Freeport Metrics

Freeport Metrics is an American-owned and managed offshore software development company based in Portland, Maine with a Polish office in Warsaw. Founded by Andrew Gauvin, a veteran digital product architect with over 25 years in the field, the company specializes in generative AI integrations for startups and enterprises. The firm has completed 125 projects on time and on budget over 15 years in the market, working across fintech, healthtech, insurance, logistics, and food tech verticals. Their positioning centers on pragmatic, cost-effective AI solutions rather than generic chatbots—moving from proof-of-concept to production systems that solve operational problems.

Services and capabilities

Freeport Metrics offers eight distinct AI solution packages, each built around specific operational needs. Their telephony agent automates inbound and outbound calls, integrates with Google Calendar API for scheduling, and can query or update database records with user consent. Calls under 10 minutes cost roughly $0.50. The task-doing assistant goes beyond Q&A—it executes workflows, updates CRM records, scans invoices, processes orders, and accepts voice, text, or image uploads. The pattern investigator analyzes large or unstructured datasets to surface fraud, anomalies, and performance gaps; it ingests data through custom anonymization and outputs reports, alerts, or triggers other agents based on set criteria.

Retrieval augmentation feeds the AI confirmed data sources (internal documentation, knowledge bases, ticketing systems, CRM platforms) so it answers from verified information rather than hallucinating. Embeddings update automatically on schedule or via manual trigger. Human prompting provides real-time contextual guidance during calls or meetings—it listens for pauses, offers tips and warnings, runs background checks, and generates after-action reports. Responses take 1-2 seconds for simple tasks, 6-7 seconds when executing actions. Costs for 20-35 minutes of call assistance run roughly $4.70 using OpenAI's standard model or $1.27 with a smaller variant.

The versatile parser extracts structured data from documents with varying layouts, watermarks, or handwriting—invoices, bank statements, research tables, grades. It doesn't store data; all is encrypted in transit. Azure Document Intelligence and OpenAI power the scanning ($0.01 per page) and parsing ($0.0025 per 100,000 input tokens). Voice controller handles voice activity detection, speech-to-text, and text-to-speech with customizable sensitivity thresholds and calibration to background noise; it integrates with web applications via TypeScript or JavaScript. The document reviewer leaves timestamped, signature-marked comments in Word or Google Docs using domain-specific prompts and retrieval augmentation, ensuring consistent feedback without modifying the original text.

Notable work

Animal Ethics engaged Freeport Metrics to build an AI chatbot reflecting their complex philosophy on sentient welfare within eight weeks. The client's core challenge: create an assistant that wouldn't appear biased to people holding opposing views. The team established a feedback loop with the organization to identify biases, test Q&A, and refine prompts. They shipped a microservices architecture using OpenAI embeddings, Claude, a database, a document importer, and an API gateway. The result: chatbot response accuracy reached 70% through iterative refinement, delivered in the constrained timeline. Client quote: "They built a flexible and scalable solution that addressed our biggest problem—overcoming pervasive biases that reflect common human misconceptions. They made sure we were part of the process every step of the way, and got it all done in lightning time." Leah McKelvie, co-founder, also noted.

Other projects span healthcare (building systems to prevent fraud and waste for Smile Train), entertainment (VoyceMe scaled from web to mobile, securing $10 million in funding), and generative AI product builds (Gaide.me, an AI-powered travel planner; Keepler, a relationship-first dating app). The firm has also worked with CashStar's co-founder on a local food platform and serves fintech, insurance, and logistics sectors. Clients consistently cite proactive problem-solving—anticipating roadblocks and proposing solutions before they surface.

How they work

Freeport Metrics employs a hands-on, collaborative engagement model. The Animal Ethics project illustrates their approach: two-week discovery and proof-of-concept phase, followed by six-week sprint cycles. They use agile development with dedicated project managers, designers, and developers who invest time learning the client's business before proposing solutions. Communication appears seamless—clients describe them as thinking ahead about constraints and potential issues. They ship fast using tools like Google Stitch and Claude Code to accelerate prototyping and software delivery. Engagement models include dedicated teams, staff augmentation, and fixed-price projects. Team composition typically spans user/business/technical discovery, architecture consulting, UX/UI design, and Agile development.

Team and credentials

Andrew Gauvin founded and leads the company; his LinkedIn is referenced on their site. The team size is not explicitly stated in the source material. Freeport Metrics holds an American business license (Portland, Maine address: 254 Commercial Street, Portland, Maine 04101, USA) and operates a Polish entity (Antoniego Józefa Madalińskiego 73/75, 02-549 Warszawa). They work with OpenAI and Azure Document Intelligence; data is encrypted at rest and in transit, and OpenAI data is not used to train models. The company demonstrates technical depth across Jupiter Notebooks, BrowserStack, Figma, Azure DevOps, Go, TypeScript, Java, Claude AI Chat, and OpenAI Embeddings. Client testimonials come from co-founders and CTOs at Animal Ethics, Keepler, Smile Train, Zipline, Forager, FoodTrux, Community Health Options, and Solomon Leadership Program.