AI Solutions

AI built into how your team works.

We design, build, and deploy custom AI tools and automations that remove manual work from your operations. From a first pilot to a production system your staff relies on every day, one team owns the outcome.

What it is

What AI Solutions means at New Block

AI Solutions is the practice of applying large language models, automation, and data pipelines to specific jobs inside your organization: answering resident questions, routing support tickets, drafting first-pass documents, reconciling records, or surfacing what changed in a dataset overnight. The goal is measurable time saved and fewer errors, not a chatbot for its own sake.

We build on frontier models, including Anthropic's Claude, and on the systems you already run. That means integrations with your CRM, ticketing, document stores, and internal databases rather than another disconnected tool your team has to remember to open.

For public agencies and enterprises, we design for the constraints that matter: data handling, access control, audit logs, human review on anything that reaches the public, and accessibility of every interface we ship.

What you get

What AI Solutions includes

01

AI opportunity assessment

A fixed-scope review of your workflows that ranks automation candidates by hours saved, risk, and data readiness, and ends with a build plan.

02

Custom AI assistants

Internal or public-facing assistants grounded in your own documents and policies, with citations, guardrails, and escalation to a person.

03

Workflow automation

Multi-step automations that move work between systems: intake to routing to response, with approvals where you want them.

04

Document and data processing

Extraction, classification, and summarization across PDFs, forms, emails, and legacy records at volume.

05

Integrations

Connections to your CRM, ERP, ticketing, and databases through APIs and the Model Context Protocol so AI acts on live data.

06

Training and governance

Usage policies, staff training, evaluation sets, and monitoring so the system stays accurate after launch.

How we deliver

One accountable team, start to finish.

  1. Discover

    We map the workflow end to end, measure current time and error rates, and confirm what data exists and who can access it.

  2. Prototype

    A working prototype on your real data within weeks, tested by the people who will use it, before any long build begins.

  3. Build

    Production engineering: integrations, permissions, evaluation suites, logging, and an accessible interface.

  4. Launch

    Staged rollout with staff training, documentation, and a clear human review path for anything customer or public facing.

  5. Improve

    Monthly review of accuracy, usage, and savings, with model and prompt updates as your needs and the technology change.

Every engagement starts with a conversation, not a proposal.

Book an Intro

Proof

We build with the tools we write about. Our team was in the room at Anthropic's Code with Claude Extended 2026, and we publish practical guides on putting Claude to work in small businesses and agencies.

Read our field notes

Questions

Frequently asked about AI Solutions

What kinds of AI projects does New Block take on?

Internal assistants grounded in your documents, public-facing service assistants with human escalation, workflow automation between systems, document and data processing at volume, and AI features built into products. We scope each engagement around a specific job with a measurable outcome.

Do we need a data team or in-house engineers?

No. We handle the engineering, integrations, and evaluation. We do need a point person who knows the workflow and can give feedback on prototypes. If you have IT or security staff, we work with them on access and review.

How do you handle sensitive or regulated data?

We design access control, logging, and data handling around your requirements from the start. Sensitive workloads can run under enterprise agreements that exclude your data from model training, and anything that reaches the public gets a human review path. We document the setup so your counsel and IT can review it.

Which AI models and platforms do you use?

We build on frontier models, including Anthropic's Claude, and choose the platform based on your data, budget, and existing systems. We are not tied to a single vendor, and we design so you can switch models later without rebuilding.

How long before we see results?

Most engagements produce a working prototype on real data within the first few weeks. Production rollout depends on integrations and review requirements. We define the success metric, usually hours saved or response time, before the build starts and report on it monthly.

Find the first workflow worth automating.

Bring one process that eats your team's time. We will tell you whether AI can take it and what it would take to build.

Book an Intro