Specialty Tokens
DEContact us
Training · Financial Operations

AI Workshops for Finance Teams — Agents in Microsoft Teams or Slack, Security First

Finance runs on systems chat tools cannot see. We train finance and fund-operations teams on their own ledger, ERP, and chat stack — Microsoft Teams or Slack — and stay on as advisors through to full transformation.

Finance teams are the most system-bound people in any company — and the least served by generic AI training. The ledger, the ERP, the payment runs, the reporting pack: none of it is visible to a chat window. So the course ends, the certificate arrives, and the month-end close is exactly as manual as before.

AI workshops for finance teams, built on the systems finance actually uses

Specialty Tokens runs hands-on AI workshops for finance, accounting, and fund operations teams. Before the session we connect your approved AI tools — Microsoft Copilot, ChatGPT, Claude — to the systems of record through the Model Context Protocol (MCP): the ERP, the accounting core, the reporting stack. The exercises are your own close: reconciliation prep, invoice and AP triage, intercompany checks, the pack someone assembles by hand every month.

The workshop happens where the team already works. For Microsoft houses that means Copilot agents inside Microsoft Teams; for Slack companies, the same pattern runs as Slack agents. Either way, the agent reads from the ERP and accounting core, prepares drafts — a reconciliation, a payment proposal, a variance commentary — and posts them into the channel for a named person to approve. No new tool to adopt; the approval happens in the flow of work.

Much of finance also runs on systems no connector catalogue covers — a proprietary ledger, an in-house data warehouse, a niche fund-administration platform. The Model Context Protocol was designed for exactly this: we build custom MCP connectors for private systems, running inside your perimeter, so the AI tools you have approved can reach the systems your team actually uses.

We also train teams to think in specialised AI coworkers — one per desk, not one bot for everything. The reconciliation desk gets an agent scoped to bank and ledger data; accounts payable gets one that triages invoices; treasury and investor relations get their own, each limited to that role's systems and permissions. Specialisation is what keeps agents useful and auditable at the same time.

Security is the operating pattern, not a promise: read freely, write behind approval, log everything. Financial data stays inside your perimeter, nothing trains public models, and every agent action carries its inputs, approver, and timestamp — the control structure your auditors already understand, the same one behind our AI for fintech work.

The people teaching have run this kind of work themselves. Thomas Schlossmacher worked at a family office (Corecam) and built revenue-operations software for ERP systems; Nik Redl built financial infrastructure. Both have served companies on financial operations with security in mind — see Talentir, a payouts company that designed AI-native financial operations early, and Speedinvest, where enablement and connected fund workflows landed together.

Workshops run in two tracks — a CFO and leadership session on what to automate and govern, and an operator session where the team builds automations it keeps. Most end with a ranked shortlist of processes worth automating; AI agent development picks up at the top of the list.

We help finance organisations go AI native — and the workshop is only the entry tier. We also advise CFOs and finance leadership on the AI operating model, and run embedded AI transformation engagements that redesign the finance function end to end. We are based in Vienna and work across DACH, the UK, and the rest of Europe, in English or German. For deal teams and funds, see AI training for private equity; the umbrella format is AI workshops & training.

What you get

Agents where finance already works

Copilot agents inside Microsoft Teams for Microsoft houses, Slack agents everywhere else — drafts prepared from live system data and approved in the flow of work, not in another tab.

An AI coworker per desk

Reconciliation, reporting, accounts payable, treasury, investor relations — each desk gets a specialised AI coworker scoped to its own systems and permissions.

Private and in-house systems included

Ledger in a proprietary tool? Portfolio data in a custom warehouse? We build MCP connectors for the systems no vendor catalogue covers — and they stay inside your perimeter.

Approval gates where money moves

Reads are free; writes are gated. Drafted postings, reconciliations, and reports queue for a named person to approve — and every action is logged.

A CFO track and an operator track

Leadership gets the judgment questions — what to automate, what to govern, what it costs. Operators build automations on their own month-end tasks and keep them.

Taught by people who have run financial operations

Thomas worked at a family office; Nik built financial infrastructure. In both careers, security was a requirement, not an afterthought.

Training, advisory, transformation

Workshops are the entry tier. We also advise CFOs on the AI operating model and run embedded transformations — through to agents in production, built with you, owned by you.

Why teams pick a partner over a platform.

Off-the-shelf AI stops at chat. Consultancies leave with the deck. We do neither.

Off-the-shelf AI

ChatGPT Enterprise · Glean · Copilot

  • Reaches your ERP, CRM & back-office
  • An agent platform your team can build on
  • Custom skills built around your workflows
  • The IP stays yours, not the vendor's
  • Works across Claude, ChatGPT & Gemini
  • Upskills your team to run it themselves
  • Embedded with your team until it works

Traditional consultancy

Decks, then exit

  • Reaches your ERP, CRM & back-office
  • An agent platform your team can build on
  • Custom skills built around your workflows
  • The IP stays yours, not the vendor's
  • Works across Claude, ChatGPT & Gemini
  • Upskills your team to run it themselves
  • Embedded with your team until it works

Specialty Tokens

Partner

  • Reaches your ERP, CRM & back-office
  • An agent platform your team can build on
  • Custom skills built around your workflows
  • The IP stays yours, not the vendor's
  • Works across Claude, ChatGPT & Gemini
  • Upskills your team to run it themselves
  • Embedded with your team until it works

Who we are

We get serious companies through the shift. Here's who does it.

We have a specific view of where software is headed, and we're watching it shift in real time. Our job is to wire AI into the systems you already run, then leave your team able to build on it — until it becomes one of your sharpest competitive advantages.

Nik Redl

Director

Engineer, operator and AI advisor.

Thomas Schlossmacher

Director

Builder, operator and educator.

Common questions

What do AI workshops for finance teams cover?

Two tracks, both on the company's own systems. The leadership track covers where AI carries weight in the finance function, what it costs, and how to govern it. The operator track is hands-on: the team connects approved AI tools — Microsoft Copilot, ChatGPT, Claude — to the ERP, accounting core, and reporting stack, and works on its own tasks: reconciliation preparation, invoice and AP triage, intercompany checks, and the reporting pack that gets assembled by hand every month. Participants leave with automations already running in their own environment.

Can AI agents run inside Microsoft Teams?

Yes. For companies on the Microsoft stack, Specialty Tokens builds Copilot agents that work inside Microsoft Teams: they read from the ERP and accounting core through Model Context Protocol (MCP) connectors, prepare drafts such as reconciliations, payment proposals, or variance commentary, and post them into the channel for a named person to approve. The team adopts no new tool — the approval happens where the conversation already is, and every agent action is logged.

What is a specialised AI coworker in finance?

An agent scoped to one desk rather than one bot for everything. The reconciliation desk gets an AI coworker limited to bank and ledger data; accounts payable gets one that triages invoices; treasury and investor relations get their own, each restricted to that role's systems and permissions. Specialisation keeps agents useful and auditable at the same time — narrow scope means grounded answers and a clean audit trail.

Do you only work with the Microsoft stack?

No. Microsoft Teams and Copilot are one common path; Slack is another — agents post drafts and take approvals in whichever chat tool the team already uses. The layer underneath is vendor-neutral: approved AI tools such as ChatGPT, Claude, Copilot, or Gemini on top, Model Context Protocol (MCP) connectors below. For private or in-house systems — a proprietary ledger, a custom data warehouse, an industry-specific portfolio tool — Specialty Tokens builds custom MCP connectors, so the AI tools a company has approved can reach the systems its finance team actually uses.

Do you also offer AI advisory and transformation for finance teams?

Yes. Workshops are the entry tier of a three-part engagement model: Training (workshops and enablement), Tech (integrations and agents in production), and Transformation (an embedded, end-to-end AI-native rollout of the finance function). Advisory runs through all of it — from helping a CFO decide what to automate, what to govern, and what to leave alone, to redesigning the close so agents carry the repeatable work and people keep the judgment.

How is financial data kept secure during a workshop?

The integration layer runs in the company's own environment, and access mirrors the permissions each participant already has. Financial data is not used to train public models. The operating pattern is read freely, write behind approval: any posting, payment proposal, or report an agent drafts queues for a named approver, and every action carries its inputs, approver, and timestamp. It is the control structure auditors already understand, applied to AI.

Do you run AI workshops for finance teams across Europe?

Yes. Specialty Tokens is based in Vienna and runs workshops for finance and fund-operations teams across Europe — DACH, the UK, and the rest of the continent — on site or remote, in English or German. Clients range from finance functions in mid-market companies to fund operations at private equity and venture firms, family offices, and hedge funds.

Bring one month-end task to a workshop

Book a scoping call. Tell us which close task eats the most hours, and we will show you how it runs with an agent, an approval gate, and an audit trail.

Book a scoping callTalk to the team