Build on Prscnt

Prscnt is the orchestration and intelligence layer for a talent-management agency. It already knows how to price a deal, vet a brand, pick the right creator, draft outreach in your voice, run the deal funnel, invoice and chase payment, and redline a contract. This documentation is for builders who want to put that brain behind their own tool — a custom manager portal, a no-code app, an internal agent — without rebuilding any of it.

It is product-agnostic. The outcome is the same whether you run on Softr + Airtable, Retool + HubSpot, a custom React app + Salesforce, or a spreadsheet and a cron job. Everywhere these docs show a specific tool, read it as one row of a table — swap in yours.

The mental model: four layers

Every integration, no matter the stack, is the same four layers:

YOUR UI Softr · Retool · custom app · Slack task + context ORCHESTRATION AGENT the LLM that composes — in Claude, or your runner Prscnt MCP intelligence + pipeline Your connectors your keys, your side intel read / write Prscnt intelligence + pipeline rate · reliability · spend · voice deals · campaigns · contracts Your data & tools HubSpot · Airtable · Salesforce Monday · Gmail · QuickBooks Prscnt holds pointers — never your data or credentials
Layer What it is Examples (pick yours)
UI Where your team clicks Softr, Retool, a custom web app, a Slack command
Data Your system of record Airtable, HubSpot, Salesforce, Monday, a Postgres table
Orchestration An LLM agent that decides what to do Claude with the Prscnt MCP connected (in-Claude, or the runner template)
Connectors The MCPs the agent composes Prscnt + your own (Gmail, QuickBooks, your CRM, …)

Prscnt is the orchestration layer plus a set of intelligence and pipeline tools. You bring the UI and the data; you keep your connectors. Prscnt never replaces them.

The one principle: zero connectors

Prscnt builds no third-party API clients and stores no third-party credentials. When work needs to touch your tools, Prscnt returns a proposed result (a draft, an invoice payload, a folder layout) and the agent executes it through your connector. Prscnt holds only thin pointers back to your records — a Prscnt deal linked to a record id + URL in your system — plus its own proprietary intelligence (rate corpus, brand reliability, partnership graph, voice profiles, deal pipeline, contract playbooks).

Your customer data and your credentials stay in your accounts. See Security & guardrails.

The one thing to know up front

There is no plain-REST endpoint. Every Prscnt tool is reached over MCP (JSON-RPC at POST https://mcp.prscnt.com/v1/mcp), which assumes an LLM in the loop choosing tools and shaping arguments. A raw button or automation can't call Prscnt directly and get an answer back.

That's a feature, not a gap: the LLM is the orchestration. The integration pattern is always your UI → an agent runner (LLM + the Prscnt MCP + your connectors) → write back to your data. The orchestration runner is a copy-paste starting point for that runner.

What you get for free

You build the surface your team sees. You never rebuild the intelligence.

Where to go next