Agent Personas, new in Uncountable, let you give Bodie a specific job and put it on a schedule. One persona can clean up inconsistent data every Monday. Another can draft your weekly lab report. Each one runs on its own, and each one waits for your approval before anything in your data changes.
Much of the work that keeps an R&D platform useful is repetitive. Someone has to notice that "Viscosity" and "VISCOSITY" are the same measurement, or summarize what ran last week. Agent Personas hand that work to an AI agent built for the task.
What an Agent Persona is
An Agent Persona is Bodie with a defined role. Each persona has two parts:
- A description that says what the persona does, in one line. Data Cleanup Bodie, for example, "finds inconsistent data and proposes bulk modify jobs for the user to approve."
- A system prompt that sets out its job in detail: what it is responsible for, what to look for, and how to act.
Uncountable ships system personas such as Data Cleanup Bodie, and you can write your own. Once a persona exists, you create agents from it: give the agent a name, pick the persona, add instructions, and set a schedule.
Example: a weekly duplicate cleanup with Data Cleanup Bodie
Lab data arrives from many sources, so the same measurement often ends up under several names. Data Cleanup Bodie is built to find that drift: duplicates that differ in spelling, case, unit or language, information buried in a name that belongs in a structured field, and inconsistent units.
Here is how one agent built on it works:
- Create the agent. Name it "Weekly output cleanup," choose the Data Cleanup Bodie persona, and give it one instruction: merge duplicate outputs where the condition is in the name.
- Schedule it. Set it to run every Monday at 9:00 AM. It joins your other agents in the Agents list, where you can see each one's persona, schedule and status.
- Let it run. On Monday, the agent read 600 output names across 6 listings and found 3 duplicates.
- Review what it proposes. It returned one proposed job per duplicate:
The first job shows how the agent handles a condition stuck in a name. "Capacity at 100 Cycles (mAh)" becomes "Capacity (mAh)" with "100 cycles" recorded as its condition, so the data is searchable and comparable again.
Nothing changes until you press Start
Agents propose; you decide. Data Cleanup Bodie changes no data on its own. Every fix it finds becomes a bulk modify job in the Proposed phase, and it stays there until you open it, check the before and after, and press Start.
Agents also run as you. Every run uses your permissions and connections, so an agent can read and change only what you can. It never has wider access than the person who created it.
Write your own persona in a few sentences
Creating a persona takes a name and a system prompt. A Weekly Report Bodie, for instance, needs one line: "You write the Monday lab report: what ran last week, what passed, and what needs attention."
From there, the agents you schedule can cover the routine work around the lab:
Agents run in the cloud, so they keep working whether or not anyone is logged in.
See Agent Personas in action
Agent Personas put Bodie to work on the jobs your team repeats every week, with your permissions and your approval at every step. Watch the demo video, or request a demo to see how agents can work for your lab.
.png)
.png)
.png)
