Reads invoices, drafts pages, checks schedules and answers questions across every part of the platform. Works in chat, on events or on a schedule, with limits you set. Bring the AI you already use.
The AP Clerk reads each invoice, matches the PO, codes it and posts or escalates. The Submittal Tracker follows every submittal and nudges when one is overdue.
In Studio, Aquifer AI drafts the page from a sentence and labels every change it makes. In Canvas, ask about the page you are reading and it answers with your filters applied.
Search across your systems and documents from one place, with the index kept in your own datastore. Ask which RFIs are open on a project and who owes the answer.
Query your data from ChatGPT through an Action, or from Claude and Copilot through an MCP server. Build a custom GPT on it for your project managers.
The same part of the platform, used by the owner, the general contractor and the subcontractor. One example each.
The AP Clerk reads each contractor pay application, codes it to project and funding source and routes it for approval. The Associate Project Manager spots a slipping milestone and notifies the program manager.
The AP Clerk matches each sub and supplier invoice to its commitment, codes it to job and cost code and sends it to the PM. The Schedule Analyst flags critical-path risk and drafts a fix for the PM to approve.
The AP Clerk codes supplier bills to jobs. The office asks Copilot which pay application is still unpaid and where retainage stands, and gets the answer from Aquifer.
AP Clerk, Associate Project Manager, Data Quality Monitor, Submittal Tracker, Schedule Analyst and Support Agent. Each does one job, and the Support Agent answers questions inside the app.
Upload a PDF and the AP Clerk reads it, matches the purchase order and codes the GL from the vendor's history. It posts within your threshold or asks for approval.
The Associate Project Manager spots delays, assesses the impact and notifies the PM. The Submittal Tracker follows each submittal and nudges when one is overdue.
The Schedule Analyst flags risk on the critical path and drafts fixes for the PM to approve. It never changes the schedule itself.
When a source changes shape or a value lands outside the usual range, the Data Quality Monitor flags it. Schema drift and odd values are caught by an agent that watches the sources.
In Studio, Aquifer AI drafts and edits pages with every change labeled "Changed by Aquifer AI". In Canvas it answers questions about the page in front of you, filters included.
Ask an agent in chat, let it react to an event such as an invoice arriving, or run it on a schedule. The same agent, three ways to call it.
Every proposal waits for a person to approve, reject or edit it, and agents learn from those decisions. Agents cannot delete records or approve their own requests, every action is audited, and anyone can pause an agent instantly.
Run Aquifer AI on Anthropic, OpenAI or AWS Bedrock with your own keys and endpoints, under your agreements with the provider. Change the model without changing a workflow.
One search across your systems and documents, with the index kept in your own datastore. Find the RFI, the commitment and the drawing in one query.
Your project managers may already live in ChatGPT, Claude or Copilot. They can ask those assistants about your projects, and the answers come from the same connected data Aquifer AI reads.
Add Aquifer as an Action and ask about commitments, RFIs or the schedule from ChatGPT. Build a custom GPT on the same connection for the people who need it.
Claude and Microsoft Copilot connect through an MCP server and read the same records, with the same sign-in.
An assistant sees what the person asking is allowed to see. The search index stays in your own datastore, and nothing is copied out to a third party to answer a question.
Some companies have a provider agreement, a region requirement or a security review that decides which model they can use. Aquifer AI runs on the provider you choose.
Bring your own keys and endpoints. Usage runs on your account with the provider, under the data terms you already negotiated, and Bedrock runs in your own AWS account.
The agents, Studio drafting and search work the same whichever model is behind them. Move to a new model or provider without rebuilding anything.
Only the task at hand, under your provider's terms. The search index, your documents and your records stay in your datastore.
Start with one job, such as the AP Clerk on every invoice that arrives. Decide whether it works in chat, on an event or on a schedule.
The AP Clerk posts within the threshold you set and asks above it. No agent deletes records or approves its own request, and every action lands in the audit log.
Review what the agent proposes. Approve it, reject it or edit it, and the agent learns from the decision. Pause any agent instantly.
Pricing follows what you use. Every part of the platform can be taken on its own and kept in sync with the systems you already run. Talk to us about pricing for your company.
Talk to us →Can an agent change my books or my schedule on its own?
Only within the limits you set. The AP Clerk posts within your threshold and asks above it. The Schedule Analyst drafts fixes for the PM to approve and never changes the schedule itself.
What can an agent never do?
Delete records or approve its own requests. Every action is audited, and a person can pause any agent instantly.
Can I keep using ChatGPT, Claude or Copilot?
Yes. ChatGPT reaches your data through an Action, and Claude and Copilot through an MCP server. Custom GPTs work on the same connection.
Can we use our own model keys?
Yes. Aquifer AI runs on Anthropic, OpenAI or AWS Bedrock with your own keys and endpoints, so usage is on your account and under your agreement with the provider. Change the model or the provider without changing a workflow.
Where does the search index live?
In your own datastore. Search runs across your systems and documents, and the index stays with your company's data.
What does Aquifer AI do in Studio and Canvas?
In Studio it drafts and edits pages, with every change labeled "Changed by Aquifer AI" and multi-object changes held in Review. In Canvas it answers questions about the page you are reading, filters included.
How do agents get better?
Each time a person approves, rejects or edits a proposal, the agent learns from that decision. The AP Clerk also codes from the vendor's history.
We use a small number of first-party cookies to understand which pages visitors find useful. No advertising, no third parties. Privacy notice
