# Services — Abatetech AI

Everything between "we should do something about AI" and AI that works.
Five services, one method: understand the work first, put governance in place,
train the people, then build.

## AI Strategy & Workflow Assessment

- **Workflow mapping** — we walk through day-to-day operations with the people
  who do the work and find where AI makes a measurable difference.
- **Opportunity scoring** — scored on time savings, complexity, and risk. A
  process eating 10 hours a week across the team is where we start.
- **Tool selection** — recommendations that fit your workflows. No vendor
  kickbacks, no commissions.
- **Roadmap & budget** — a concrete plan with phases, costs, and expected outcomes.

## Training & AI Enablement

- **Getting past the chatbot** — structured prompting, context, iteration; AI
  as a thinking partner instead of a search engine.
- **Role-specific training** — accounting, marketing, operations each learn
  what matters to their job.
- **Verification & ethics** — AI output is a draft, not a source of truth.
- **Ongoing coaching** — check-ins and new use cases; the goal is self-sufficiency.

## Enterprise Platform Deployment

Platform-neutral, set up as enterprise tools with proper data handling:

- **Microsoft Copilot** — M365-embedded; readiness audits, tenant cleanup,
  licensing, deployment. If SharePoint permissions are a mess, we fix that first.
- **Anthropic Claude** — long-form analysis, document work, and the Claude
  Code / Agent SDK platform we use to build custom agents.
- **OpenAI** — broad ecosystem, Codex for code-heavy automation.

## Custom Automation & AI Agents

- Agents that connect to your systems (M365, CRM, databases) and do real work:
  triaging requests, reviewing contracts, producing reports.
- Power Automate / M365 flows where a full agent is overkill.
- Built, tested against your real workflows, deployed with guardrails and
  human checkpoints. The point: more output per person, not fewer people.

## AI Policy & Governance

- Acceptable use policy — approved tools, data boundaries, review expectations.
- Data classification — what can go into which tools, and what stays out of AI.
- Compliance — HIPAA, PII, funder requirements; the technology side set up right.
- Vendor evaluation — data handling, enterprise agreements, SOC 2, residency.

## What we don't do

Custom model training or fine-tuning · unsupervised customer-facing AI ·
replacing human judgment in critical decisions · promising magic.

Contact: https://abatetech.ai/contact