Build with Lumopath

Your revenue engine.
Our plumbing.

Not another place for your team to live. Every KPI, every play, and every agent is available over MCP — so you can build your post-sales operating model inside your own stack, on a data layer you don't have to spend nine months building.

Book a technical review Request the MCP docs

Your stack

CRMInternal assistantBI toolSlack

Effort GraphLabeled activity · KPIs · plays · closed loop · agents

CRM · email · calendar · chat · support · product usage · transcripts · warehouse

The division of labour

You keep the logic and the interface. We keep the data engineering.

Lumopath providesYou build

Labeled activity and effort — every hour mapped to an account and an outcomeThe scoring logic that reflects how your business actually works

A catalog of KPIs, computed across 12 months of history on day oneYour thresholds, segment definitions, and what "good" means per tier

Plays with triggers, owners and SLAs — including the custom plays we build with youAgentic workflows to execute next best actions end to end

Also ours, and none of it yours

Connector upkeep as ten vendors change their APIs
Activity-to-account attribution
Effort labeling and duration scoring
KPI definitions drifting between teams
A security review for every new data source
Backfills when someone changes a field

The interface

Everything we compute, callable from anything you build.

KPIs

Query any metric in the catalog by account, segment, owner or period. Retroactive across 12 months, so a new metric has history the moment you ask for it.

query · list · compare

Plays

Read the library, read live fire state, create your own, and subscribe to fires as they happen.

list · read state · create · subscribe

Effort & activity

The labeled graph itself: who spent how long on what, mapped to accounts and outcomes.

query · aggregate · attribute

Agents

Invoke any agent with your own context and get the artifact back.

invoke · retrieve artifact

Plays × agents

How teams build on the Effort Graph.

The split is consistent: we fire the trigger and hand over the grounded context; you own the execution layer — your agent, your format, your surface.

Play or triggerWhat Lumopath deliversThe execution layer you build

Business Review DueEvery relevant activity, outcome and open commitment from the last 90 days, packaged as structured contextYour agent renders it into the deck format your team actually presents

Champion DepartureThe departure signal, the org map and the full relationship history behind itYour succession workflow — who gets assigned, what outreach goes out, in your tooling

Save Plan RequiredContributing risk signals, ranked, with the evidence behind each oneThe save plan in your format, routed through your approval path

Renewal Runway StartDecision-process map, value delivered to date, and open commitmentsYour renewal packet and the stages it moves through

Whitespace Cross-SellThe expansion signal, the category evidence and the account contextOpportunity creation and the brief your sellers actually use

Owner TransitionA structured record of everything the outgoing owner knewYour handoff checklist, and how it lands in your CRM

Health Score DropThe labeled inputs and the change that drove the movement, written back continuouslyThe score itself, and the objects and fields it lands in

Advocacy AskOutcome data and the relationship history behind the winThe case study, in your brand and your publishing flow

Detractor ResponseThe sentiment signal and the thread history that produced itThe response, in your tone and your escalation rules

Coverage Gap on BookWhat actually happened across each rep's book, assembled per personYour 1:1 format, and what you coach to

Your own health scoreLabeled inputs and 12 months of history to backtest againstThe scoring logic and weights that reflect your business

Your internal AI assistantEffort, KPI and play context over MCP — not a vector search over documentsThe assistant your team has already asked you for

Any custom workflowThe trigger, and the grounded context behind itWhatever you run it in — Slack, n8n, or your own services

On every row: the play fires automatically, the closed loop records whether it was done and inside the window, and your team's feedback tunes what we deliver next time.

How it gets builtEvery customer gets a forward-deployed data scientist.

They build the first agents with you, or set your team up to build them in your own environment — whichever way you want to work. Custom work isn't a services line item; it's how the layer is meant to be used.

See how plays work →

The questions we get from people who build.

Can't we just connect an LLM to our data over MCP?

For a blank-page draft, sure — generation is commoditized. The value isn't producing the output, it's being right. A general model doesn't know this customer's history, your KPIs, or which signals matter, so you spend your time feeding it context and checking what it invented. The model is a weekend. The labeling and KPI layer underneath it is the nine months. That's the part we sell.

We've already built half of this.

Then keep it. The half that almost never gets built is the closed loop — whether anyone acted on the signal, inside the window, and what it moved. Plenty of systems will now tell you an account is at risk. Very few can tell you who ignored it.

What happens to our build if we stop paying you?

Your data continues to flow to your warehouse for the life of the contract, and everything we've computed is exportable. If you've built warehouse-first, your dashboards and models keep running on the data you already hold. What you'd lose is the ongoing labeling and computation, not your work.

Is this services or product?

Product, configured. We stand up your scoring logic, plays and agents on the same layer every other customer runs on — so new workflows come online in days and stay consistent. Nothing here is a bespoke one-off.

Live and validated before the kickoff deck would have shipped.

"The integration was live faster than any tool we have rolled out. We were looking at twelve months of our own history in the first weeks, not building toward it."