A clinic co-pilot
Metabolic rate
Fat-burn efficiency
Heart rhythm variability
For companies and teams
Put your company name on Plexon and pre-load it with your own agents, personas, and skills. Share them with your clients in a form they can use but cannot read or copy. It all arrives through the official app, so there is nothing for your team to build or host.
Build a gold-standard persona for each role (Engineer, Designer, Analyst) and publish it to your company once. Every employee signs in and gets the same one: the same system prompt, the same agents, skills, and tools, the same defaults. No setup, no drift between machines.
Publish once
Curate the persona, push it to your company source, and it lands on every machine the next time the app opens. Change a rule or add an agent and everyone picks it up.
Role-scoped
Sort people into groups and give each group its own persona. Engineers get the engineering setup, analysts get theirs, from the same company source.
Not every company pays list price. We can set a custom plan for yours at a price we agree on, built around your own personas and flows. It appears on your company page in Plexon like a normal plan, with a link you can share with whoever needs to buy.
On your company page
The plan shows at plexon.ai/companies/your-name, beside the content your company ships. People subscribe right there.
A link to share
Each plan comes with its own Stripe checkout link. Hand it to an affiliate or a client and they subscribe on their own.
Package your best work as secret items. Your clients install them and use them like anything else. They cannot open the system prompt, read the knowledge base, or copy the files to hand to someone else. This is how a company shares its know-how without giving it away.
What the client gets
A working agent or persona. They chat with it, run it, and rely on its results, the same as any other item in their app.
What stays hidden
The prompt, the instructions, the knowledge base, and the files behind it. They never see the source and cannot pass it on.
Secret items cover agents, personas, and skills.
Your team opens Plexon and sees your name. The splash screen, the window title, and the header carry your company branding in place of ours.
A new hire signs in and your setup is already there. No checklist, no manual install. Hand someone a laptop, have them sign in, and they are working with your AI the same minute.
Pre-loaded and pre-activated. The items you mark as active turn on by themselves.
You keep one curated set of content. Push a change and it lands on every machine the next time the app opens. You curate. Plexon delivers.
When someone on your team signs in, Plexon can prove to your own backend that they belong to your company, and which group they are in. No call back to us at request time.
Keep the app focused on what your people actually use. Hide or lock features per company or per group.
Showcase
The overview deck for a live Plexon deployment, rebuilt for the web. The partner is a metabolic health company; its name is withheld and the figures shown are illustrative. Everything else is how the rollout works.
Metabolic rate
Fat-burn efficiency
Heart rhythm variability
Lower cost
than calling a frontier model raw
Right-sized models
A cheaper or local model for planning, a stronger one for execution. The company chooses.
Prompt caching
The stable context is cached, not re-billed on every turn of the conversation.
Results to disk
Large tool outputs are written to disk, not re-sent through the token stream.
Prompt caching
The stable context is cached, so it is not re-billed on every turn.
Auto context-condense
As a chat nears the model’s limit, older turns are summarized to keep it small.
Repeat-read cache
Read the same file or run the same search twice, and it reuses the first result until something changes.
Big results to disk
Large tool outputs are written to disk, not re-sent through the token stream.
Isolated subagents
Big jobs split across helper agents, each in its own memory. Only a short summary returns.
Right-sized models
Cheaper or local models for planning, frontier models for the hard reasoning. You choose.
Observation masking
Inside a helper, stale tool output is swapped for a placeholder so its window stays small.
Teams share a board
Agents coordinate through a shared board and mailbox, instead of one context holding it all.
Persistent memory
It remembers facts across sessions, so it does not re-learn them each time.
Self-improving skills
It saves a procedure once and reuses it, instead of re-deriving it later.
Code by symbols
It jumps to the exact symbols a change touches, instead of re-reading whole files.
Tidy memory
A background pass keeps saved notes compact and deduplicated, so recall stays lean.
Most levers are on by default; a few are opt-in.
Elena’s in the room. What do I lead with?
Core metrics
Metabolic rate
excellentFat-burn efficiency
limitationHeart rhythm variability
goodMetabolic rate is strong. Fat-burn has the most room, so lead the plan with Zone 2 training and shift the crossover point.
Consultation flow · RMR walkthrough · 5 steps
Likely drivers
Recommended protocol
Biological age
31
years
3 years younger than chronological
Consultation priorities
Patient follow-up
OnRuns on each new report
Connected tools
CRM, EHR, email, and messaging
Clinic ROI & growth · conservative model
Tests / month
18
Ramp 5 → 18Avg revenue / test
$240
Blended menuMembership conversion
25%
$179/moRetention · month 12
68%
StickyNet over 3 years+$130K from membership
Where to invest next
Year-1 net · after platform
$76,300
18 tests/mo · 25% membership · $890 tier
Sales
Sells the device & platform to facilities
3,214
Demos booked
287
Closed won
26%
No-show
Conversion by lead source · win rate
Dietitian
Plans meals on the live nutrition database
180
Active clients
142
Active plans
9
Reports due
Live consultation co-pilot. Fetches and renders the patient’s report, tracks the walkthrough, writes the notes.
Company API
Turns a clinic into a case study with the ROI math built in. Branded reports and email sequences.
Agent Browser
Qualifies leads, preps demos, and works the live pipeline. Drafts, never auto-sends.
HubSpot · Gmail · WhatsApp
Plans meals and tracks nutrition against a live database of foods, recipes, and client health data.
Nutrition DB · Excel
2
Marketplaces
4
Personas
22
Skills
4
Agents
18
Workflows
6
Dashboards
In plain language
A trigger wired to typed steps: connector calls, AI steps, conditions, approvals, waits. The engine runs locally, snapshots every run, and sends at most once.
Boards of live widgets fed by tools, workflows, or the AI: stat tiles, charts, tables, full HTML reports. Refreshed on a timer or in the background.
Any HTTP API becomes tools every persona can call. Import an OpenAPI spec or build endpoints by hand; OAuth, API key, or Bearer, with keys encrypted on the machine.
Any agent or workflow on a cron or a simple interval: nightly CRM syncs, morning digests, weekly ROI reports. It notifies the team and pauses itself after repeated failures.
Plain language,
Describe the change and Plexon rewires the workflow, redraws the board, or moves the schedule. The clinic team edits its own automation, no engineer required.
The company’s knowledge & data
EHR, CRM, 20+ biomarkers, and clinical protocols
Plexon personas
Doctor, Dietitian, Analyst, and Sales, built to order
Every clinic desktop
Runs locally, any model, your data stays yours
Plexon turns expertise into a digital workforce, on any desktop, with any model.
Tell us your brand, the content you want pre-loaded, and whether you need secret items or custom theming. We will set up your company and walk you through it.
Talk to us