercialize content.
Terms of use, robots directives, copyright, privacy obligations, authentication restrictions and contractual licenses vary by source and jurisdiction. A legal review is necessary for commercial collection and redistribution.
Who might buy Nimble—and who might not
Likely fit
- Teams monitoring many changing public sites without a large scraping-maintenance group.
- AI product builders that need schema-constrained, source-linked web data.
- Market-intelligence, pricing, compliance and due-diligence teams with recurring workloads.
- Organizations already using a warehouse or lake and seeking an external-data ingestion layer.
Poorer fit
- Businesses extracting a few stable pages where a simple parser is reliable and inexpensive.
- Workflows that require licensed, authoritative datasets or guaranteed historical completeness.
- Use cases involving authenticated sources Nimble cannot support.
- Organizations unable to accept variable usage costs or external processing.
- High-stakes decisions with no independent validation or human review.
How Nimble compares with alternatives
| Category | Strength | Trade-off |
|---|---|---|
| Search and answer APIs | Fast adoption for conversational research, citations and general answers | May offer less control over repeatable field-level schemas and site-specific crawling |
| Scraping and browser infrastructure | Maximum control over browsers, proxies, rendering and parsers | More engineering, monitoring and break/fix work remains in-house |
| Licensed data vendors | Contracted schemas, provenance, support and often historical coverage | Narrower domain coverage and potentially higher fixed cost |
| In-house pipelines | Control over proprietary logic, residency and long-term architecture | Requires browser, data-engineering and maintenance expertise |
Nimble’s differentiation is therefore workflow abstraction: it combines search, extraction, crawling, agent generation, rendering and proxy capabilities behind common interfaces. The company still has to demonstrate that this abstraction delivers acceptable accuracy, freshness, latency, compliance and unit economics for each workload.
Technical starting point
The documented API base URL is https://sdk.nimbleway.com/v1. Requests use a bearer token:
Best Value
Authorization: Bearer YOUR_API_KEYA documented extraction request follows this pattern:
curl -X POST 'https://sdk.nimbleway.com/v1/extract'
--header 'Authorization: Bearer YOUR_API_KEY'
--header 'Content-Type: application/json'
--data-raw '{
"url": "https://www.example.com"
}'Endpoint names, SDKs, parameters, limits and authentication details can change; consult the SDK examples and current API documentation before implementation.
Questions to answer before signing up
- Is the required information public, authenticated, paywalled or personalized?
- Do you need live retrieval, scheduled crawling, or historical snapshots?
- What fields, types, currencies, languages and regions must be normalized?
- What latency, throughput, concurrency and retry behavior are required?
- Which pages need JavaScript, stealth or geo-targeting?
- Must source URLs, timestamps, raw content and audit logs be retained?
- How will missing, conflicting or inferred values be reviewed?
- What retention, residency, PII, access-control and customer-isolation terms apply?
- What is the cost per completed workflow after rendering, answer generation and storage?
- Do source licenses and terms permit collection and downstream use?
The investment thesis
Nimble is betting that enterprise AI’s next bottleneck is dependable external data, not simply access to another language model. A managed retrieval layer could be valuable if it reduces the browser, proxy, parser and maintenance burden while delivering evidence that downstream systems can audit.
The unresolved questions are measurable: how often agents extract the right field, how quickly they recover from site changes, how complete search coverage is, how costs behave under retries and rendering, and how customers can prove lawful use. The $47 million round gives Nimble resources to pursue that infrastructure position; it does not, by itself, validate the company’s accuracy or reliability claims.
Recommended Free Tools
The Bottom Line
Nimble’s Series B supports an ambitious attempt to turn the changing public web into a governed input for enterprise AI. Its value is less the ability to fetch a webpage than the promise of repeatable, structured retrieval. Buyers should test representative sources, preserve provenance, model full workflow costs and require human oversight for consequential decisions.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
