Promptev App
The App is where a team builds and runs agents. Connect the systems you already pay for, describe an assistant in plain words, decide what it may do on its own — then put it in Slack, in WhatsApp, on your website or behind a URL. Everyone asks the same agent; nobody gets an answer from a document they could not have opened themselves.
2,500 free credits · no card · model calls go to your provider on your own keys
How it is arranged
Three levels, and you meet all three in the first ten minutes. A workspace is the wall around your company's data. A project is a room inside it with its own knowledge and tools. An agent is someone in that room who does a job.
One workspace per company. It holds your people and their roles — owner, admin, editor, viewer — your billing, your audit trail, and the choice of where everything runs. People who join through single sign-on land on the role you set as the default.
Finance, Support, Legal — whatever your teams are. A project carries its own connected sources, its own tools and its own members, so what a Support agent can reach is not what a Finance agent can reach, by construction rather than by prompt.
Described in plain words: "answer HR questions from the handbook", "review this month's invoices and flag anything odd". Give it the knowledge it may use, the tools it may call, the model you want, and the channels it should answer on.
Knowledge
Point a project at a Drive folder, a SharePoint site, a Confluence space, a Jira project, a Dropbox, a database — or upload files. Whatever a person could open, the agent can read for them, including scanned pages and spreadsheets.
Nadia, Finance Directormember of the finance group
"What did we pay Northwind in Q2, and is the renewal signed?"
You paid Northwind $184,200 across four invoices in Q2. The renewal was countersigned on 12 June at a 6% uplift, effective 1 July.
Tom, contractorproject member, no finance groups
"What did we pay Northwind in Q2, and is the renewal signed?"
I can't see finance records for Northwind. From the project wiki: the integration is live and the renewal was announced as signed on 12 June. For amounts, ask the finance team.
Why this matters. An index that retrieves first and filters later has already read the document; it leaks into summaries, "related" answers and follow-ups. Here the finance files are never fetched for Tom, and the agent says so instead of guessing.
Reach
Eight apps to approve, sixteen services inside them, and 198 tools between those services. The list is where agents start, not where they stop — three routes reach anything else you run, so the honest answer to “do you integrate with X?” is yes, without waiting for us to build it.
| App you approve | Services inside it | Counts as |
|---|---|---|
| Google Workspace | Drive, Docs, Sheets, Slides, Gmail, Calendar | 6 services |
| Microsoft 365 | OneDrive, SharePoint, Outlook mail, Calendar | 4 services |
| Dropbox | Files and folders, shared spaces | 1 service |
| Confluence | Spaces and pages | 1 service |
| Jira | Projects, issues, comments | 1 service |
| Azure DevOps | Repos, work items, wikis | 1 service |
| Slack | Channels, messages, files | 1 service |
| Conversations with your team and your customers | 1 service |
An internal service, a partner endpoint, a product with no ready-made connector: point Promptev at it, describe its inputs, and it becomes a tool an agent calls — under the same approval rules and the same audit trail as every built-in one. Not second-class.
Agents query your databases and return tables or charts. Read-only by default, and you choose which schemas and tables are even visible. Eight engines: PostgreSQL, MySQL, MariaDB, SQL Server, Oracle, SQLite, ClickHouse and Snowflake.
Attach a Model Context Protocol server you run or trust and its tools appear beside the built-in ones. It works in the other direction too: every project is itself an MCP server at its own URL, so an MCP-speaking developer tool gets that project's knowledge and tools with the caller's permissions intact.
Point your own agent, your editor or a script at a project and it gets that project's permission-aware knowledge and its governed tools — searching the knowledge base, calling a tool, calling an agent — scoped to what the key is granted. Exactly as your people do, no more.
{
"mcpServers": {
"promptev": {
"url": "https://app.promptev.ai/mcp/acme/ops",
"headers": { "Authorization": "Bearer pk_…" }
}
}
}Month-end close · runs Mondays 07:00
Orchestrations
Compose the agents you already have into a run: on a schedule, on a webhook, or when someone asks. Steps hand work to each other, a step that needs a person pauses instead of failing, and a run that was interrupted picks up where it stopped.
Channels
One agent, every door. The permissions, the sources and the audit trail are the same wherever the question arrives — a person in Slack and a customer on your website are talking to the same thing, with different reach.
The agent answers in a channel or a DM, in the thread it was asked in, as the person who asked.
For the people who never open another tool. Same agent, same permissions, same audit trail.
A hosted page for one agent — the fastest way to give a team something to use before anything is embedded.
A chat box on your own site from one script tag. In-page, not an iframe, and only on the origins you allow.
A callable URL per agent or orchestration. Post to it from anything, and it answers or starts the run.
Every project is an MCP server at its own URL, so your own developer tools get its knowledge and its tools.
The widget renders in your page rather than in an iframe, so it inherits your layout and your colours. It answers only on the origins you list, and it is the same agent your staff use internally — with a customer's reach, not a colleague's.
<script src="https://app.promptev.ai/embed.js"
data-agent="support-website"
data-theme="auto"></script>Approvals and audit
For every tool an agent can call you pick one of three: no approval, ask once, or always ask. Nothing that spends money or writes to your systems happens without the rule you set.
Approval neededE-mail · Slack · Inbox
The run waits here. Approve it and the agent carries on from the same step; reject it and the run records that it was refused, by whom, and when.
Privacy first, in the literal sense
Bring your own model keys, your own database, your own storage bucket, your own connector apps. Run it managed by us, in your cloud, or fully on your premises — and take your data with you whenever you like.
Credits
No seats, no per-user minimum, no plan to outgrow. Work costs credits; a workspace that has a quiet month pays for a quiet month.
Start with 2,500 free credits and no card. Buy credits when you want them, or switch on auto top-up so a busy week never stops mid-run. Receipts and a payments ledger sit in the workspace, and money notices go to every owner and admin.
A custom deployment in your own cloud account or on your own premises, with single sign-on, audit exports and support. Sized to the estate rather than to a seat count — and the route for data that has to stay in a particular jurisdiction.
What you get on day one
No add-on tier and no module to buy. Every workspace has all of it from the first login.
Every agent in a project: who built it, where it answers, live or draft.
Files, images and voice notes in; answers out with the documents they came from.
Connect a source or upload. Per-file sync, live progress, scanned pages read for you.
Connector tools, HTTP tools, database tools and your own MCP servers, side by side.
Teams of agents on a canvas. Schedules, webhooks, retries, pause for approval.
Approve in the App, by e-mail or in Slack. A request expires if nobody answers.
Owner, admin, editor, viewer, plus project roles. SSO groups become access rules.
Every change and every run, who did it and what happened, readable and exportable.
Slack, WhatsApp, the portal, a widget on your site, a webhook. One agent, every door.
Approvals, held runs, a connector needing attention, credits running low.
Pay as you go, optional auto top-up, receipts and a payments ledger.
Scoped keys per project, and the embed snippet for your own site.
Plain answers
The product your team opens. A workspace is the wall around your data; inside it, projects hold knowledge, tools and the agents that use them. Editors build agents in plain words, viewers simply talk to them, and every answer carries the documents it came from.
Yes, and without waiting for us. Sixteen services arrive ready, and three routes reach everything else: point Promptev at any REST or HTTP API and describe its inputs, and it becomes a tool; connect a database and agents query it in plain language; or attach any MCP server you run or trust. All three obey the same approval rules and land in the same audit trail as a built-in tool.
The person asking is part of the retrieval query itself, so a document they may not open is never fetched for them — not filtered out afterwards. Everyone talks to the same agent and each gets what they are entitled to, which is what makes it safe to put in front of contractors.
Managed by Promptev, in your own cloud, or fully on-premises. SOC 2 Type II. Model calls go to your provider on your own credentials — chat, embedding and OCR alike — so you can switch model or provider without changing what you built, and a self-hosted or OpenAI-compatible endpoint of your own works too.
Credits, not seats. Start with 2,500 free credits and no card, then buy credits from $5 when you need them and pay for what your agents actually do. Enterprise is an annual licence with a custom deployment in your own cloud account or on your own premises, plus single sign-on, audit exports and support — sized to the estate, not to a seat count.
Yes, in both directions. Each project is an MCP server at its own URL, so an MCP-speaking developer tool gets that project’s knowledge and governed tools with the caller’s permissions intact; and each agent or orchestration has a callable URL you can post to from anything you already run.
Start here
2,500 free credits, no card. Or talk to us about running the App where your data already lives.