# Datasmarts > AI intake for law firms: a firm-branded assistant installed on the website a firm already > has, which answers the question a visitor types after hours, writes it up as a memo on the > firm's letterhead, and books the consultation. Also an AI automation consultancy for > businesses with a repeating process that costs real hours. ## About Datasmarts is an AI automation consultancy working remotely with clients in the United States and Europe. Its product is AI intake for law firms, and the wider practice is finding the process that costs a business the most hours, building the system that removes it, and measuring the result against the number it started at. Engagements begin with a paid assessment that produces a ranked shortlist with a cost against each candidate process, including the ones not worth automating. One person does the work: Jesús Martínez, founder, 14 years in software and AI engineering. - Website: [datasmarts.tech](https://datasmarts.tech) - Contact: [jesus@datasmarts.tech](mailto:jesus@datasmarts.tech) - Who you would be working with: [About](https://datasmarts.tech/about) The site is bilingual. Every page below has a Spanish counterpart at the same path under `/es/`, for example `https://datasmarts.tech/es/services`. ## AI intake - [AI intake](https://datasmarts.tech/ai-intake): The product. A firm-branded intake assistant installed on a law firm's existing website, with no rebuild. - [AI intake for law firms](https://datasmarts.tech/ai-intake/law-firms): What the assistant does at 11pm, the compliance guardrails an attorney approves before launch, who the engagement fits and who it does not, and a client-reported funnel from the first 2.5 weeks after launch at a practice that is not named. - [How AI intake works](https://datasmarts.tech/ai-intake/how-it-works): The five steps between a question typed at 11pm and a consultation on the calendar: the answer inside the guardrails, the memo on the firm's letterhead, urgency triage, and the booking. Reported at 20 to 90 seconds from question to memo. - [AI intake compliance](https://datasmarts.tech/ai-intake/compliance): How the assistant stays outside the practice of law. Outcome predictions are banned at the prompt firewall rather than discouraged in a style guide, the firm's disclaimer closes every answer verbatim, an attorney approves the written rules before launch, and every memo reaches the firm by email. - [AI intake pricing](https://datasmarts.tech/ai-intake/pricing): What every engagement includes, the four things that move the number, and what can be added once intake is working. No figure is published: the scoped quote follows a fifteen minute call. ## Services - [Services](https://datasmarts.tech/services): Four services covering AI agents and LLM systems, workflow automation, AI strategy assessments, and the data engineering underneath them. - [AI agents and LLM systems](https://datasmarts.tech/services/ai-llm-systems): Retrieval, extraction, and assistant systems that read your documents, answer questions, and trace every answer back to its source. - [Workflow automation](https://datasmarts.tech/services/workflow-automation): Mapping the process that runs on manual copying, costing what it takes today, then building the automation that removes it and measuring the difference. - [AI strategy and assessments](https://datasmarts.tech/services/ai-strategy): An inventory of manual processes, each costed from your own numbers, returned as a ranked shortlist saying what to automate first and what to skip. - [Data and backend engineering](https://datasmarts.tech/services/data-backend-engineering): Fixing the data layer underneath the automation: ingestion that fails loudly, query performance measured before and after, and numbers that agree. ## Case studies Clients are named where they agreed to be named; the rest are anonymized by descriptor and industry. Every figure is the one the work produced, and figures that are estimates rather than measurements say so. - [Case studies](https://datasmarts.tech/case-studies): Nine engagements and the numbers they produced. - [Legal intake consultation funnel](https://datasmarts.tech/case-studies/legal-intake-consultation-funnel): A single-shot question and answer assistant on a law firm site, with a compliance layer the attorney authored and the pipeline enforces. Client-reported, from the first 2.5 weeks after launch at a practice we do not name: roughly 100 visitor questions, about 33 booked consultations, about 5 signed matters. - [Political ad intelligence platform](https://datasmarts.tech/case-studies/political-ad-intelligence-platform): A four-stage LLM pipeline turning FCC contract filings into queryable weekly ad spend, with every dollar traceable to its source document. - [Agency automation cost migration](https://datasmarts.tech/case-studies/agency-automation-cost-migration): Migrating JIA NOMADS' automation stack off a hosted no-code platform onto self-hosted n8n and Python, which cut monthly cost 90% and response times from about 3 minutes to under 20 seconds. - [Legal lead operations reporting](https://datasmarts.tech/case-studies/legal-lead-operations-reporting): Re-engineering a manual weekly lead report with n8n and AI-powered validation. - [Restaurant AI operations assistant](https://datasmarts.tech/case-studies/restaurant-ai-operations-assistant): A bilingual operations assistant answering daily staff questions about recipes and procedures across Padrino's Cuban Cuisine, a five-location restaurant group. - [Enterprise database re-engineering](https://datasmarts.tech/case-studies/enterprise-database-reengineering): Founder experience from a prior enterprise engagement, not a Datasmarts engagement: re-engineering the query layer of a critical data process. - [Creative agency multi-agent platform](https://datasmarts.tech/case-studies/creative-agency-multi-agent-platform): A master agent reads high-level requests arriving in Slack and delegates them across eight specialized assistants running on nested n8n workflows, built for Crescendo Creative. This engagement was never instrumented, so it reports what was built and no before-and-after figure. - [Guest messaging RAG rebuild](https://datasmarts.tech/case-studies/guest-messaging-rag-rebuild): Rebuilding the retrieval layer behind Cortado's AI guest messaging agents from first principles, consolidating property data into one vector store and moving ingestion off a monolith, which cut agent response latency from minutes per response to under a minute. - [Contextual AI assistant MVP](https://datasmarts.tech/case-studies/contextual-ai-assistant-mvp): The agentic-orchestration backend behind Tomo AI, JIA NOMADS' contextual assistant for tasks, email, and calendar, carried from a messaging-first prototype to a Flutter iOS MVP. No product-side figure was recorded, so the study counts what was built rather than inventing one. ## Insights - [Insights](https://datasmarts.tech/insights): Notes on choosing which process to automate and how these systems are built and measured. - [How to decide which process to automate first](https://datasmarts.tech/insights/which-process-to-automate-first): A ranking method for choosing the first automation: count the hours, price the errors, check the process is stable, and rule out the tempting ones. - [Migrating an agency from Make.com to self-hosted n8n](https://datasmarts.tech/insights/no-code-to-self-hosted-n8n): When a hosted no-code platform stops earning its price, and how to leave one safely. ## Reference - [About](https://datasmarts.tech/about): Who you would be working with, and whether the engagement is a fit. - [FAQ](https://datasmarts.tech/faq): What AI automation is, how pricing and engagements work, and what happens to your data. - [ROI calculator](https://datasmarts.tech/roi-calculator): Estimate the hours and annual cost a repetitive task is taking, using conservative assumptions listed on the page. Figures in USD. - [Contact](https://datasmarts.tech/contact): Describe the process that is costing you time and get a reply from the founder within one business day. - [Privacy policy](https://datasmarts.tech/privacy-policy): What the site collects, who processes it, and how to withdraw consent.