ActivoCEO · ActivoRules · with DecisionRules
Autonomous decisions, 24/7. You pay for the outcomes.
Our premium agent runs your workflows around the clock, enforces your policy in DecisionRules and calls dedicated solvers for the hard maths: price calculations, stock transfers and fulfilment routes.
ActivoRules gives your business a decision twin: a live, executable copy of how it decides. Simple decisions are answered by rules in milliseconds. Complex ones are optimised by a solver in seconds, then checked against your policy before anything reaches your order, warehouse or finance systems. Fees are tied to the outcomes we measure together, not to seats or licences.
- Startups
- SMBs
- Corporates
- E-commerce & retail
- Operations & finance
Built with DecisionRules
Activo designs, runs and monitors the agents, solvers and human checkpoints. DecisionRules is our rules engine partner: it holds your policy as rules that business owners can read and change, and that our agents call through its MCP server and APIs.
- Rules people can readDecision tables, decision trees and lookup tables
- CalculationsJavaScript scripting rules for pricing and margin maths
- FlowsDecision flows chain rules, APIs and databases; integration flows run long jobs
- GovernanceVersions, test bench and audit logs, in the cloud or on-premise
How it works
Three engines, one autonomous loop
Each part of ActivoRules does one job well. Agents handle language and judgement, rules make every policy call the same way, and solvers find the best answer when there are too many options to check by hand.
1 · Orchestrate
The agent runs the workflow
ActivoRules agent · 24/7Picks up every order, invoice, stock alert or request as it lands, reads the documents, gathers the data and decides which rule or solver to call next.
- Reads emails, PDFs, orders and tickets
- Builds a clean, typed payload
- Routes exceptions to the right person
- Executes approved actions only
2 · Enforce
Rules make the policy call
DecisionRulesYour policy as versioned rules. Same input, same answer, with reason codes and the rule version recorded for every decision.
- Eligibility, limits and approvals
- Price and margin calculations
- When to call a solver, and why
- Checks every solver answer
3 · Optimise
Solvers do the hard maths
Dedicated solversPurpose-built optimisation code that runs on scalable cloud compute, separate from the rules engine, and returns the best plan in seconds.
- Dynamic price optimisation
- Stock transfers between sites
- Fulfilment and delivery routes
- Built with OR-Tools, PuLP or SciPy
Right tool, right decision
Simple and complex decisions, each in the right place
Most decisions are simple and should be instant. A few are hard optimisation problems that need serious compute. ActivoRules sends each one to the right place.
A rule answers
millisecondsDecision tables and trees in DecisionRules apply your policy directly.
Refund eligibility, discount bands, credit limits, approval routing.
A scripting rule calculates
milliseconds to a secondJavaScript scripting rules inside DecisionRules handle moderate maths and chain lookups together.
Landed cost, price and margin calculations, commission plans.
A solver optimises
secondsA dedicated solver searches thousands of combinations for the best plan, then the rules check it.
Multi-warehouse fulfilment, stock rebalancing, routes, dynamic pricing.
Why keep heavy optimisation outside the rules engine? Scripting rules run in a secure, shared sandbox built for fast decisions. Problems like route optimisation or bin-packing can take many seconds of compute, so they run in their own solver service and never slow down or time out your policy decisions.
Hybrid architecture
The decoupled solver pattern
The rule decides when and why a solver should run. The solver does the maths. The rule checks the answer before anything is committed. Your agents trigger the whole loop through MCP, so it runs end to end without a person in the middle, unless your policy says one is needed.
Swipe sideways to see the full diagram →
Rules decide when and why
Policy says which cases deserve optimisation, so solvers only run where they pay for themselves.
Solvers do the maths
Stateless optimisation code scales on demand and can be improved without touching your rules.
Rules check the answer
No solver output reaches your systems until it passes the same policy limits a person would apply.
Worked example · e-commerce
Dynamic multi-warehouse fulfilment
A customer orders 15 items. Stock is spread across three warehouses, each with its own labour limits and carrier rates. Which warehouses should ship what, to keep cost and carbon low and still hit the delivery promise? It is a classic integer programming problem, and ActivoRules answers it on every qualifying order, day and night.
- Agent
Pick up the order
The agent receives the order the moment checkout completes and sends the order context to DecisionRules through MCP.
- DecisionRules
Apply the policy
A decision table checks the customer’s SLA tier, fraud risk and whether the order qualifies for optimised grouping, and returns
run_optimisation = true. Orders that do not qualify follow the standard rule in milliseconds. - Solver
Find the best split
A webhook starts a Python solver on serverless compute. It weighs stock, labour limits, carrier rates and carbon, and returns the optimal warehouse split in seconds.
- DecisionRules
Check the plan
The plan is checked against policy: maximum parcels per order, a delivery date inside the SLA and a cost ceiling. Anything outside the limits goes to a person with the reasons attached.
- Agent · systems of record
Ship and measure
The agent creates the fulfilment orders, updates stock and tells the customer. The saving against the standard rule is logged: that is the outcome you pay for.
| Measure | Standard rule | Optimised |
|---|---|---|
| Parcels shipped | 4 | 2 |
| Shipping cost | £38.60 | £24.10 |
| Carbon (kg CO₂e) | 9.8 | 6.1 |
| Delivered by | Day 3 | Day 2 |
| Time to decide | Instant | About 3 seconds |
Figures show how the outcome is measured, not a promised result. Real savings depend on your network, carriers and order mix, and are measured against the baseline we agree in the Decision Audit.
What we automate
Decisions that run without you
Every use case follows the same pattern: the agent orchestrates, DecisionRules enforces the policy and, where the problem needs it, a solver optimises. Each one has an outcome we can measure.
Price calculations
Rules apply price floors, margin targets and promotion limits; a solver tunes prices to demand and stock.
Outcome: margin protected
Stock transfers
Rules set reorder points and transfer limits; a solver plans which site sends what to whom, at the lowest cost.
Outcome: fewer stock-outs, less cash tied up
Fulfilment & delivery routes
Rules decide which orders qualify; a solver picks warehouses, carriers and routes.
Outcome: cost and carbon saved per order
Quote & discount approval
A discount matrix by tier, volume and margin floor approves, counters or escalates.
Outcome: quotes approved in minutes
Returns, refunds & support
Eligibility and goodwill tables decide; the agent drafts the reply and issues the refund.
Outcome: tickets resolved on policy
Supplier invoices
Three-way match tolerances and an approval matrix by amount and cost centre.
Outcome: invoices posted touchless
Credit & onboarding
A scorecard and rule chain set limit, terms and deposit for each new trade account.
Outcome: same-day onboarding
Commission payouts
The plan as versioned rules: tiers, accelerators, splits and clawbacks.
Outcome: payouts without disputes
Always on, always in control
Runs 24/7, inside limits you set
The agent never sleeps, but it never acts outside your envelope either. Autonomy is earned process by process, and the routing rule that decides when a person steps in is itself a versioned rule you own.
Assist
The agent prepares; a person acts. For new processes and anything irreversible.
Approve
Rules and solver prepare the decision with reasons; a person approves or overrides.
Supervise
Runs automatically inside your limits. People handle only the exceptions.
Autonomous
Runs 24/7 with sampled review. You see outcomes and drift in a weekly digest.
You own the policy
Process owners approve every rule version. Agents can propose changes and backtest them, but never edit production rules.
Autonomy is earned
A process moves up a level only when its override rate stays low for weeks, and drops back automatically if exceptions spike.
People take the hard calls
High-value, out-of-policy or irreversible decisions go to the right person, with the evidence and reasons attached.
Who it is for
Scale new kinds of decisions at every stage
The bottleneck looks different as you grow. The fix is the same: clear rules, solvers for the hard problems and an agent that runs it all around the clock.
For startups
Decide like a much bigger company
Every discount, refund and shipping choice waits for the founder. There is no ops team to optimise anything.
- Write your first policies as clear, owned rules
- Automate routine calls from day one, inside your limits
- Get optimised pricing and fulfilment without hiring analysts
For SMBs
Turn spreadsheets into autonomous operations
Rules live in spreadsheets and long-serving staff. Stock sits in the wrong place and shipping costs creep up.
- One source of truth for pricing, stock and approval policy
- Solvers that rebalance stock and pick the cheapest route
- A 24/7 agent that keeps orders moving overnight
For corporates
Agentic AI with control and auditability
Business rules are locked in legacy systems, and optimisation lives in one-off models nobody can run in production.
- Business owners change rules without code, with versioning
- Solvers deployed as governed services, checked by rules
- Every decision logged with inputs, outcome and rule version
Commercial model
You pay for outcomes, not seats
We only earn when ActivoRules delivers. Together we set a baseline in the Decision Audit, prove the difference in a pilot, and then base fees on the outcomes the system produces, measured from the decision logs.
Baseline
The Decision Audit measures how decisions are made today: volumes, cycle times, costs and errors.
Pilot
One decision runs in shadow, then with approval, so the gain over the baseline is proven on your own data.
Run 24/7
The agent takes the process live, inside your limits, and every outcome is logged against the baseline.
Pay on results
Fees follow the agreed outcome units, reported monthly from the same logs you can audit.
- Decisions run on policyRoutine decisions completed without a person
- Money savedShipping, transfer or stockholding cost against baseline
- Margin protectedPrice and discount decisions kept inside policy
- Time returnedApproval and processing hours no longer needed
Outcome units, rates and the baseline are agreed in writing after the Decision Audit. Examples only.
Governance by design
Clean, controlled and auditable
- Versioned rules, each with a named owner and full history
- An audit trail for every decision: inputs, outcome, reason codes and rule version
- Every solver answer checked against policy before it is committed
- Autonomy levels per process, from assist to supervised automation
- Backtesting and scenario tests before any rule or solver change goes live
- Least-privilege, governed write-back to ERP, OMS, WMS, CRM and payroll
- A kill switch and rollback to any previous rule version
- Cloud or on-premise rules engine to suit your data policies
Decisions teams automate first
- Price calculations
- Stock transfers
- Multi-warehouse fulfilment
- Delivery routing
- Quote & discount approval
- Returns & refunds
- Supplier invoice matching
- Credit limits
Start here
What you get from a Decision Audit
A focused working session with the ActivoRules team on your highest-volume and highest-value decisions. No platform commitment: if there is a fit, we follow up with a scoped pilot.
Decision inventory
The recurring decisions in your business, who makes them and how often.
Rule and solver map
Which decisions are simple rules, which need calculation and which are worth optimising.
Outcome baseline
Today’s volumes, costs and cycle times: the yardstick for the outcomes you pay for.
Pilot blueprint
A phased plan for the first decision, with autonomy levels and success measures.
Go deeper
The decision stack
Want to see every layer before we talk? The decision stack is the full blueprint behind ActivoRules.
ActivoRules: the decision stack
How people, agents, DecisionRules, dedicated solvers and your systems of record fit together, and how to roll it out one decision at a time.
- One architecture for every process
- Humans in the loop: autonomy levels and routing tables
- Scenario management and optimisation with solvers
- Ten worked use cases, from discounts to fulfilment
- A phased rollout plan and responsibility matrix
Questions leaders ask
Is ActivoRules a chatbot?
No. Agents do the reading, routing and drafting, but every policy decision is made by deterministic, versioned rules in DecisionRules, and every optimised plan is checked by those rules before it is used. The same input always gets the same answer, with reason codes and the rule version that produced it.
What is it built on?
The policy layer runs on DecisionRules, our rules engine partner, which provides decision tables and trees, scripting rules, decision and integration flows, version history, a test bench, audit logs, an MCP server and APIs, in the cloud or on-premise. Optimisation runs in dedicated solver services built with tools such as Google OR-Tools, PuLP or SciPy. Activo designs and runs the agents, solvers, autonomy levels and controls around them.
Why not run the optimisation inside the rules engine?
Scripting rules in DecisionRules are ideal for moderate calculations such as prices, margins and commissions. Heavy optimisation, such as routing a shipping network or bin-packing, can need many seconds of compute. Running it in a separate solver keeps your policy decisions fast and reliable, lets the solver scale on demand, and keeps a clean line between what the policy says and how the maths is done.
How do outcome-based fees work?
The Decision Audit sets a baseline for the decisions in scope. After a pilot proves the improvement, fees are based on agreed outcome units, such as decisions completed on policy without a person or cost saved against the baseline, and reported monthly from the decision logs. Units and rates are agreed in writing for each client.
Does it really run 24/7?
Yes. The agent reacts to events such as orders, invoices and stock levels as they happen, and to schedules such as nightly replenishment. Anything your routing rule marks for a person waits in an approval queue with the reasons attached, and you get a digest of what ran overnight.
Do we have to replace our ERP, OMS or CRM?
No. Your systems of record stay the golden source. ActivoRules reads from them through connectors and writes back only through governed, logged workflows.
Can the AI change our rules?
Agents can spot problems, propose changes and backtest them on past decisions, but they never edit production rules. A process owner approves each change, it ships as a new version, and it can be rolled back.
We’re a startup. Isn’t this overkill?
Writing policies down early is one of the cheapest ways to scale without the founder in every decision, and paying for outcomes keeps the cost tied to value. Start with one decision, such as discounts or shipping choices, and add solvers as volumes grow.
- The decision stack
- External & internal triggers
- ActivoCEO
- AI Models & Agents
- Supply Chain
- Actionable Analytics
Start with one decision
Find the decisions worth running autonomously
Tell us where decisions slow you down, cost you margin or leave stock in the wrong place. We will map your rules and solver opportunities, set the baseline and plan a pilot you only pay for when it delivers.
