Step 1.
Give it a book and a patch.
Upload a 100-row slice from a list you have written off. Define the postcode districts and account types your brokers actually work.
Broker Book Engine Lead intelligence for UK energy brokers
Turn written-off broker data into calls worth making.
A custom, evidence-first AI engine built and operated by Inbound AI. It revives stale records, watches your patch for new businesses and hands over exclusive, ready-to-call leads — each with the reason to call attached. It runs at 0% invented facts on our eval set.
Operator-led deliveryKaylum — founder, Inbound AI
The operator
You work with the person who builds it.
Inbound AI is Kaylum — one operator who scopes, builds and runs every system. I'm a network engineer with six years in the field — three of them keeping infrastructure reliable at Amazon — now building agentic AI to the same standard. The person on your first call is the person who ships it and keeps it live.
6 yrsin engineering
3of them at Amazon
3live systems in production
0%invented facts on our eval set
Latest client system
Live
SyncPostrsyncpostr.com
Post once and publish across TikTok, Instagram, YouTube, Facebook, X and LinkedIn — with every comment and DM handled in one unified inbox. Designed, built and operated end-to-end for a client.
Book a callHow it works
From a tired list to an evidenced reason to call.
Step 2.
Verify before anyone dials.
The engine checks the current occupier, phone, trading status, premises signals and screening fields. Ambiguity stays visible instead of being filled with guesses.
Step 3.
Work a ranked call sheet.
Callable businesses are ordered by usefulness. Each row expands into its evidence, confidence and screening status in the portal, Sheet export or digest.
The engine
Four agents with one job: produce better reasons to call.
01 / Revive
Recover the useful part of a written-off book.
Verifies the present occupier and separates callable, changed-hands, dead and unresolved records.02 / Detect
Watch your patch for businesses taking shape.
Starts with new non-domestic EPC lodgements, then waits for corroborating business signals before minting a lead.03 / Watch
Flag likely renewal windows with the basis attached.
Uses EPC or broker-supplied anchors and models 12, 24 and 36-month terms without pretending an inferred date is known.04 / Learn
Use broker feedback to improve the next list.
Useful, wrong and dead outcomes feed tenant-specific evaluation so qualification rules improve instead of resetting every week.Sample output
Every opportunity carries its own evidence.
Broker Book EngineDEMO DATA
Patch: M20 / M21
TODAY'S CALL SHEET
18 records ready for review
RankBusinessReasonConfidenceStatus
01Sample Engineering LtdM20 3ABOccupier verifiedHighCALL TODAY
02Demo Kitchens LtdSK4 2HUChanged handsHighHOT COT
03Example Storage LtdM21 8PLNew EPC + signalMediumWATCHING
What changes
Better inputs for the people making the calls.
01Cleaner calling time
Prioritise records with a reason to call now.
02Fewer dead records
Closed and unresolved rows stop wasting dials.
03Evidence per row
Sources and confidence stay attached.
Broker BookEngine
04Patch visibility
New premises remain on watch until signals agree.
05Renewal windows
The inference basis is visible to the broker.
06A feed that learns
Useful and wrong outcomes tune qualification.
Engagement
Start with measured yield, not a volume promise.
Pilot
100 rows
14 days
Use a written-off slice to measure what is still callable and what your patch produces.
Book a callIncluded
- Revive across 100 supplied rows
- Detect sample in your chosen patch
- Evidence and screening fields per row
- Callable, changed-hands, dead and unresolved split
- Deposit credited against month one
Rollout
Live call
operation
Move the measured process into a tenant-scoped workspace your brokers can use each day.
See the workspaceIncluded
- Portal access and ranked call sheet
- Book upload and processing flow
- Google Sheet export
- Feedback controls on each record
- Patch configuration and exclusions
Operate
Watch and
learn weekly
Add renewal-window intelligence and tenant-specific learning once the underlying yield is known.
Read about the agentsIncluded
- Weekly renewal-window flags
- Daily patch detection ticks
- Digest email delivery
- Useful and wrong feedback loop
- Threshold tuning against broker outcomes
Comparison
Bought data versus a working broker engine.
Static broker dataVSBroker Book Engine
Name and address only
Every row looks equally callable
A phone number with no screening state
No source context
One file that ages immediately
Contract dates presented as fact
Broker outcomes disappear
Volume promised before quality is measured
Current occupier and trading status checked
CALL TODAY, CHANGED HANDS, DEAD or UNRESOLVED
Phone found and eligible to call kept separate
Evidence and confidence attached
Patch watch refreshed on a schedule
Renewal-window inference basis shown
Feedback improves tenant rules
The pilot measures your real yield first
FAQs
Questions brokers ask before the pilot.
There is no volume promise before we have measured the quality of your own book and patch.
What do you need from us?
A 100-row CSV from a list you have stopped working, plus the postcode districts and account types your brokers care about. The mapping step handles imperfect column names.
What does the 14-day pilot prove?
It measures your callable percentage, changed-hands percentage, dead and unresolved percentage, and the cost per revived row. Those numbers decide whether a rollout is sensible.
Are renewal dates guaranteed?
No. Broker Book Engine provides renewal-window intelligence. It shows the anchor date, assumed term and evidence behind the inference. A modelled window is never presented as known contract data.
What happens to unresolved rows?
They stay unresolved and are reported as such. Rural and industrial premises can have weak public coverage. The system does not invent the missing answer to improve a headline number.
How is calling eligibility handled?
Every row keeps phoneFound separate from eligibleToCall, with TPS or CTPS screening status and the screened-at time. Your team can see the state before anyone dials.
Does this replace our CRM?
Not by default. The first delivery routes are the portal, Google Sheet export and digest email. A two-way CRM connector is added later only when a retained broker names the system they actually need.
Is Inbound AI only for energy brokers?
No. Inbound AI builds and operates custom AI systems across industries. Broker Book Engine is the current specialist product for UK energy brokers.
Put 100 written-off rows through the engine.
We'll measure what's still alive, what has changed hands, what's dead and what can't be resolved. Then you decide whether the full engine is worth running.