Synthesised audiences
Distil interviews, support transcripts and analytics into personas an agent can think alongside.
AGENTS SYSTEMAgents that
Agents that
know your team.
Personas synthesised from your audiences. Workflows replacing manual ops. Runs grounded in your mind — every agent reads from memory, not the void.
v1.4 · macOS · Windows · Linux
synap.app/acme/agents
PERSONAIndie FounderShips fast, hates ceremony, reads in 3-min bursts.
WORKFLOWOn new release-notes.md → publishTrigger: file change · Steps: lint, summarise, post.
RUN#218 · 14s · $0.0412 steps · 4 tool calls · audit trail attached.
25+AGENTSprebuilt across product, content, ops
∞PERSONASsynthesised from any audience source
100%AUDITABLEevery step, every cost, logged
0BLACK BOXgrounded in files Claude can re-read
01 · PERSONAS, WORKFLOWS, RUNSThree primitives.
Three primitives.
Every automation.
Personas describe who. Workflows describe how. Runs prove what happened.
Steps that replace ops
Compose triggers, tools and prompts into a workflow that does what someone used to do by hand.
Every execution, traced
A run is the receipt: every step, every tool call, every prompt, every cost — readable later.
02 · WORKFLOWSTriggers fire, steps run,
Triggers fire, steps run,
audits compound.
File-change, schedule and manual triggers kick a workflow off. Each step writes a body trace into your Mind so the next run starts smarter than the last. Stop a workflow at any boundary; nothing is opaque, nothing is irreversible.
VS GENERIC AUTOMATION
Plumbing isn't intelligence.
Why agents grounded in your mind beat node-graph automation tools.
TODAY · ZAPIER-LIKE
Generic plumbing
Steps connect APIs but understand nothing about your team
No memory — every run starts from scratch
Audit trails are scattered across vendor dashboards
Cost surprises only show up on the next invoice
WITH SYNAP · AGENTS
Agents grounded in memory
Every workflow reads from your Mind — files, decisions, personas
Runs leave traces that the next run can build on
One auditable trail next to the docs the agent touched
Live cost meter and hard caps — no surprise bills
03 · TRUST & OBSERVABILITYAuditable
Auditable
by design.
Every prompt, every tool call, every byte of context — recorded and replayable.
Every call,
logged.
A run is a tree of steps. Every tool call, prompt, response and file mutation is captured — viewable, exportable, and re-runnable from any point.
Real cost in real time
Live token and API cost per step, per run, per agent. Hard ceilings stop runaway spend automatically.
Stop the agent at any boundary
Pause on file write, on tool call, on cost ceiling. Nothing irreversible without your hand on the brake.
Agents that
work your way.
Free for personal mode, Pro plan for teams.
Free · No credit card · No telemetry