1,000+ specialised AI agents · Do the work
Hand over a deal. Get back the underwriting.
- 125_Old_Broad_Street_OM.pdf
- Rent_roll_Q3.xlsx
- Leases.pdf
Underwrite this office for the investment committee.
214 of 1,000+ agents
Verified- Source 58
- Underwrite 91
- Diligence 65
It’s an office, so 214 agents ran on it, with the Company Brain and the Financial Model Engine. Here is what 5 of them found:
Office · 214 agents| Agent | Read | Finding |
|---|---|---|
| Lease Abstraction | Leases · Rent rolls | WAULT 6.3 years |
| Tenant/Covenant | Rent rolls · Alternative data | Top 5 tenants, 61% of NOI |
| Rent Comp | Comps · Transactions | 17 comparables |
| Cap Rate | Transactions · Comps | 5.4% entry yield |
| Re-letting Risk | Macro · Leases | 37% of income expires in 5 years |
1,000+ specialist agents for every real estate decision.
Purpose-built agents across asset classes, markets and underwriting workflows are dynamically assembled for each opportunity.
- Source 58
- Underwrite 91
- Diligence 65
125 Old Broad Street, London · Office · 329,200 sq ft · £385m acquisition
- Company BrainKnows your firm
- UK Office IntelligenceKnows offices
- UK Market IntelligenceKnows the UK
- Lease Abstraction AgentWAULT 6.3 years
- Tenant/Covenant AgentTop 5 tenants, 61% of NOI
- Rent Comp Agent17 comparables
- Cap Rate AgentEntry 5.4%, stabilised 6.2%
- Re-letting Risk Agent37% of income expires in 5 years
- Building Quality AgentReads surveys, EPC
- Debt AgentReads term sheets, rate curves
- Tax AgentReads SDLT, VAT position
- ESG AgentReads EPC, MEES
- Planning AgentReads the planning register
- Financial Model Enginecash flowsdebt scheduleswaterfallsreturns
- IC Memo AgentIC memo
On the London Office Acquisition, the orchestrator reads a bench of 1,000+ specialist agents and assembles a team of 214 agents with the Company Brain and the Financial Model Engine. Shown here: the Company Brain; UK Office Intelligence, Knows offices; UK Market Intelligence, Knows the UK; Lease Abstraction Agent, WAULT 6.3 years; Tenant/Covenant Agent, Top 5 tenants, 61% of NOI; Rent Comp Agent, 17 comparables; Cap Rate Agent, Entry 5.4%, stabilised 6.2%; Re-letting Risk Agent, 37% of income expires in 5 years; Building Quality Agent, Reads surveys, EPC; Debt Agent, Reads term sheets, rate curves; Tax Agent, Reads SDLT, VAT position; ESG Agent, Reads EPC, MEES; Planning Agent, Reads the planning register; the Financial Model Engine, which computes cash flows, debt schedules, waterfalls, returns; and the IC Memo Agent, IC memo. The IC memo shows a base IRR of 14.8% and a 1.9x multiple: proceed to IC.
The agent library
Agents for every kind of real estate.
No single agent covers an asset class. The asset class decides which specialist agents run. Pick one and see its set.
Land and development
Natural capital / specialist land
Office
01 · 10 agents
- Lease & Rent Roll Agentabstracts leases, breaks, expiries, incentives, indexation and contractual rent.
- Tenant Credit Agentanalyses covenant quality, concentration, guarantors and tenant financial health.
- Occupancy & Leasing Agentphysical/economic occupancy, downtime, expiries and rollover exposure.
- Market Rent AgentERV, achieved rents, incentives and competing stock.
- Supply & Demand Agentpipeline, absorption, vacancy and submarket dynamics.
- Building Quality Agentgrade, floorplates, amenities, age, specification and competitive positioning.
- CapEx Agentrefurbishment, tenant improvements, leasing commissions and lifecycle CapEx.
- ESG & Obsolescence AgentEPC/energy standards, stranding risk and retrofit requirements.
- Valuation & Comps Agentinvestment transactions, yields, £/sq ft and comparable adjustments.
- Office Underwriting Agentintegrates leasing, CapEx and exit assumptions into office-specific scenarios.
What an agent is
One job, one set of sources, one thing handed back.
Not a chatbot with a prompt. An agent is scoped to one part of a deal, reads the sources that part depends on, and returns a result the next agent can use.
One agent
Lease & Rent Roll Agent
abstracts leases, breaks, expiries, incentives, indexation and contractual rent.
Reads
- Leases
- Rent rolls
- OMs
| Level 04, west | Break | 24 Mar 2028 | Lease cl. 14.2 |
|---|---|---|---|
| Level 07 | Indexation | RPI, collar 1% | Lease cl. 8.1 |
| Ground, retail | Contractual rent | £412,500 pa | Rent roll, row 11 |
Every figure carries the clause it came from. That is what the next agent reads.
Ten of these make an Office.
The named agents
Eleven task agents, one part of the deal each.
Each investigates one part of an investment
In parallel, on the same deal, from the same sources.
These eleven worked the London Office Acquisition. On the next deal, the asset and the market decide which of the 1,000+ are assembled.

- Lease Abstraction
- Tenant/Covenant
- Rent Comp
- Cap Rate
- Re-letting Risk
- Building Quality
- Debt
- Tax
- ESG
- Planning
- IC Memo
The agents read every market through Market Intelligence.
Country-specific regulation, taxation, planning, transaction processes, financing, registries, macro data and local market conventions.
Market Intelligence →
Orthoimagery 2024 © City of New York (NYC OTI), CC BY 4.0 · Building footprints and street centrelines: NYC Open Data
One model per micro-location, not one per city.
Prices move block by block, so the model is built at that level and the specialist agents run on top of it.
What comes back
Each agent hands back one finding. The memo is built from them.
An investment recommendation, underwriting and scenarios, the maximum supportable price, risks and exceptions, a client Excel model and the IC materials.
Smart Bricks Investment Score
87 / 100
Evidence confidence: 95%
Source note 1, Investment Score, 87 / 100
- Method
- Smart Bricks Investment Score, 0 to 100
- Stands on
- 28 external sources, 17 comparables
Illustrative example.
- Entry yield of 5.4%, stabilising to 6.2%.Cap Rate
- A 14.8% five-year levered IRR and a 1.9x equity multiple.Financial Model Engine
- WAULT of 6.3 years, with 37% of income expiring within five years.Lease Abstraction
- The top 5 tenants carry 61% of NOI.Tenant/Covenant
- The downside case (ERV -10%, exit yield +75bps) holds a 7.1% IRR.Re-letting Risk
- 17 comparables selected.Rent Comp
In the memo
- Investment recommendation
- Underwriting and scenarios
- Maximum supportable price
- Risks and exceptions
Files
- Client Excel model
- IC materials
Material judgement
Humans review, challenge and approve material judgement.
- Review
- Challenge
- Approve
Follow one number back to its source.
The sources it read, the engine that did the arithmetic, the review it passed.
Finding · Lease Abstraction
WAULT of 6.3 years, with 37% of income expiring within five years.
- Read from
- Leases · Rent rolls
- Formulas
- Unexpired lease terms, weighted by rent. Income expiring within five years, over contracted income.
- Arithmetic
- A coded engine, not the language model
- Re-run
- Identical inputs, identical number
- Status
- Verified
In numbers
The agents, counted.
207
Countries covered
7 in 10
Agent tasks run on Smart Bricks models
30+
Frontier AI models, each task routed to the best one
Source note 2, 207 countries covered
- Period
- As of September 2026
- Method
- Every country Smart Bricks covers, one row each in the Market Intelligence catalogue
- Source
- Smart Bricks, company-reported
Coverage is listed country by country, down to the named dataset.
Market Intelligence →Source note 3, 7 in 10 agent tasks
- Period
- As of September 2026
- Method
- Share of agent tasks routed to self-hosted, proprietary models on our own GPUs
- Source
- Smart Bricks, company-reported
Model routing sends each task to Smart Bricks models, frontier models or coded deterministic engines.
Smart Bricks models →Source note 4, 30+ frontier models
- Period
- As of September 2026
- Method
- Frontier AI models under orchestration, routed per task
- Source
- Smart Bricks, company-reported
Frontier models handle language. Smart Bricks models handle real estate.
Smart Bricks models →Smart Bricks isn’t one chatbot doing everything.





