We connect AI workflows to HubSpot, Salesforce, Microsoft 365, SharePoint and Holded — bidirectional read and write, with every operation audited and the firm’s confidentiality walls intact.
Firm knowledge already lives in HubSpot, Salesforce, SharePoint and the folder of the partner running the matter. An AI that doesn’t read that information or write back is one more chat window — the associate ends up copy-pasting, and the proposal sits outside the CRM until someone imports it by hand two weeks later.
We build the connector between your AI and the systems you already pay for, with role-based permissions mirrored from your AD or Workspace, idempotency so you never duplicate contacts or opportunities, and an audit log the responsible partner can review line by line. If the productised integration doesn’t exist, we build it — and we maintain it when the vendor quietly changes the endpoint.
Reads the account, contact, opportunity and notes history in HubSpot or Salesforce, drafts the proposal and stores it as an attachment on the opportunity — without touching fields it shouldn’t.
Keeps contacts, matters and time entries in sync between the commercial CRM and the legal or accounting PMS. Each change carries its origin, timestamp and an identifier so it can be reversed if needed.
Reads the shared inbox, links each thread to the right matter in the PMS, writes the activity note and creates the follow-up task — with the role of the user who would have done it manually.
The agent only sees documents the asking user could open themselves. We mirror the SharePoint or Google Drive permission matrix and apply it on every query — no back doors.
Pulls validated hours from the PMS, cross-references them with the client rate and creates the pre-invoice in Holded or your billing system — with the line editable before issuing.
Proposals are drafted outside the CRM in Word, saved in the partner’s folder and attached to HubSpot "when there’s time". Hours are exported from SharePoint to Holded by hand at every close. Every time a new AI tool is trialled, ops ends up opening another spreadsheet to link what the system wrote with what the CRM already had.
The agent reads the HubSpot opportunity, drafts the proposal from an approved template, saves it as an attachment on the right object and leaves a signed activity note. The connectors push validated hours from SharePoint to Holded in a nightly batch with automatic reconciliation. Agent permissions are inherited from the firm’s Microsoft 365 — an associate never sees documents from a matter they weren’t assigned to.
Mid-market CRM with broad APIs — a natural fit for sales agents and lead enrichment.
Enterprise CRM with fine-grained permissions — AI workflows that respect the data model.
CRM with strong adoption among Spanish SMBs — automations and agents at a contained cost.
Enterprise CRM/ERP suite in the Microsoft ecosystem — native fit with 365 and Power Platform.
Spanish cloud ERP widely adopted by SMBs — invoicing, expenses and reconciliation automation.
Modular open-source ERP — AI agents and workflows on top of sales, inventory and project modules.
Email, calendar and SharePoint as channel and context — triage, drafting and RAG over your inbox and files.
No. The rule is that the agent never has more permissions than the user issuing the query. We mirror the role matrix from the CRM, SharePoint or Drive and apply it on every read. If an associate asks about a matter they aren’t assigned to, the agent says it has no visibility — it doesn’t improvise with another associate’s data.
Every write is logged with the previous value, the new value, the logical user and the timestamp. We keep a queue of reversible operations: with one click you can undo any agent write within a defined window. And before going live we evaluate the connector against a CRM sandbox with anonymised real data.
No. The premise of the service is the opposite: we work on top of the stack you already have. If a productised connector for your tool doesn’t exist — common with sector-specific PMS — we build it against the API and maintain it. The rule is no forced migrations to get started.
Every write is idempotent: each operation carries a unique key the connector checks before executing. If the agent retries because the network failed, the operation is applied exactly once. For new entities we use domain, email and matching rules you approve before deployment.
We design for privacy from the start, human control, traceability, usage limits, permissioning and documentation. For sensitive processes, we help assess risk and applicable obligations under GDPR and the EU AI Act.
Every engagement is led personally by one of the partners. If there's a fit, you get a personal first read of your case within one business day — not a canned demo.
