What we do
Agents our clients run every day. Not demos
Estimates chased, reports checked, deals turned into jobs, land scored across whole counties. The numbers here are from the first month in production, not a pitch.
- 400+
- estimates captured in month one · Future Group
- 50+ hrs
- a month of re-keying removed · Earl Kendrick
- 80 → 20%
- false positives on fire risk assessment QA · Tetra
- 42
- local authorities of land scored in under two hours · Verafind
Four agents in production
What each one does, and what changed for the client.
-
Reads the inbox · chases · reports
600+ emails
An AI agent that chases every estimate
Nine users across three divisions adopted it in four weeks with no training. Spreadsheet tracking stopped.
Read the case study -
Extracts · creates the job · no re-keying
50+ hours a month
Won deals become Deltek jobs without anyone re-keying them
A deal marked won in Pipedrive becomes a Deltek job with the right fee structure: an 89.2% field-by-field match with human-reviewed jobs.
Read the case study -
Checks · flags · explains
80% → 20%
AI that checks fire risk assessments against the QA team's own rules
Section-level checks cut false positives from 80% to 20% on real Tetra reports, with around 800 reports a month in scope.
Read the case study -
Scores · ranks · maps
42 local authorities
AI land sourcing across 42 local authorities in under two hours
About a million title polygons scored across two local authorities in under an hour; 42 in a single run of under two.
Read the case study
What an agent can do
Six verbs. Most agents we build do two or three of them, with a person on the gate for the rest.
-
Read
The inbox, a shared mailbox, a document drop
Every email classified as it arrives: an estimate, a reply, a tender, noise.
Future Group: 600+ emails handled in month one -
Extract
Contacts, estimate details, line items, fee structures
Structured data out of unstructured email and PDFs, into the system that needs it.
Frankham Group: 15,000+ contacts from email -
Chase
Sequences tuned to the client's situation
Section 20 consultations and tender timelines change when a chaser should go. The agent knows.
Future Group: 400+ estimates chased -
Check
Reports against your own QA rules
The rules the QA team already applies, run on every report before a reviewer opens it.
Tetra Consulting: false positives 80% → 20% -
Classify
Which ones need a human
Won or lost, needs review or not, which document is which. The routine goes through; the rest is forwarded.
Spinach AI: which pull requests need a human -
Report
Pipeline, win rate, forecast, by division and client
The spreadsheet nobody kept up to date, kept up to date.
Earl Kendrick: every job's invoice forecast from its fee proposal
AI agents & automation, delivered for
- Spinach AI
- Penny House
- Leach Rhodes Walker
AI agents & automation case studies
-
Contractors
An AI agent that chases every estimate
400+ estimates and 80+ hours saved in month one
Future Group
-
Surveyors
Won deals become Deltek jobs without anyone re-keying them
50+ hours a month spent re-keying won deals
Earl Kendrick
-
Fire safety · Pilot
AI that checks fire risk assessments against the QA team's own rules
80% → 20% false positives on real Tetra reports, before and after section-level checks
Tetra Consulting
-
Land & development
AI land sourcing across 42 local authorities in under two hours
~1 million title polygons scored across two local authorities in under an hour
Verafind
-
Contractors
A platform that chases what block managers owe their suppliers
Block Debt
- Spinach AI PropTech Telling engineers which pull requests need a human and which don't 5.2× more likely to need a later fix: PRs sent for focus vs those left alone Spinach AI
-
Surveyors
One platform and one vocabulary for a surveying group
5 EK service lines that organise EKi's navigation
Earl Kendrick
- Penny House Finance & insurance · In build Meet Penny: AI that drafts and checks while the adviser decides Penny House
-
Surveyors · In discovery
Three AI agents designed for BREEAM, embodied carbon and SAP
3–5 days mapping each bill of quantities into One Click LCA by hand
Consult Sustain
-
Surveyors · In discovery
An agent designed to turn finished projects into case studies
Earl Kendrick
-
Surveyors
Fire compartmentation reports built from pins, not memory
3–4 hours on site per survey
Earl Kendrick
-
Surveyors · In build
Every job's invoice forecast, read from its fee proposal
15+ min of copy-paste to forecast each job by hand
Earl Kendrick
-
Surveyors
Over £2m added to the pipeline from contacts buried in email
15,000+ contacts extracted from email
Frankham Group
- Leach Rhodes Walker Surveyors · In discovery Four AI opportunities mapped for an architecture practice 3,000 documents in a single CDE drop, to find the 200 that matter Leach Rhodes Walker
-
Fire safety · In discovery
Designing AI to sort building safety case documents and draft reports
18 hours average to write a building safety case report
Tetra Consulting
-
Land & development
Land sourcing extended into Scotland in 13.3 days
13.3 days from registry sign-up to Scottish parcels importing
Verafind
Not sure it's an AI problem?
Some of the studies above started as discovery and ended with four opportunities mapped and no agent built yet. That's a good outcome too.
Which of these is your inbox?
Tell us where the time goes. Or call 0333 444 2298.